*====================================================* *If you got that far in the video: NO BOARDS remaining! ALL GONE* *====================================================*
A couple of months ago I created an STM32F103 design and inadvertently put a GD32 in it. Inadvertently because these were counterfeit chips: the GD32 markings had been removed and replaced with STM32 markings. After writing my firmware to the GD32, it only half worked. Nothing on the I2C bus was responding. I wrote a test application that toggled the I2C wires as GPIO pins to be sure I didn't have the lines bridged or something. In GPIO mode, the I2C lines toggled fine but they did nothing in I2C mode. After acquiring genuine STM32 parts, everything on the board worked. I never solved the problem, but it did drive home one very important point: the GD32 is not a drop-in replacement for the STM32! I guess the moral here is: don't declare victory if you've only tested GPIO. It's comparatively easy to get GPIO right, but I2C, SPI, ADC and DMA state machines are much more complex. The I2C incompatibility of the GD32 is well known and it was a mention of this incompatibility problem on the stm32duino forums that led me to suspect, test, and verify that my first STM32s were in fact counterfeit remarked GD32s.
To be sure, having a standard test harness of some kind would be useful with these chips. I discovered that all the problems I had with the Blue Pill STM32 not connecting to USB were caused by the *fake* STM32 chips! I acquired some genuine ones and guess what? Yes, they connected to USB exactly like everyone else's. Thanks for raising this important point, good to hear from you.
Maybe this has been said but I don't want to wade through 80 comments to find out. I have used both paste and drag soldering on fine pitched chips. With the solder paste a good deal of the paste is flux. When heated the little metal balls melt and migrate to the solder pads with none remaining outside the pads. This works even without a solder mask. Your friend capillary action at work. Should be UA-cam content on drag soldering. Good video
Well, I've had a lot of different suggestions, Dan, including this method and variations thereof. I don't know why it was so difficult. Probably because it was so difficult to see what I was doing. I shall practice some more, thanks for the suggestion.
For the chip. Solder just one corner pin first, get it aligned , do not put solder on pads etc. then solder opposite corner, do not worry about extra solder or joins between other pins. use your iron on an angle put flux on the pins first, then flow a molten ball of solder down the row of pins. then to clean up use 2mm solder wick on the END of the pins to suck the extra solder down and away, moving the solder wick and iron together, no pressure on the pins. for the 0603 components use flux to stick them to the pads, then use a trimmed ice cream stick leaning on the component with a small weight, (I use the solder roll) to hold it then touch a very thin solder (I use 0.56mm) to the end of the component touch the pad and solder with the iron remove the solder and approach the component, and it should flow up the end. slide the iron back away from the component and pad. Takes practice. Other wise, use the flux again to hold the chips, and put a tiny bit of solder paste on the pad with a syringe, and use a hot air pencil to flow it. Either way clean up the flux with IPA after.
You've described more-or-less what I did (badly) with the STM32. I discovered you need LOTS of flux and when I re-did this yesterday it worked MUCH better. Regarding the 0603 I don't think I want to use them again, too small. And if you get the caps mixed up it's the devil to measure them all to find out which is which! I guess if it were that easy everyone would do it, right?
I coat the pads and legs with *liquid* flux, then center the chip, and leave it dry for quite a while to paste it on there ... then do a couple of corners lightly, and so on. Plain rock-hard resin, melted on the solder tip, can help the solder bead on to the legs and pads.
Hi. If you have problems in soldering these tiny IC's on a board (like me too), just let JLCPCB solder it for you. They have very good component prices too. Now I have no more fear in using SMD Components like 0402 capacitors or resistors in my designs.
0402 components are not my choice of size, they are but specks of dust. 0603 are OK and 0805 are relatively huge so easy to solder by hand. Under normal conditions when/if my eyesight returns! I've used JLCPCB's assembly service for SMT components (even did a video on it) and they are very good, as you say, Poppo..
Your comment on soldering. I have those days too. I just can't solder at all. Especially SMD stuff. After burning myself twice I usually find something else to do that day and leave the soldering to when I remember how to. Other days the solder flows like a glove every time as soon as it's touched, nothing sticks to the iron and everything ends up straight!
I've soldered through-hole components since I was 14 years old (I'll let you work out how many years that's been). I'm relatively new to SMD soldering but have got a good enough grasp on it now that I actually prefer it, due to the smaller size of components (so a smaller PCB), no lead cropping and it just looks so much neater (and some components are now only made in SMD format).
@@RalphBacon For SMD stuff I bought a load of 1.99 solder practice boards of Aliexpress and sat and did them until it was easy. That said, I just botched an LPFQ64 soldering. I dont think the chip will have survived. A reminder I need to practice again it's been a few years.
I have sometimes been successful with the drag soldering technique, where the tip is loaded up with solder, and dragged across the pins. You need a decent soldering iron and a good amount of flux, but when it works, it's great. You can also do reflow soldering using the hot air gun and just putting down a line of solder paste across all the pads. The solder only sticks to the pads and the legs of the chip, and surface tension tends to pull it all into place. You do have to do a visual inspection, or check with a multimeter, but generally a slight touch with a soldering iron will get rid of any short circuits.
I've been practicing with the drag soldering method and it works well, even more so now that I have decent soldering iron and a soldering tip called a Gull Wing, which is slightly concave and holds a reservoir of molten solder, specifically for this kind of soldering. I've even used my Hot Air Gun Soldering station with good results. Practice makes perfect! Thanks for your post, good to hear from you.
When surface soldering try putting a strip of paste right across all the pins (right angle to the pins). When you place the chip and hit it with heat the paste will melt and suck into place.
Your last two videos have encouraged me to dig our a number of Blue Pill boards I have stored away. It would appear that since V1.6 of STM32 Cores was released that all uploads require the installation of STM32CubeMX this provides the command line API to allow STLink and Serial uploads to the board. Another nice feature is the compiler setting to use USB Serial as the default UART thereby freeing up the serial GPIO pins for other uses, I've had no luck installing the Bootloader on these boards so I'm using STLink. Keep up the great work, I look forward to your video every Friday as they are always interesting and informative.
I'm really glad it's inspired you to dig your old Blue Pills out. I waited quite a while before jumping into the STM32 pool, to let the technology mature enough for it to be useable for beginners. Now is the time! Good luck and thanks for posting, Dave.
You need reflow solder paste and a heat gun or toaster oven to solder down small parts like that. Once the solder melts, removing the part is very easy. Unpowered, electronics will tolerate heat of melted solder quite well. A small heat gun will cost $65 or so and is highly adjustable. Toaster ovens are very cheap, but they don't regulate tempurature as wel las a digital heat gun will. In either case you are applying heat to the board until the solder melts. Be careful to only touch the part you want to remove. You'll work inside the toaster oven so the board stays hot and remove or place the part. From that perspective the toaster oven is less convenient. The heat gun since you are in open air makes accessing the board easier, but you have to keep the heat on the board so it doesn't cool down. For reflow soldering a new blank board, put solder paste on all the connections. It's somewhat sticky so you can use the paste to stick down the SMD components to the correct positions. Heat the entire board in the toaster oven or play over it with the heat gun and the solder will melt and the parts will pop into place into the solder pools. It's much easier than trying to solder down tiny parts with your iron.
I've got a heat run (sorry, digital soldering station!) and it works well, Richard, but not so great for actually soldering stuff onto the board (great at getting it off) as I am far too worried about all that heat being pushed at those components. Maybe I should not worry so much given what you are saying about that toaster oven - sounds quite tricky! Thanks for posting, good to hear from you.
@@RalphBacon Consider how those components got soldered down originally. They were done in an oven. The components were roboticly placed on the board and then the entire board and electronics were heated to the point where the solder melted. The parts were "stuck down" with the aid of solder paste until the solder had melted around them. Call it tricky if you like, but this IS how it's done. I do a fair amount of reflowing and I can tell you factually that the electronics survive this process quite well. If you are still skeptical, watch some videos on reflow soldering. For simple stuff, like reflowing an LED to a star board, you can use a frying pan and your kitchen stove. I know I have many times for LED's. For more complex stuff like a populated board, you need a way to hold the board so it heats uniformly and doesn't sit on or touch anything. A heat gun such as this will push very hot air onto the board to the point that the solder melts and then any component in the hot air path will simply lift off the board. A lot of engineering goes into making the electronics heat tolerant so that the solder mounting them can melt. www.amazon.com/DEWALT-D26960K-Heavy-Duty-Display/dp/B007X2ATZE/ref=sr_1_3?keywords=dewalt+heat+gun&qid=1570992759&sr=8-3
@@Lamster66 Sure...strawberries and apples are both sweet so they are the same thing...lol Solder is not all the same...not by a long shot! Buy some chinese solder and then come back with that argument. I've done enough reflow soldering to say that solder paste works best. I have used solder and yes...it can be done, but you get loads of solder bridges while a thin coat of solder paste does not typically bridge. I did say use a heat gun or toaster oven..
@@Lamster66 You just go have a nice day alright? I'm not even remotely interested in arguing with you. Congrats...on winning absolutely nothing that matters! I hope that settles you right on down into a reasonable human being.
Hey Ralph! The "O3" compiler flag is activating all optimizations for speed, e.g. that does stuff like unrolling loops and the like, to gain the most execution performance, even when this takes up significantly more program space in the outcome. The more moderate in regards to program space is "O1", which tries to minimize code size if possible. And "O2" is somewhere in the middle, where some optimizations can cause some increase in code size. --- Best wishes from Berlin!
Servus Sebastian! In which case, if the -O3 optimised version fits in the available space that should be the one to use - otherwise it's just wasted Flash space, running a relatively slower version of the code, right? MfG aus MK.
*Thumbs up* I got GD32 chips from LCSC as well. First idea was to design a custom board with different kind of external components for testing I2C, SPI and other things. After couple of design ideas, I decided to take advantage of the breadboard friendly BluePills that I already have. I also replaced chip on board with the GD32. There are two ways I like to use when soldering fine pitched SMD ICs. I think first one is easier, if one doesn't have good eye sight or hands are shaky. 1) Add thin line of solder paste (Doesn't cost much on eBay/Aliexpress, if you don't already have it) across the pads, place the IC on board and apply heat with reflow station / adjustable hot air gun. Alignment isn't too important, surface tension will take care of small errors in alignment. or 2) Add flux to the pads, place the IC on board and tack it in place with tinned soldering iron. Go on and touch each leg with soldering iron (dragging, or "poking" one-by-one, which ever feels more natural), adding flux and solder when needed. Solder bridges can be cleaned by adding flux and touching them with clean soldering iron. Larger bridges can be cleaned with solder wick. And of course there are reflow ovens and other ways to do it, but those two are simple methods for simple tasks of soldering one IC.
I'll give the reflow oven a miss, but your other method is not one I've tried with the solder paste to I will that a whirl too, next time I solder one up. I now use much more flux and it works a whole lot better!
Thanks Ralph and LCSC. I didn't know the GD32 was actually a viable alternative. I looked into that chip around a year or two ago, found some forum posts that said it had bootloader issues, and didn't go any further with it. Now I know better, thanks. While I'm no expert on soldering, I've soldered chips on several boards with TTSOP, and TQFP 32/64/100 chips. If I have a board I am going to etch several of, I make a solder mask and a stencil first. I make my stencils with a similar process to my single sided boards with a photoresist on a piece of aluminum cut from a soda can. I only use solder paste if I have a stencil. That will change soon though. I will be forced to use paste with a project that has my first QFN footprints/chips in the near future. I haven't had much luck using hot air with larger SMD chips or anything without a stencil. I actually use my main, large chisel tip on my iron. First a bit of background. I have the old RadioShack "Digital Soldering Station" still. It's an Atten 60w station that was just rebranded for R.S. It is a Hakko 900 series-ish station with a 2 wire element instead of a Hakko 900 type 4 wire, but it has an identical handle and form factor otherwise. The important bit is that it is a true 60w and uses Hakko 900 series tips. My main tip is a genuine Hakko 900M-T-3.2D. I got a bunch of them on eBay second hand NOS in the original individually sealed Hakko packaging. That tip makes a big difference for me in a lot of different circumstances like this. I have tried several dozen tips, many were clones and the last dozen or so have been genuine Hakko's. I have several fine pointed tips that I will use if I am forced to, but I use my main chisel tip 95+% of the time. This tip holds an enormous amount of thermal energy compared to all the other tips I've tried. I usually set it to 575F (302C) as my general setting. This is hot enough to work on almost anything. It is also hot enough that intentionally bridging standard 0.1in pins using good (MG Chemicals No Clean leaded 0.8mm) solder is nearly impossible. My low temperature setting with this iron for stuff like cheap perfboards and bridging pins is 505-515F (263-268C). All of my other tips I use regularly require settings that are around +100F (+50C) higher than this tip to achieve the same results. I credit this tip's thermal mass as the biggest contributing factor when it comes to soldering fine pitch SMD stuff. I also use a block of Sal Ammoniac. That's the traditional name for a block of ammonium chloride. This has traditionally been used as a soldering iron tip cleaner. It will remove any oxidation from a tip. I think of it like soap for a tip. If you rub the tip around on the block with a bit of solder, then clean off the junk with brass turnings and a wet sponge, when you put fresh solder on the tip it will be just like a brand new tip. This is also the condition when real Hakko tips do really well IMO. I always do a cleaning routine like this before soldering a large SMD chip. I start by applying a lot of Amtech 559 UV flux to the clean footprint. My rule for flux is: "Too much flux = I can't see the footprint any more. AKA There's no such thing as too much flux."
Wow! Thanks for such a detailed reply, Jake. Seriously, I'm very grateful you took the time to give me your experience. I could imagine in my mind's eye exactly what you meant too, as I was reading it. I think perhaps, that I just don't have enough experience at the moment and need to practice more on those Banggood practice boards. One thing I discovered to my cost it to *not* press the chip's pins using the iron or you will end up with bent, misaligned pins. So the dragging is just for the solder, nothing else! Experience lesson #1. I am thinking of getting a better iron than my Antex, which has served me well for 20 years or more but is not up to SMD soldering (and certainly not desoldering, not even using soldering wick to clean up afterwards, for which I have to use a much bigger, hotter iron). The range of tips for my iron is small, I'm not even sure they make it any more. Anyway, I will re-read you post a few times and see what I can extract from it to improve my soldering (although I did get both working so I'm not too disappointed!).
@@RalphBacon Getting it working is the important bit. Sorry if I'm too wordy. I'm not the brightest and feel like I need to explain too much to get my point across. My software code is no better. If I were to get a new iron it would probably be a T12 clone unit. Voltlog has some good uploads reviewing it. Those T12 cartridges with the integrated element and tip make a big difference. I didn't mention it earlier, but on my soldering station I actually made my own PCB's for the part that goes inside the handle. I did this to control exactly how long the heating element extends into the tip. I also tried to offset the ceramic element slightly so that it sits closer to one side of the tip. The Hakko 900 series tip bore is about 1-2mm larger than the ceramic elements that go inside of them. I made sure my iron has maximum thermal contact between the tip and heating element. That made a big difference. The original PCB/element didn't extend to the end of the tips. I'm about to add a dual iron setup too so I don't have to take things apart to change tips, but that's a whole nother topic. If your iron has enough thermal mass (or total power if you've got several thousand dollars for something like a JBC) you won't have problems with anything sticking to the pins, solder braid, iron, or otherwise. If you want something nicer than the T12 with it's 70w rating, watch the uploads from Marco Reps about his JBC setup. He only bought a JBC handle/ cartridge and built his own power supply and controller. It's one of the few DIY projects you can watch and see the CC actually using in later projects and uploads. Marco has been sent a sponsored T100 and used it before, but you will see the JBC most often in peripheral use in other content he has uploaded when he's recorded more serious bits. He also references a killer open source soldering station project on dangerous prototypes' website in his uploads about his JBC project. -Jake
You may find this interesting Ralph. I have this STM32_Nucleo F0 board and I have played around with it in the past. I used the STM development software Cube and HAL to program the board because at that time the only Arduino core (wrapper) was for the 103 chip. That Arduino core is known as STM32_Arduino core or "Rogers core". Your last video made me pull out the board and program using the Arduino extensions through Platformio. VSC/Platformio IDE uses STM32duino core. This core supports many more boards Nucleo, disconvery, and chips, F0, F1,F2,F3,F4,F7. The STM32duino was written by ST themselves and is now considered the "official" Arduino core. The official core uses HAL where as "Rogers" core (STM32_Arduino core) uses the Standard Peripheral library. I think in this video you are using the Rogers core which many people still use. No need to use Platformio as the STM32duino core can be installed into the arduino ide . The ST official core became available in May 2017. ST supports the core with a ST employee on Rogers STMduino forum, crazy days !
Check out a program called Zadig. It allows you to specify which USB driver is loaded. An alternative would be ImpulseRC driver loader. The Naze (pronounce Nah-zeh in a Japanese manner ) bootloader is what was loaded onto early multicopter fight controllers. Windows has a nasty habit of overwriting the USB driver so if stuff suddenly stops working that may be the issue.
I downloaded this (thanks for the heads up, Rob) but it could not install anything as it said the driver was already installed. I could see no way of 'forcing' it to use a particular driver for the STM32 (which has a weird, ie invalid, VID of 0000).
You need to select an appropriate driver from the combo box. The reference below relates to OpenTX used on the Taranis RC transmitter which uses an STM micro. STM is also popular on many other RC related hardware products. open-txu.org/home/undergraduate-courses/fund-of-opentx/using-zadig/.
17:40 It is pretty simple to solder that stuff. Actually tehre are 2 good ways to do so. One is with your hot-air gun, apply the paste generously across the pins, do not worry about the single pins - just lay a small saussage across each side across the pins on the board, place the ic onto it, about where it would sit when soldered in correctly and heat the stuff up. The solder paste will draw to the contacts on the board on its own and with a tweezer you simply make sure the ic goes into the right position by rudly but gentil pushing around the ic to keep it in the position you want it to be afterwards. --- the second way is almost as simple. Take some solid solder and put the solder with some liquid flux onto the board, but without the ic first. When all connections have soem solder on them put the ic on top, heat the stuff up and use a tweezer to push the ic around to fit the right position when the whole stuff becomes liquid. --- In both ways the solder substance as well as the ic want to get into the right position on their own anyways, since there is a certain addiction caused by surface tension of the solder pads and the solder substance. Just make sure there is enough flux in play to lower the surface tension on the pads on the pcb. --- There is also a third method working without reflow oven, which relies on using superglue. You simply glue the ic in the corret position and then apply paste onto the pins like a saussage on each side of the ic and heat the thing up to make the paste become liquid and find it's place. Excess is removed afterwards with a desoldering braid (?) ... [Entlötlitze]. Do not worry to have pins soldered together at the first try ... it doesn't matter. Just solder the thingy into place and remove the excess afterwards...that will do the job. [edit] So do not worry to get the paste onto the pads. It doesn't matter at all, which also makes stenils pointless for small series done at home. If there is too much, get the excess off afterwards with a soldering iron / heat gun. Not worries about placing solder paste. It pointless fumbeling about effort at the end of the day ...
Yes, I never understood how a stencil would really work for a single hobbyist board, Sebastian. The solder paste sausage has been suggested by others too; I'm waiting for an opportunity to test it out. Like most things, it's trial and error and lots of practice. And when you can do it you have no idea why you couldn't do it before. It's all about creating new neural pathways in this aging brain of mine! Thanks for the info, I shall practice soon, I suspect.
@@RalphBacon Well, there is one thing to use stencils for as a hobbyist, but that is not about soldering. The stencils are great as a template for spray painting things, like inscriptions/symbols on a front panels or for inscriptions on a solder-stop of a pcb and similar stuff. It is just pointless for applying liquid solder paste, since that really is a waste of time.
Great video, Ralph. Just one thing though - the number of primes per 30 sec is not a good indicator of speed, because the amount of time required to check if number N is prime is proportional to N^2. I think it would be better to fix the amount of work the CPU should do (let's say, calculate the first 1000 primes) and measure the time it takes to do it.
I'm sure you're right, Dima, about it being a strange test, but then it is the _same_ test for each processor, so no matter how it determines whether a number is a prime, they are all doing the same work? Funnily enough, the even odder method used in the previous video showed how long it took to generated the first 500 primes. Dang! I should have left it alone.
Sorry Ralph, I was not quite right. The amount of time required to check if a number N is prime is not N^2, it's sqrt(N). My original point still stands - it's not a linear dependency. Try calculating the first 500 primes and the first 1000 primes on the same CPU, the latter will take at least 4 times as long as the former. So, if two CPUs can do 500 and 1000 primes in the same amount of time, you could've thought that one is twice as fast as the other, but it is in fact 4 times as fast.
Oh, and one more thing. Your arduino sketch uses sqrt(num), which does floating point calculations and returns a floating number which then gets converted into an integer. ATMega and STM32F103 don't have hardware support for floating point, but ESP32 does, so it's not a fair comparison. Try removing the sqrt() call and changing line 52 to this: for (long cnum = 2; cnum*cnum < num; cnum++) Then your sketch would use only integers. I wonder how much difference it'll make..
Thanks for the tip about the naze32 firmware, I spent a couple of hours trying different ways to upload over USB. It is possible that you may have restored my sanity.
I hope so, although I have no idea why that particular bootloader works and the "official" one doesn't. And yet others have no problems with the USB and the official one. I blame Bill Gates.
@@RalphBacon naze32 didn't work for me. Sanity is back in question again. Serial programming will do for now. I think it depends on what clone you after using.
Hey Ralph, fun and interesting comparison. I guess the primes/clock is roughly the same for GD32 and STM32, with the ST one having a bit of an upperhand. Some suggestions: 1) overclock the STM32, you can make it work at 108Mhz by modifying the PLL clock multiplier which is usually at x9 for 72Mhz but can be set higher. 2) check out those ST-Link V2 programmers, they are really cheap and use a STM32F1xx chip inside and if you open them they include headers so they be reprogramed like any other STM32 Thanks for the great content.
Yes, the GD32 does seem OK and is, apparently, used as the _de facto_ STM32 in Chinese markets. And it will be able to run at 108MHz too, just as you say. I do have a couple of ST-LINK programmers now (after I blew up my previous one, no idea how I did it) but haven't really used them in earnest yet. Thanks for posting, great to hear from you.
I picked up one of those "blue pill's" about six years or so ago. The Ebay listing said it was a super fast arduino! Never have gotten the blasted thing to work. Yesterday I was cleaning in my old backroom shop, long abandoned in favor of my recliner, when I came upon it, in a little static proof envelope that I pushed it in after many disgusting attempts to make it work, so thought I would view some of your videos and see if I could get the damn thing to do something. Still haven't tried but it is laying next to me on my end table right now. As far as micro soldering, I have the perfect tool! My unaided vision is at 20/400 20/400 (my doc says it is much worse then that but that is the top of the chart) With glasses, I do have 20/20 in both eyes. So all I need to do is remove my glasses and add a bright light, and instant microscope so long as it is about 7 inches in front of my nose. For an iron, I picked up some curved needle type tips, I use the backside of the curve for most joints but for tiny SMD's the tip works good so long as I have a good hold with my curved tweezers. I built one of those little SMD wristwatches with the clear case and leather band that are being peddled on Ebay using that, all very tiny soldering some in very close proximity to some melt-able cables that hook the screen to the board. I was very surprised indeed when the blasted thing actually worked when I finished. It had a couple of chips like the one on that board. The way I got them to work was to flood them with solder, then use the iron to flick off till only the links remained. I didn't think it would work but saw someone on a video do that so figured it was worth the try.
I don't really know what 20/400 means but it sounds bad. I've found a source of +6.0 and +5.0 glasses on Amazon so with my contact lenses I can now do SMD soldering with just a pair of bottle bottom glasses. Cool. (No, it doesn't look cool, as my eyes are as large as a watermelon with these on, but at least I can focus real close again).
I'm getting the age related vision problems too. For small work like that, I use the +3.5 magnification reading glasses that my local dollar store sells. They work great.
Yes, I got some of those too. But I'm more short-sighted than that which makes it harder to focus really close up. I'm now using +6.0 reading glasses in one eye, +5.0 in the other (I swapped over the lenses from two pairs). Now I can solder _really_ close up. I can't see a thing further than 6 inches away but for soldering it's even better than my jewellers headset magnifier. All in all, I'm using 3 sets of glasses (over my contact lenses) for computer, middle distance and close work until I get my cataract surgery done. What a pain!
Yet another great video, thank you. For soldering the chips, definitely look at drag soldering (search UA-cam for David Watts and scroll back a year or two and he has a great video doing it). Alternatively spread some paste across the pads, drop the chip on and heat with your air gun. So long as there is solder mask between the pads then surface tension will pull the solder to just the pads and pull the chip into place. If you get the amount of paste correct there will be no shorts as the solder mask repels the molten solder. Alternatively take a look at Unexpected Maker's channel and search for the video on his reflow master. More expensive though as that requires a toaster oven.
I'm (slowly) getting the hang of dragging properly, Gavin, (I was using far too little flux paste) but I might get a better soldering iron too (or tip, at least) that has the "well" of solder.
One time I was eating chocolate chip cookies in front of my PC, when I picked up and ate a very crunchy crumb. I'm pretty sure it was one of the SMD parts I was picking out previously, when going through a list of needed components for a PCB, but suddenly vanished out from the tweezers. But don't worry, the parts were conforming to RoHS, so should be fine ;) I do like the 0603 or even 0402 size, not necessarily when designing a circuit, but for bodges and repairs. They fit in between pins or almost anywhere on a PCB after scratching away the solder mask. And they work even better on one off projects that are quickly done on perfboard, instead of using "huge" THT components. I'm quite myopic (5 diopters), so I see squat in the distance without glasses, but it's very useful when doing closeup stuff. Over 40 now and definitely recognizing that I can't go as close with my glasses on anymore, but the myopia will stick (luckily?), so no reading glasses for me.
This is probably a true anecdote, and I laughed a lot when I read it! Is it the first step towards Borg assimilation, do you think? I just hope it was a capacitor you swallowed. Resistance, RoHS or not, is futile.
That's CS32 from CKS... GD32 (F103) uses serial flash (like ESP32) to store program. That flash then copies itself at boot to SRAM (yes, it has more SRAM) which allows microcontroller to run at higher speeds with no waitstates (and elaborate accelerators). CS32 - I don't really know how they handle things. However ARM Cortex Core performance isn't actually the most important thing in microcontroller world. It's more interesting if GD32 and CS32 has at least similar ADC, Timers, GPIO, SPI, I2C, DMA performance.
Yep, and someone have decapped it more than 3 years ago. For those interested in die-shot and interesting analysis visit: zeptobars.com/en/read/GD32F103CBT6-mcm-serial-flash-Giga-Devices
The other parts being so close may make it hard to use a soldering iron but I'd try the "Flood and Suck" approach. Clean the board well, flux the pads, place the part and solder it with the iron, not worrying at all about the bridging, just get more than enough solder on the part. Then use solder removal braid and suck out all the solder from the pins you can. The bridges go away but the surface tension will leave enough solder between the pins and the pads so that you end up with a nice solder job. To be sure, after soldering, use a magnifier to inspect. I use a xacto knife under a magnifier to gently push each pin. If the pin moves, it's not soldered. Rinse and repeat.
I like the idea of pushing the pin to see if it soldered, David. I had a couple of unsoldered pins. But now I have a better solution overall so keep watching, all will be revealed in a future video!
I just got hold of an STM32F401CCU6 WeAct board. I ran the Primes test and it finds 88049 primes in 30s, which is impressive. Programming was straightforward and the USB bootloader worked first time. I think its work looking at both this board and the 100MHz STM32F411CEU6 WeAct board. Both of these boards have Floating point processors so that brings other performance enhancements over the STM32F103 blue pill boards
Ah, yes, a FP processor will undoubtedly assist in speed (when calculations are required). It's quite amazing how quick some of these STM processors are, although the ESP32 seems the fastest and most rounded in capability right now IMHO.
It's obvious it's more performant. STM32F4 is using ARM Cortex-M4F, which adds a FPU compared to STM32F1, which is Cortex-M3 and has no FPU. You might want to have a look at STM32F405, these run at 168MHz and can work as a USB OTG host at USB 2.0 (480Mbit/s) speed
if you dont have a stencil and solder paste, i just slightly tin the contacts on the board with regular soldering iron, then add flux and hot air the chip on works pretty well
Age, and a dodgy eye, has already got me, Michael, so I was foolish even attempting this (but now that I've done it, and recorded it, I'm feeling just a little pleased with myself that it eventually worked). But never again with components (specks of dust) that small!
Like others have said, I'm a fan of spreading some solder paste over the board and running over it with a soldering iron, effectively tinning the pads without putting solder everywhere. Then start in one corner and work your way around. Also, plenty of practice can be done on old or broken boards- why practice on good components 😀😀😀
16:11 - the caps around the LDO are insane and will _cause_ usb problems. The USB specification lists 10uF as max (to lower the inrush current), so a 10uF and a 100nF on each side is just about right.
Indeed. Madness. Think of the inrush current! Perhaps it was a misprint and was supposed to be 33µF? I've never seen such a large cap after an LDO. 10µF with a 10nF in parallel perhaps is standard, as you rightly say.
Ralph take a look at the ADSM301 and the ADSM302 USB microscopes I got the ADSM301 about 6 months ago and I love it, it was around 150$ US, now looks like it's around 180$, for a 20$ or so more, the ADSM302. Before, I was using a magnifying lens and I would get headaches and my eyes would "wig out" :) for a few hours. These have excellent working height unlike the cheap 10-20$ USB microscopes. My next planned upgrade would be to mount it on a articulating stand so I can examine larger boards like PC motherboards and XBOX boards
Oh I did, believe me Sylvain. Long, long hours evaluating them all. But it's too expensive to just go out and buy so I'm starting off with a cheapie that most will probably shudder at but others have said good things about it so it's worth trying it out, just to see how it goes. Buy cheap then buy the tool you really need, hey?
The best way to solder these devices is by "drag soldering" where you wipe plenty of liquid flux over the pins and then drag a ball of solder on the soldering iron tip along all the pins. The key do success in this method of soldering is the use of liquid flux.
Not true. Even if we run the GD32 at 72Mhz it is still quicker due to the zero wait state for flash memory. I'll do a further demo in due course on this aspect.
@@RalphBacon Ralph, I agree with you. These clock crystals are specified to 100 PPM (or better) or an error of 0.01% (note:PPM Parts Per Million) The performance difference shows the "Blue Pill" 4.4% faster than the "Maple Mini". If these clocks were that different, I would expect to see errors on the serial communication due to baud rate issues. (although not guaranteed).
16:12 the guy changed the caps but never thouoght about adding a free wheel diode from out to in to protect the AMS1117 from the surge of current coming from the C6 discharging and going thru the reg everytime power is down, which will surely kill it in no time! The kind of thing I directly notice in a circuit and shun reading the rest.
Is a fly back diode really required across a voltage regulator? The spec sheet makes no mention of this and the internal circuit diagram of the AMS1117 doesn't indicate that it would be helpful. If the flyback is to protect from inductive loads then the place for the diode is across the coil there, surely?
You can solder the LQFP just by putting a thin stripe of solder paste over the area the pins are to solder to. Hot air, wicking action combined with the solder mask will flow the solder to the pins and unless too much solder paste was use should not cause bridging. Use the desolder wick as indicated by David Westernall to clean up any bridging you see under microscopy. The hard part of this method is the chip will tend to float on the solder pads so you need to use some kind of a pick to position and then hold the chip down for a few seconds while the solder cools. Using a USB microscope to solder under does not work well. It is very difficult to position components and tools without stereo vision.
I wondered how the paste would fare, Carol; but when I use it, it does seem to get everywhere! Practice is the answer I suppose. Perhaps the USB scope will work OK for me as my stereo vision is pretty poor anyway! Yet another cross to bear... Thanks for posting, nice to hear from you.
Drag soldering, with 0.5mm soldering iron tip works fine down to 0.4 mm pitch chips. Micro scpe look at ADM503, good work clearance, many mscopes too close no space to work
That's pretty much what I ended up doing, Helmuth, but I was very unimpressed with the result (as you may have noticed). Others have suggested solder paste and a hot air gun (which I used to desolder, not solder). I might experiment a bit more. Thanks for the suggestion about the microscope.
I have been meaning to ask you how you were getting along with the YOCTOSUN LED Light Hands Free Headband Illuminated Magnifier that you bought when starting to use SM components
I guess you saw me in the video with them then, Bob? Essential Stormtrooper headgear, I would say. I could not have done this soldering without them, that's for sure. 20 years ago I would not have needed them. Another comment here says that they bought a strong pair of reading glasses from one of the Chinese websites (x4 magnification) and they worked OK. I did a test against my x2 readers and my x2 headset; the headset was more powerful so I don't know where the x2 comes from. More like a x4. And they go up to x3.5 (which is a bit like x7).
@@RalphBacon Thanks for the reply. I originally wrote my post when you said that you had used your camera when soldering the SMDs. In fact my post ended by saying something like "I presume, therefore, that the headband was not a success". Of course, as soon as I saved it you mentioned the headband so I edited it to remove the last sentence. It sounds like it was useful after all. I can feel an illuminated headband in my future !
Believe me I watched _lots_ of videos before attempting this, Graham, as well as practicing on the Banggood boards. Everyone make it look so easy! Which it probably is, once you have mastered the technique. So I shall practice some more.
I believe the right technique to solder chips like that is to use some of that jelly like flux that comes in a syringe, you tin your iron tip and drag the tip across the pins, the flux does it's magic and the solder just sticks to the pins/pads perfectly. Look at this video from Snazzy Labs where he's building a RPi Zero retro gaming handheld - in particular where he solders the ribbon cable for the screen. I also believe iFixit have done a series on micro soldering too.
I have that flux in a syringe (I used it lots for this video!) so I will search for SnazzyLabs (no link! - I will find it). [Edit] Found it! ua-cam.com/video/YrofIDloHgw/v-deo.html
I got a blue pill about a week ago. I was trying to hook a literal USB to serial link into the uart for programming and of course that would not work. Then I tried to program it using a ttl serial converter today. It does not work for whatever reason, including me blowing the uarts. I soldered a 2K resistor I found on some scrap on. I finally decided I had to use tweezers while soldering to force it into place because you just can't to it with just your soldering iron. Very small stuff.
@@RalphBacon oh later in the story it was a learning experience,. The serial program does work if you (de)press reset at just the time the cube programmer pauses to connect. Also I used the stduino and pc13 generic bootloader and it windows it is pretty much the same experience as an arduino at this point. Linux needs some examination. Does not go into DFU mode as it stands.
You're saying board X is so many times faster than board Y. However, the test you are using doesn't scale linearly for multiple reasons. For one checking if a number is prime is faster for smaller numbers. Furthermore you seem to be printing the number of primes found and not the total numbers that have been checked. The distance between prime numbers is bigger for bigger numbers.
@@RalphBacon Yes, using a fixed number of calculations is easiest. You could pick a low number such that the time is acceptable. If you find that the results are too noisy then just repeat it and take the average. Although I would expect on microcontrollers like these the time is quite predictable. Just be aware that the Arduino framework might do things in the background.
@@RalphBacon Yes, exactly. Pick a number that the UNO can do in a reasonable time (50 to 60 sec ) by trial and error, and then stick with that number for all the tests. The times for the other boards are then inversely proportional to their computation speed.
@@RexxSchneider well and you're depending on oscillator consistency if it is timing itself... Swapping the chip actually would have helped a bit (if you weren't using any of the internal oscillators)
18:00 Lots of Flux (like flooded with paste flux) then it is super easy since the flux will help prevent the pins from bridging. (Louis Rossman has a channel on Apple board level repairs, it is awesome). Here's an older vid where he does a QFN chip smaller than what ours ua-cam.com/video/mr1UVPsExiE/v-deo.html
Yes, he uses so much, James, but it works! But I always clean mine off, I don't trust the "no clean" label one bit. I like his microscope too but beyond my reach, that's for sure! Thanks for posting.
@@RalphBacon The 'official' term is a Rossman unit of flux /s I agree it is excessive as is his iron temps but it is hard to argue with his results! :) I have also fallen in love with using the Amtech flux he uses, it has transformed my ability to perform SMD repairs and builds. I used to think drag soldering tiny adjacent traces was magic, but nope its just the flux :)
You can get them up to +4.0 on places like Banggood and AliExpress. In the shops here they only seem to go up to +3.0. But then you can get really close for SMD soldering. Cheers, David, thanks for posting.
Hi, I don't go below 0805 either I think that's it at 53 without any expensive gear! as for the chips I thought the same way and now I'm playing with a dirt cheap (£6.70) hot plate, all I have to do is get a kindly Arduino to automate it and bingo, Roberts ya relative!, seriously look up "hot plate reflow" see what ya think, I mean there are folks getting good results using an upturned Iron!
Hotplates are quite popular. I think the biggest issue is heat uniformity. It's also more difficult to follow a proper temperature reflow profile as you're applying heat to the back of an insulator (pcb). Most people seem to monitor the surface temp of the hotplate but the part that matters is the chip pins and footprint. I have a "PCB fabrication" playlist saved with all the different hotplate, toaster ovens, and reflow rigs I've come across that other people have uploaded. The coolest cheap and unusual rig that comes to mind is one by Andy Brown. He used a halogen flood light with an enclosed container and μC to make a reflow oven. That one is definitely worth checking out :-) -Jake
@@UpcycleElectronics Oh my!!!...My name is Andy Brown and I have just bought a Halogen hotplate from Ebay and I plan to use an Arduino to control it. I also want to monitor the temperature by embedding a thermocouple in a PCB in order to give me a reading closer to the actual board temp...this is so spooky please give me the site address to check out..as I haven't done this yet!!!!
@@andymouse Lol The impostor/Other Andy's upload: ua-cam.com/video/6bxpy1Ret6Y/v-deo.html My playlist with a few references I've come across: ua-cam.com/play/PLPIwHuVy9EyNsNgV-mOPapYaKjEw2C_kP.html Probably the most notable upload I've seen about hotplates and reflow: ua-cam.com/video/075D_tuswVU/v-deo.html -Jake
I hear what you are saying about small components and being older. I would recommend a large illuminated magnifying glass. The board is likely to have lead free solder on it also using flux paste will help a lot.
Hi Ralph, I think I'm about your age and I have the same kind of eyesight issues. I have a couple of suggestions. Drag soldering works surprisingly well. You need to use plenty of a quality flux. I use water soluble organic flux. It works well and cleans up beautifully with soap and water. I'm not sure when you made this video, but now you can have JLC PCB assemble your boards as well as fabricate them. I've also had success with solder paste and a toaster oven. Thanks for the video! Cheers!
Ralph ... I can TOTALLY relate to your eyesight/soldering etc. thing and everything else you're talking about. I'm 65 and can't see my hands in front of my face without one or another pairs of reading glasses. HOWEVER ... there is help. First ... Forget about USB microscopes for soldering. What you want (trust me) is an "Optical" stereoscopic microscope. I heard the part about 'you can't afford much', but honestly, these are a game changer, and they are a LOT less expensive than you might think. (USB microscopes are great for other things, but not for soldering.) I bought an Eakins 'trinocular' (the 'tri' means there is a 3rd tube for mounting a camera), but the 'binocular' is more than adequate if you just want help seeing small stuff while soldering. Get the 'lowest' magnification range, typically 3.5x-45x. (trust me, you don't need any more than this.) You can find more on their AliExpress site: www.aliexpress.com/store/836673?spm=a2g0o.detail.1000002.2.5aeb643f8U2BIV They have a billion different options on their site, and it can be confusing, but really everything comes down to a few options: 1) trinocular vs binocular, 2) high vs low magnification, 3) type of stand, and 4) which camera if you pick the trinocular option. Eakins is not unique, there are several other suppliers selling very similar equipment, for similar prices. Further, every month or two they have sales. ... Wait for the price drop. Why the pitch? Recently, I bought about 15,000 0402 resistors and capacitors from LCSC for a total of only a few dollars. Look for yourself; you can purchase 1000 resistors for as little as $0.38CDN/1000. I've never heard of anything SO CHEAP. 38 cents for 1000 of anything. I haven't done much with these yet but importantly, with my microscope, I don't have ANY PROBLEM at all soldering them. Did I mention ... my eyesight SUCKS, and is probably even worse than yours. And, soldering 0603 parts is literally ... EASY and routine. You didn't mention what kind of soldering iron/solder you are using. I have learned over the years that these make a world of difference. Some time ago, I bought a 'BAKON 950D' soldering iron. This is one of the 'new' type irons, which is VERY responsive to temperature changes. Unlike others however, this sells for about $50, and honestly it is ABSOLUTELY AMAZING! Again, there are other similar options. Solder is a whole other issue, but I use 37/63 lead based solder, 0.3-0.5mm. It is FAR easier to work with than lead-free solder. But if you think lead is a dirty word, then spend a LOT of money and get GOOD lead-free, otherwise you'll find soldering to be tedious and difficult. I have no commercial interest in supporting the above ... I've simply figured these things out from experience. I hope this helps.
I feel your passion and enthusiasm in your answer, thank you for all the information. Now, my eyes are slowly improving, to the extent that I can solder 0603 with just magnifying (jeweller's) glasses +3.0 / +4.0 although because my eyes are not the same (they got the replacement lenses for my eyes wrong when they did my cataracts) I still have some difficulty with stereoscopic vision. I go back to the eye clinic on 10th August, when the consultant will decide whether we can continue eye drops or whether some further laser surgery needs to happen to remove scar tissue from my retinas. Regarding the USB microscope, I find them very good even for soldering (no 3D vision, admittedly). I did investigate the type you mention (binocular) but eventually decided the cost vs the benefits could not be justified - but then again it sounds like your eyes are (even) worse than mine! I have a great soldering iron now, the Atten ST-100 which uses Weller tips and heats up in seconds and has great temperature control too. As you say, a decent iron makes a huge difference in soldering success and I never use lead-free, it is the devil's spawn. I use 0.4 - 0.6mm leaded solder and it works well especially with copious amounts of extra flux for SMD components. Your comment could have been written by me, we are on the same page. At the moment (and until the eye surgeon signs me off) I'll stay with my current setup then consider that binocular scope again, thanks for the information.
@@RalphBacon Thank you for your quick reply. I feel a bit awkward now, since I didn't mean to pry into your medical details. Good to hear things are getting better though. For the price, I'm thrilled with my little BAKON soldering iron, but clearly there's no point in dumping your perfectly good Atten iron. There is actually a 'significant' difference between 60/40 and 63/37 solder ... not sure if you and/or others are aware of it. Briefly, 63/37 forms what is called a 'eutectic', which in a nutshell means that at this ratio, the alloy behaves more like a pure element with a very precise melting temperature (183°C in this case), making it ideal for hand-soldering electronics. 60/40 tends to stay semi-liquid (sticky) in an around the higher melting temperature (190°C), making it ideal for plumbing. This can make a lot of difference especially for tiny parts. (Lead-free solders by comparison, (there are many), often suffer from stickiness and higher sloppy melting temperatures, making them difficult to hand solder.) As for the microscope, it was a big, much considered decision for me. I also considered the less expensive USB option, but I'm very happy that I bought the optical. It made a huge difference. USB microscopes can still be useful for inspection and making videos etc.
Save up your coins and buy a true stereo microscope you will never look back. Get it with a .5x Barlow lens for better lens to board distance & easy access for soldering. For the small parts use .020" or .015" diameter solder (use lead based 63/37, none of this lead free crap) and a soldering iron with a fine conical .016" diameter tip. I use a Metcal iron but others irons with small tips should work fine.
I looked very hard at true stereo microscopes (I would actually want a trinocular one so I could film the stuff, or output the image onto a big screen for recording). But at around £600 ($750) I am going to have to forgo this until I have at least 100,000 subscribers! But you never know, Amazon Prime Day is coming, perhaps they will knock 50% off!
I agree. Had similar problem with ESP32 Cam. I added a capacitor to the supply input and it shut down my supply. Had to re-set and reconnect the supply 3 times before it stayed on. All down to inrush of current.
@@EdFrench_uk Yes, that is why I had to fit the cap too. It was when connecting the supply, it was 3rd time before the cap had charged enought so as the surge didn't trip the supply.
@@EdFrench_uk Wouldn't that drop the voltage. The ESP-32 cam suffers from a lot of noise lines on the image if the voltage is low. A few 100mV and the picture is unusable. I had to use 6 volts from a boost converter to power it enought to get a cleaner picture. Could be down to the onboard LDO not handleing the higher current draw of ESP-32 and camera!
I am impressed, soldering these small parts is not easy, I prefer 0805 parts. I have done some chips with about 0.6mm pitch, they are not easy (I cannot remove shorts with the soldering iron, but has to use solder wick). I often use a USB microscope to check stuff, but it is not easy to solder under. It looks like you are getting away with comparing the performance with the long data type, both mega and ESP32 use 32 bit values and I supposed the STM32 will also do that.
Yes, the power of a 32-bit processor. Poor old Arduino Uno/Nano never really had a chance to compete.Don't be too impressed with my 0603 soldering, Henrik, I thought it was awful. I will stick to 0805 in future. Perhaps I can stretch that r3bers board up to that size _and_ add in an FTDI or other USB controller too. Then I can really call it a Green Pill and retire on the proceeds. Meanwhile, back on Planet Earth...
A STM32 or similar that is cheap (I.e. cheaper than ESP32), small and works perfectly with the Arduino environment would be nice. The best/easiest way to get money is probably to start a succesfull company and sell it (This can be a cooperation between multiple people).
Since GD is running at a higher frequency it is not a very fair comparison. To make it more reasonable, you should try overclocking those chips and check what is their maximum frequency. Chinese manufacturer may have just claimed higher specs but in reality, it may be pretty much the same thing. Stm32f103 is known to work fine at 128MHz and some even claim 160 and over 200mhz As I know the limiting factor here is flash speed, if you copy your program into ram you can increase speed even more and make it closer to esp32
You're the second person to use the word "fair", as though these chips need defending! They were just out-of-the-box tests to see what they offered. Yes, the GD32 _might_ be internally overclocked to appear to give an advantage. But we also know that there are zero wait states on the flash memory for the GD32 whereas the STM32 requires 2-3 wait states - thus the GD32 gets a speed boost. That's not smoke and mirrors, that's better chip design. Without copying anything anywhere! But I'm sure the STM32 will continue its unabated reign of Blue Pill excellence, regardless!
@@RalphBacon Well, I am not some Of STM company fan. In fact, I am even using GD chips myself, because of the better price, ST cost $ 1.25 and GH costs $0.81 on Aliexpress with free shipping. That's a pretty decent difference. But your tests are extremely primitive and useless if you do not do a proper comparison. I am not defending STM design but proper testing would be way more useful than comparing default clock speed not just as a marketing ploy but as really valuable information. If you claim that GD made a better design you could compare their performance at the same clock speed and then later you can try overclocking them all to the max to test their capabilities. Measuring power consumption also would be useful.
OK, I'll see how easy it is to set the clock speed on either of these chips and see what I come up with in a future video. That said, the fact I desoldered one STM32 chip and soldered in another (the GD32) and it ran quite a bit faster is still a valid test, IMHO. You wouldn't ask BMW to disable the turbo when comparing it with a Ford, just because the Ford didn't have one, so the road test would be 'unfair', would you? You get what you get with the manufacturer's design. We'll have to agree to disagree (not that I _am_ disagreeing exactly!) and I'll do that clock speed setting research asap. I suspect the GD32 will still be faster due to the zero wait states on flash memory, but we'll see...
@@RalphBacon Thank you. In fact, I wanted to do such test myself so you will save me some time. Setting frequency is quite easy, you can even change it anytime so to avoid problems with peripherals you can even change it when doing the test and then set it back to print results. or if you set it to some obscure value you need to reconfigure UART and USB speeds again Well if Ford does not have turbo I do not ask to disable it on BMW but if both have a turbo and one comes with turbo witch on and another with turbo switch off from the factory I would ask to put then moth in the same position ;) You don't have anything to disagree because I also want to find out if GD is actually faster at the same clock speed than STM Also I just found out that GD has another advantage that it can use fractional frequency PLL multipliers so it gives you more frequency choices what is pretty important because you can adjust core speed without affecting peripherals. that multiplier usually happens to be the main problem when setting the desired frequency.
Thank you so much for pointing this out. I had to look with my video camera zoomed in to spot this. As a result of fixing it I also attached a few more components including the USB socket, and my SMD soldering skills are definitely improving. Great to hear from you.
@@RalphBacon Comes out perfect every time! I use the heat gun for most small items like that, otherwise, I use the drag method, which works great if you use a good flux.
Oh definitely, that will be my next project. I'm building a full blown Arduino Uno on a board just 1cm squared, using 0402 components. Gosh, I can hardly wait.
I didn't dig that deeply (yet). What was interesting is that the GD32 Green Pill used an 8MHz crystal (not 12MHz) so I'm guessing that the multiplier is set to 13.5 but maybe not. I know the zero wait states on the flash in the GD32 give a big improvement over the STM32 (1 or 2 or 3 wait states). It will all come out in the wash!
even a cheap binocular microscope is far better than the best usb microscope. a good binocular microscope and an SMD soldering station (hot air) is the only way I've been able to get good results trying to solder fleas (0603s) to PCBs. It's amazing that you were able to solder those 0603s with a soldering iron. Great skill.
Pure luck, Dirk. I was depending on several decades of THT soldering to see me through. I shall have to see what binocular microscopes are out there that allow room for a soldering iron.
Hey Ralph, Do you have fixed set of schematics for the GD board. I would like to put it together myself. Keep up the good work and pet the kitty for me
I do have everything for the Green Pill board, Skip, including all the schematics and PCB from r3bers. I'll put the schematics etc up on my GitHub. Stand by... It's there now.
I know it's an old video, but if you haven't bought a microscope yet, I highly recommend. I recently bought one off Aliexpress with a 7" LCD and even though it's a cheap piece of crap which overstated it's magnification level by like 1195x and will probably burst into flames at any moment, it's a game changer compared to the magnifying glass on my "helping hands" and my eyesight isn't even that bad (yet).
Yes, I did this some years ago now. In fact, I got a slightly larger one from Andonstar and I use it most days, even if to just read the names of certain chips! Deffo worth having. Oh, also great for checking soldering, with my eyes I need to double check.
If you don't know how to solder SMD chips it doesn't hurt to watch a few tutorials. Btw. it's actually easy to solder with an iron with a chisel or bent tip just by holding the chip down with tweezers in the correct position, applying a Rossmann of flux(aka a lot), picking up some solder on the tip, and just dragging it over the pins, relying on the flux to keep the pins separate.
Are we to tailor the sketch then to what each chip does best? Not a level playing field, which is why I changed to this sketch. I suppose I should really be using a Dhrystone/Whetstone sketch to calculate a proper VAX MIPS rating. As it happens I know the STM32C103 has a VAX MIPS of 48.81. In comparison, an Arduino Mega 2560 has a VAX Mips rating of 7.23. Other than that I don't know what it tests but I bet division must be included in there somewhere, surely? Anyway, benchmarks are really just a bit of fun and entertainment and give you a a good idea of what performance to expect in Real Life (a bit like car 0-60mph tests). That's all.
Hi Ralph, A microscope is undoubtedly the best way to go, especially if you have tired eyes like I do but, they are tremendously helpful even for young eyes. Once you try one you will see for yourself that everyone that works with PCBs should have one. All of them have an independent adjustable focus on one of the eyepieces and in my opinion is a must! Just a note on magnification: You will likely never need magnification over 20X for PCBs! In fact, 7X to 10X seems to be the most common magnification range used for our application. 45X and 90X is just too much for what we do (in my experience). You mentioned in a response to a comment below that you would want to buy a Trinocular scope so you can use a video camera at the same time. "Be aware that Trinocular scopes only allow one of the eyepieces to be used while using the video camera" The only scope that allows you to use a video camera "While" using the scope with both eyepieces at the same time are "Simul Focal" stereo scopes! Simul Focal scopes aren't really more expensive, there are lots of deals on the net for Simul Focal scopes. Check out this video for an demo of using a Simul Focal scope as well how to successfully drag solder while using the scope and.....you'll see just how much flux he is using for the process. ua-cam.com/video/C_eQrbop-J4/v-deo.html You have received lots of great tips for soldering SMD components. It's obvious that you aren't willing to spend the money for the "Tools You Need" to do the job (which is kind of ironic since PCBWay is investing in you and your channel). You're going to have to bite the bullet at some point and "Invest" in the right tools for the job. I have found that one of these circuit board holders is also a must, this one allows for rotating the board so you can see the back of the board without having to remove it from the holder and it's pretty cheap (worth every penny IMHO). It helps a lot to be able to steady one hand with the other hand but to do that, the circuit board has to be held in place by something. This holder works very well under a microscope as well. www.amazon.com/Aven-17010-Adjustable-Circuit-Holder/dp/B00Q2TTQEE/ref=sr_1_3?keywords=PCB+holder&qid=1563118302&s=gateway&sr=8-3 Anyway, Best of luck to ya Ralph!
Wow! Thanks, Joe, for all that information. In no particular order, that holder you mention at the bottom of your post, I already have one, for a long time. I bought it to, ironically, be able to solder THT components and then flip the board over to insert the next one. I never thought to try it for an SMD board. My "Strange Third Hand" tool is a blessing though. But I'll give it a try. Regarding my toolset, or lack of it, I'm in the process of buying a pretty simple $40 USB microscope that many have said they're OK, and many have said, Don't do it, you need binocular vision. But, I'm testing the waters to see how far I get. The reasoning is that I tried using my video camera as a USB scope (very enlightening to zoom in) and it "sort of" worked. The monitor is as the wrong angle and position but the principle made me think I have to try it first, before spending anything like what they want for a true stereo binocular scope - a _simul focal_ one at that, thanks for putting me right. That said, the microscope's camera is eye-wateringly expensive. Anyway, at 1/20th of the cost of a 'proper' tool, this USB 'toy' will be an interesting testing of the waters - and I might be able to sell it on (or give it away) if I decide to go big time. My sponsors don't pay _that_ much so it will be my own money for this! As a result of another comment telling me I had a short between two pins on my swapped out STM32 board (I had, who knew?) I've just been practicing all the techniques and guess what? Yes, I'm definitely getting better. I've cleaned up the chip, soldered a few more components, even the USB socket, all without _too_ much trouble. Exciting times! Thanks very much for all the info, it's great that so many (including you) have taken the time and trouble in giving me their slant on things. Most appreciated.
@@RalphBacon Do you plan to begin showing viewers your soldering process? Most of what I've seen from you is "The results" which is great IMHO, there are plenty of other videos showing the process. If you are not planning to show viewers your own soldering process, you could buy a simple 10x to 20x stereo scope just for soldering and inspections off camera, not to mention the fact that a decent scope is great for much more than soldering. I'm not kidding, a good microscope is THE SINGLE MOST IMPORTANT tool in the shed, especially for tired eyes like we have!! Here's a good scope on Amazon that would be perfect for just soldering and inspections at $202.95: www.amazon.com/AmScope-SE400-Z-Professional-Microscope-Magnification/dp/B005C75IVM/ref=sr_1_7?keywords=stereo+microscope&qid=1563135891&s=gateway&sr=8-7 The best arrangement I've tried so far is a 1x objective with 10x eyepieces, the scope I just posted the link to has that very arrangement (along with a set of 20x eyepices for really close up inspections). That arrangement has the best amount of work space under the objective lens I've seen (check out the specs on that scope in the description) You can also buy an adapter for most any camera so it's not etched in stone that you have to use a camera made specifically for scopes (If you intend for sure to film your soldering process).
Hi again Ralph, I feel I should be more clear about my suggestion to buy a stereo microscope without the camera port. It was merely an observation that you usually don't film your soldering process. I was only trying to help you figure this out by voicing an observation. I too am one of those guys that only spends money on tools I feel I really need. For what it's worth, my decision to buy a microscope came from a friend who offered to let me borrow his microscope for a few days, I literally could not believe how helpful it was to be able to find the issue and be able to solder while viewing the board through the microscope. It was one of those "WOW" moments! :-) The issue I was having at the time was a really tiny strand of coil wire that had broken. I couldn't see that wire strand even with my WiFi microscope, but with the borrowed microscope I was able to re-solder the broken wire while looking through the microscope. Anyway, if you do plan to start showing your soldering process, then by all means, hold out for the Simul Focal stereo microscope. Again, just trying to help :-) Joe
Tell me, what kind of plate do u have ontop of the table? I mean the green plate with those squares. I didnt found any similar on ebay. nor on internet at all, may be because of wrong search-text.
It's a *self-healing cutting mat* and I have several. A3 size for normal use but my soldering station has a huge A1 size (900mm x 600mm) to protect the bench. You will see it in a future video but looks pretty much the same as the smaller one. Example of A1 size: www.amazon.co.uk/gp/product/B078XCKDHT
Umm in my ECE lab we used microscopes and regular old Weller irons, you start off by lin in ng up the chip, solder two corners, and then go arround getting the rest of the pins, to fix any bridges you just clean your iron tip and wick it off or over into the other pins
16:17 The 330uF is too large! It is true, that regulators cannot follow varying current demands at the speed required by some modern controlers and therefore they brown out. But 330uF needs to be charged first and that might overload the regulator already at startup. It would be better to have a much smaller one (10uF...22uF) as well as a number of 100nF (ceramic) in parallel, to lower the effective series resistance of the current reservoire on the output of the 1117-3.3. So what died might be the regulator, since the value of 330uF is quite demanding when initially asking to be charged to 3.3V ... that can draw quite some current through the tiny regulator thingy. Perhaps try to exchange that regulator for some replacement of similar type ... any "xx1117-3.3" might do.
I removed the 330µF and used a smaller one, I _think_ it was around the 100µF mark but I'm guessing now. Whatever, it worked just fine and continues to work so I guess that suggested large component was never going to work!
@@RalphBacon Well, it might at least turned off the regulator by its thermal protection when charging the large electrolyte capacitor on its output. Bigger does not always do good... But some ESD-protection o the 5v power input is missing isn't it? Perhaps something like CG0603MLC-05E(bournes) or a B72500D50A60(epcos) or at least some zener like 1SMB5918BT3G (on-semi)?
@@RalphBacon I've worked in electronics for 45 years and didn't know this until I saw it in a UA-cam video the other day. Of course I was of the days before surface mounted chips took over completely. I still like a hole. I like your show, keep sharing, its fun. My HA sytem was designed in the mid 80s and is still running 24/7. It would have been nice to have had a few of these Arduino boards. Mine was designed around a WIZ ITC232 controller which is still available to this day believe it or not. This may be TMI but the cost of energy for that 2700 sq ft home was $39 a month for the first 15 years due to the HA system integrating passive solar and a hydronic heating system. All Good Things, Cliff
Each core runs independently. So, in parallel. Thus, we could have _two_ tasks running at the same time to do twice the amount of work but it would not complete each task any quicker.
Well, its sole purpose is just to get the processor to do some work so we can measure it. The previous sketch used the sieve method but ran so fast after I removed the float I couldn't get a measurement! Zero milliseconds. Even the UNO did it in about 3ms. Glad you like the videos, thanks for posting.
Ralph S Bacon I really do enjoy your videos and appreciate the time you take to make them. They have inspired several projects here and helped out in several others. I hope I didn’t offend you and if I did I apologize. One question: where did you get the gerber files for the nano?
Hi Ralph, the green pill looks like a fun project... just wondering if you thought about hard wiring extra functionality to the board, like maybe a temp sensor, canbus chip or opto-isolated inputs?
SOLVED IT Cost me a few hours but I have a boot-loader working on stm32f103c8t6 blue pill. I can download programs but even better I get serial debugging from USB also. The 10k to 1.5k was all a wild goose chase you can get it working without any hardware changes at all. For those interested here is what to do. - Use ST-LINK to put boot-loader on blue pill. STLINK flasher and driver are on the ST site for free. -You have to match the correct boot-loader via LED pin i.e.builtin led on my board is PC13 so I used Generic_boot20_PC13. -Don't make my mistake move the jumper back to Boot 0 before reset or the boot-loader get erased. -The official ST core will not work with the boot loader it is not supported. use dan.drown.org/stm32duino/package_STM32duino_index.json. All working for me now shows as "Maple Serial (COM15)"
I did this countless times on different boards during my R&D for the video. How come it works for you and not me? Life is so unfair. Strop mode starts...
@@RalphBacon Hmm, in that case it must also involve one of the many things I tried along the way that i didn't realise, was important: I did at one stage follow the manual instructions from RogerClarkMelbourne github.com/rogerclarkmelbourne/Arduino_STM32/wiki/Installation which included: -Installing Arduino Sam boards but I assumed it was just for Arm tool chain stuff. -I also ran his maple drivers Install_drivers.bat that you don't do with the json version unless it is automated. But I am pretty sure it got replaced by Zaidg driver install The only other factor I can think of is hardware, I bought 4 6 months ago so a different batch to new onces and all are stm32 not gd or CW etc. You know the device is working properly when you power them up and they power up as DFU device then a second later disconnects and become a Serial Port. Also when you did the zadig driver install when one did you install? I will check which one I had in the end, its working on two win10 machines and all 4 clones now. I think it would have worked earlier for me but there are two pc13 bootloader files and I think I grabed the GD not generic one. Good luck.
This clears a lot of confusion!! I'm trying to replace STM32F103CBT6 from one of our products due to chip shortages and price hikes. Since I'm getting the PCBs from JLC, I am looking for IC supplies from LCSC. Now, LCSC does have GD32 F 103CBT6 (Cortex M3) in stock right now, however, they have GD32 E 103CBT6 (Cortex M4) at a very good rate and in large quantity. Although I've ordered 10 pcs of E103 from LCSC so that I can discover myself, but I'm just dying to know these: 1. My board currently has an FT232 on board. Can GD32 E 103 be programmed through the HW bootloader on Serial1 ? 2. What config changes will I need to make in arduino in order to build? 3. What config changes will I need to make in arduino in order to Upload? 4. And the most important one. I use FT232 earlier and it gives me data rates up to 2Mbps. Which is very much required by my application. Can the USB be used to program in the bootloader and also simulate Serial-over-USB while in the application? Will it give me at least this much bandwidth? (Some forums claim that with some buffer upgradations, it should go as high as 8Mbps, is it true on the standard bluepill)?
I'm afraid I just don't have that information to hand; I'm hoping you will either research it or find it out during the course of your experiments. Good luck with your project.
Make sure that you wear breathing protection when soldering small SMD. 😎 I accidentally ordered metric 0402 resistors instead of imperial. Only 100, so it wasn't an expensive lesson.
And you didn't solder them on, Bernd? All you needed was a steady hand an an electron microscope :) Regarding ventilation, I have a small USB desk fan blowing the fumes away from me but it's not great.
Hello Ralph, I hope you are doing well. STM chips are manufactured in many different countries like Korea, Malaysia, Philippines, Taiwan, and China. And I am thinking that GD, CKS, and CS are a few of these Chinese manufactures. So they don't have the need to clone them.
I don't think those chips are exactly the same as a genuine STM32, though, Georg. There are some differences (as noted in the comments on this video). The GD32 seemed most like it, and for general use the others will probably work too. What I would do is design a circuit with a genuine STM32 then swap it out for a "clone" and see if it still worked!
@@RalphBacon I've just took a look in ST's application note AN2606 that the STM32f103 supports only SUART boot. The STM32f105/07 support CAN boot, DFU boot, and so on. I think I will compare the data sheets. That may will take a while. Best regards.
Nope, I don't, Pete. Whilst solder paste works well for (a) fabrication houses and (b) if you have a stencil so you only put the paste where it should go rather than all over the place, I find that standard leaded solder works best (for me). Use copious amounts of flux on the SMD component area (which also helps stick it in position), optionally hold the component with sharp pointed tweezers on the top, and then just touch the joint with some solder on your iron. Soldered in less than a second. You will see a close up of how I do this in two videos' time using my wonderful HDMI microscope, so don't miss it!
The CKS103f is not fully compatible with the STM32F103C8 Overclocking at 128 Mhz didn't work. The library for 7 segment display TM1637Display.h also didn't work. LCD library LiquidCrystal.h didn't work properly and displayed many incorrect characters. LiquidCrystal_I2C.h did work. I tested bij swapping a number of cks blue pills with stm blue pills in a soldered circuit.
Hmm. I wonder what the underlying issue is. Registers? Hard-coded pin definitions? Something else..? I guess we shall find out as we progress. Thanks for posting your findings.
Great video Ralph- Q: does the STM Maple Mini require toggling the small yellow bridge pins after programming like the Blue Pill does? I think moving that bridge each time would drive me crazy!
No, the Maple Mini has extra circuitry so that the USB is used to upload the sketch in exactly the same way as an Arduino Uno or Nano. No jumper changes required!
Actually, this video is more useful today because of the chip shortage. Right now on Canadian Amazon the originals can go for over $100. Next day delivery is $22. If you can wait a few days, you can get 3-5 for around $30. The clones are no-brainers right now and can easily be returned to Amazon if needed.
The clones are not too bad, although I would not use one without prototyping the entire project with one to ensure compatibility. If that gets the thumbs up then it's good to go!
I've bought 3 stm32 boards. On 2 of them, when I selects boot0 to 1, the blink sketch continues working and can't program via serial. I needed to buy an stlink.
@@RalphBacon I've got an ftdi. But 2 of my 3 bluepills can't be programmed via serial. Boot0 jumper does nothing. Check this video: ua-cam.com/video/z-AR7Xl8pA0/v-deo.html
There are only four wires to connect but you must connect them correctly, of course. VCC and GND are self-explanatory, the SWCLK and SWDIO go to CLK and IO on the Blue Pill end connections. Make sure you use the correct voltage pin on the ST-LINK. I used 3v3 and connected it to the pin marked "3" on my STM32F103 device.
Indeed, sounds good. My main issue was that my STM32s were fakes (clones, not perfect ones). As soon as I bought from a reputable supplier my bootloader issues disappeared.
I am still looking for a way to add the necesary files & packages to my IDE (without being connected to the internet!) to get it working. Just the way I do with my libraries which is very very easy. Is this possible? Just downloading some packages on a USB stick and copy them into some IDE folders. I still hope so.
Creality replaced the STM32F103RET6 with the GD32F303RET6 in the Ender 3 Pro. Is there a possibility of extracting the source code from the board? I am not knowledgeable about this topic. Anyway, I like the explanations in your video and the way you present yourself. Keep it up and who knows, maybe someone or I will find a way to figure this issue out.
You can read the contents of the μC (the uploader does this to verify the upload) unless the fuse bit has been set to prevent this. How you do this on the STM32F103 escapes me but must use avrdude command line. Whether the code is going to help you (it might be very processor specific) is another question - and it's in hex.
JLCPCB will supply SMD components on a board (one side only). You pay for the components as priced on LCSC.com (where they/you source the items from) plus the assembly fee (which is on their website). For small boards with few SMD components it's possibly not worth it. But a fully populated board with lots of closely packed SMD components it could be a life (or project) saver.
I've not examined the quality, Olaf. I do know that at higher resolutions there may be some "noise" but the only way to find out whether that's a problem is to try it out.
hi there hay just wondering on avage what micro..... do you tend to use the most.. and one more i tend to used the uno in most in my projects (i just like it and pretty much has just the right amount of goodness for my projects) but what i do find a real pain in the ass is the fact that a lot of things tend to like 3.3 volts and not 5.0 volts if you hit this problem often what do you do to get around it i been using resistor as voltage dividers which seem to work ok but a real pain what's the answers lol.
Good question! If Wifi is required then an ESP8266 or ESP32 is simplest to use, although you can get some hybrid boards with an Arduino Uno and an ESP8266 WiFi module all on the UNO shaped board - but I've not used one. The Wemos D1 Mini ESP8266s are good and reliable and you can plug things together easily all on 3v3. That's what I used for my Home Alone project. For non-demanding projects I will use a Nano or a Arduino Mini which are basically Arduino UNOs in a smaller form factor. UNOs are great for developing but far too big for actual projects IMHO. If no WiFi is required but a faster UNO-compatible board is needed then the STM32 Blue (or even Black) Pill is great value and so very fast. There's even a way to update or remove the bootloader now. To interface 5v to 3v3, a resistor divider works (usually) but is fiddly; I have used bi-directional 4-way or 8-way level shifters in the past and they work really well but need a lot of connections on a breadboard but are a no-brainer when used with a PCB. I mention these in a video when I did on RFID Reader/Writers: ua-cam.com/video/g3a0X30Le-8/v-deo.html
Wait, I am confused. You run code which counts how many primes you have found, but finding the next prime should take more computational power than finding the previous one. Then how come the number of primes the ratio (primes stm32 found)/(primes arduino found) is much greater than (clock speed stm32)/(clock speed arduino)?
I give up, what's the answer? Forgive me, Beacon, this was just a rough and ready way of showing the STM32 was a gazillion times faster than a standard Arduino. I need to use the Dhrystone method in future which has something of a scientific basis!
Ralph S Bacon I just realized the reason, I’m dumb and forgot that arduino runs 8 bit and not 32 bit, if you take that into account everything makes sense :)
*====================================================*
*If you got that far in the video: NO BOARDS remaining! ALL GONE*
*====================================================*
All gone, sorry.
A couple of months ago I created an STM32F103 design and inadvertently put a GD32 in it. Inadvertently because these were counterfeit chips: the GD32 markings had been removed and replaced with STM32 markings. After writing my firmware to the GD32, it only half worked. Nothing on the I2C bus was responding. I wrote a test application that toggled the I2C wires as GPIO pins to be sure I didn't have the lines bridged or something. In GPIO mode, the I2C lines toggled fine but they did nothing in I2C mode. After acquiring genuine STM32 parts, everything on the board worked.
I never solved the problem, but it did drive home one very important point: the GD32 is not a drop-in replacement for the STM32! I guess the moral here is: don't declare victory if you've only tested GPIO. It's comparatively easy to get GPIO right, but I2C, SPI, ADC and DMA state machines are much more complex. The I2C incompatibility of the GD32 is well known and it was a mention of this incompatibility problem on the stm32duino forums that led me to suspect, test, and verify that my first STM32s were in fact counterfeit remarked GD32s.
To be sure, having a standard test harness of some kind would be useful with these chips. I discovered that all the problems I had with the Blue Pill STM32 not connecting to USB were caused by the *fake* STM32 chips! I acquired some genuine ones and guess what? Yes, they connected to USB exactly like everyone else's.
Thanks for raising this important point, good to hear from you.
Maybe this has been said but I don't want to wade through 80 comments to find out. I have used both paste and drag soldering on fine pitched chips. With the solder paste a good deal of the paste is flux. When heated the little metal balls melt and migrate to the solder pads with none remaining outside the pads. This works even without a solder mask. Your friend capillary action at work. Should be UA-cam content on drag soldering. Good video
Well, I've had a lot of different suggestions, Dan, including this method and variations thereof. I don't know why it was so difficult. Probably because it was so difficult to see what I was doing. I shall practice some more, thanks for the suggestion.
For the chip. Solder just one corner pin first, get it aligned , do not put solder on pads etc. then solder opposite corner, do not worry about extra solder or joins between other pins. use your iron on an angle put flux on the pins first, then flow a molten ball of solder down the row of pins. then to clean up use 2mm solder wick on the END of the pins to suck the extra solder down and away, moving the solder wick and iron together, no pressure on the pins. for the 0603 components use flux to stick them to the pads, then use a trimmed ice cream stick leaning on the component with a small weight, (I use the solder roll) to hold it then touch a very thin solder (I use 0.56mm) to the end of the component touch the pad and solder with the iron remove the solder and approach the component, and it should flow up the end. slide the iron back away from the component and pad. Takes practice. Other wise, use the flux again to hold the chips, and put a tiny bit of solder paste on the pad with a syringe, and use a hot air pencil to flow it. Either way clean up the flux with IPA after.
You've described more-or-less what I did (badly) with the STM32. I discovered you need LOTS of flux and when I re-did this yesterday it worked MUCH better. Regarding the 0603 I don't think I want to use them again, too small. And if you get the caps mixed up it's the devil to measure them all to find out which is which! I guess if it were that easy everyone would do it, right?
I coat the pads and legs with *liquid* flux, then center the chip, and leave it dry for quite a while to paste it on there ... then do a couple of corners lightly, and so on. Plain rock-hard resin, melted on the solder tip, can help the solder bead on to the legs and pads.
Damn. I have not the time to watch your videos now. But I save them all in my list "autumn projects". Keep 'em coming. Thanks.
Yes, when the winter nights start drawing in, these videos will keep you company, Pekka!
Hi. If you have problems in soldering these tiny IC's on a board (like me too), just let JLCPCB solder it for you. They have very good component prices too. Now I have no more fear in using SMD Components like 0402 capacitors or resistors in my designs.
0402 components are not my choice of size, they are but specks of dust. 0603 are OK and 0805 are relatively huge so easy to solder by hand. Under normal conditions when/if my eyesight returns! I've used JLCPCB's assembly service for SMT components (even did a video on it) and they are very good, as you say, Poppo..
Your comment on soldering. I have those days too. I just can't solder at all. Especially SMD stuff. After burning myself twice I usually find something else to do that day and leave the soldering to when I remember how to. Other days the solder flows like a glove every time as soon as it's touched, nothing sticks to the iron and everything ends up straight!
I've soldered through-hole components since I was 14 years old (I'll let you work out how many years that's been).
I'm relatively new to SMD soldering but have got a good enough grasp on it now that I actually prefer it, due to the smaller size of components (so a smaller PCB), no lead cropping and it just looks so much neater (and some components are now only made in SMD format).
@@RalphBacon For SMD stuff I bought a load of 1.99 solder practice boards of Aliexpress and sat and did them until it was easy.
That said, I just botched an LPFQ64 soldering. I dont think the chip will have survived. A reminder I need to practice again it's been a few years.
I have sometimes been successful with the drag soldering technique, where the tip is loaded up with solder, and dragged across the pins. You need a decent soldering iron and a good amount of flux, but when it works, it's great.
You can also do reflow soldering using the hot air gun and just putting down a line of solder paste across all the pads. The solder only sticks to the pads and the legs of the chip, and surface tension tends to pull it all into place. You do have to do a visual inspection, or check with a multimeter, but generally a slight touch with a soldering iron will get rid of any short circuits.
I've been practicing with the drag soldering method and it works well, even more so now that I have decent soldering iron and a soldering tip called a Gull Wing, which is slightly concave and holds a reservoir of molten solder, specifically for this kind of soldering. I've even used my Hot Air Gun Soldering station with good results. Practice makes perfect! Thanks for your post, good to hear from you.
When surface soldering try putting a strip of paste right across all the pins (right angle to the pins). When you place the chip and hit it with heat the paste will melt and suck into place.
I have to test this out more, Digger, more in a future video!
Hehe, the toaster method. 0.5mm pitch pins are real soldering gymnastics.
Your last two videos have encouraged me to dig our a number of Blue Pill boards I have stored away. It would appear that since V1.6 of STM32 Cores was released that all uploads require the installation of STM32CubeMX this provides the command line API to allow STLink and Serial uploads to the board. Another nice feature is the compiler setting to use USB Serial as the default UART thereby freeing up the serial GPIO pins for other uses, I've had no luck installing the Bootloader on these boards so I'm using STLink.
Keep up the great work, I look forward to your video every Friday as they are always interesting and informative.
I'm really glad it's inspired you to dig your old Blue Pills out. I waited quite a while before jumping into the STM32 pool, to let the technology mature enough for it to be useable for beginners. Now is the time! Good luck and thanks for posting, Dave.
Awesome video yet again Ralph! I'll be procuring me some of those GD clones for sure.
Experiment away, Jacob. You didn't want a (Green Pill) board then? Or did you not watch my video to the end? Shock! Horror!
@@RalphBacon haha im not in the UK I'm in Australia
It's time for a following-up. There are a lot of new stm32F103s clones, some have more flash and RAM and much higher clocks.
I'll give it some thought, I must refresh my knowledge on this device.
You need reflow solder paste and a heat gun or toaster oven to solder down small parts like that. Once the solder melts, removing the part is very easy. Unpowered, electronics will tolerate heat of melted solder quite well. A small heat gun will cost $65 or so and is highly adjustable. Toaster ovens are very cheap, but they don't regulate tempurature as wel las a digital heat gun will. In either case you are applying heat to the board until the solder melts. Be careful to only touch the part you want to remove. You'll work inside the toaster oven so the board stays hot and remove or place the part. From that perspective the toaster oven is less convenient. The heat gun since you are in open air makes accessing the board easier, but you have to keep the heat on the board so it doesn't cool down.
For reflow soldering a new blank board, put solder paste on all the connections. It's somewhat sticky so you can use the paste to stick down the SMD components to the correct positions. Heat the entire board in the toaster oven or play over it with the heat gun and the solder will melt and the parts will pop into place into the solder pools. It's much easier than trying to solder down tiny parts with your iron.
I've got a heat run (sorry, digital soldering station!) and it works well, Richard, but not so great for actually soldering stuff onto the board (great at getting it off) as I am far too worried about all that heat being pushed at those components. Maybe I should not worry so much given what you are saying about that toaster oven - sounds quite tricky! Thanks for posting, good to hear from you.
@@RalphBacon Consider how those components got soldered down originally. They were done in an oven. The components were roboticly placed on the board and then the entire board and electronics were heated to the point where the solder melted. The parts were "stuck down" with the aid of solder paste until the solder had melted around them. Call it tricky if you like, but this IS how it's done. I do a fair amount of reflowing and I can tell you factually that the electronics survive this process quite well. If you are still skeptical, watch some videos on reflow soldering.
For simple stuff, like reflowing an LED to a star board, you can use a frying pan and your kitchen stove. I know I have many times for LED's. For more complex stuff like a populated board, you need a way to hold the board so it heats uniformly and doesn't sit on or touch anything. A heat gun such as this will push very hot air onto the board to the point that the solder melts and then any component in the hot air path will simply lift off the board. A lot of engineering goes into making the electronics heat tolerant so that the solder mounting them can melt.
www.amazon.com/DEWALT-D26960K-Heavy-Duty-Display/dp/B007X2ATZE/ref=sr_1_3?keywords=dewalt+heat+gun&qid=1570992759&sr=8-3
@@Lamster66 Sure...strawberries and apples are both sweet so they are the same thing...lol
Solder is not all the same...not by a long shot! Buy some chinese solder and then come back with that argument.
I've done enough reflow soldering to say that solder paste works best. I have used solder and yes...it can be done, but you get loads of solder bridges while a thin coat of solder paste does not typically bridge.
I did say use a heat gun or toaster oven..
@@Lamster66 OK...well you go and have a nice day alright?
You want to argue and I see no reason for it.
@@Lamster66 You just go have a nice day alright? I'm not even remotely interested in arguing with you.
Congrats...on winning absolutely nothing that matters!
I hope that settles you right on down into a reasonable human being.
Hey Ralph!
The "O3" compiler flag is activating all optimizations for speed, e.g. that does stuff like unrolling loops and the like, to gain the most execution performance, even when this takes up significantly more program space in the outcome. The more moderate in regards to program space is "O1", which tries to minimize code size if possible. And "O2" is somewhere in the middle, where some optimizations can cause some increase in code size. --- Best wishes from Berlin!
Servus Sebastian! In which case, if the -O3 optimised version fits in the available space that should be the one to use - otherwise it's just wasted Flash space, running a relatively slower version of the code, right? MfG aus MK.
@@RalphBacon That is correct of course. Unused flash space could always be traded for greater speed. Beste Grüße aus Berlin!
*Thumbs up*
I got GD32 chips from LCSC as well. First idea was to design a custom board with different kind of external components for testing I2C, SPI and other things. After couple of design ideas, I decided to take advantage of the breadboard friendly BluePills that I already have. I also replaced chip on board with the GD32.
There are two ways I like to use when soldering fine pitched SMD ICs. I think first one is easier, if one doesn't have good eye sight or hands are shaky.
1) Add thin line of solder paste (Doesn't cost much on eBay/Aliexpress, if you don't already have it) across the pads, place the IC on board and apply heat with reflow station / adjustable hot air gun. Alignment isn't too important, surface tension will take care of small errors in alignment.
or 2) Add flux to the pads, place the IC on board and tack it in place with tinned soldering iron. Go on and touch each leg with soldering iron (dragging, or "poking" one-by-one, which ever feels more natural), adding flux and solder when needed. Solder bridges can be cleaned by adding flux and touching them with clean soldering iron. Larger bridges can be cleaned with solder wick.
And of course there are reflow ovens and other ways to do it, but those two are simple methods for simple tasks of soldering one IC.
I'll give the reflow oven a miss, but your other method is not one I've tried with the solder paste to I will that a whirl too, next time I solder one up. I now use much more flux and it works a whole lot better!
Thanks Ralph and LCSC.
I didn't know the GD32 was actually a viable alternative. I looked into that chip around a year or two ago, found some forum posts that said it had bootloader issues, and didn't go any further with it. Now I know better, thanks.
While I'm no expert on soldering, I've soldered chips on several boards with TTSOP, and TQFP 32/64/100 chips. If I have a board I am going to etch several of, I make a solder mask and a stencil first. I make my stencils with a similar process to my single sided boards with a photoresist on a piece of aluminum cut from a soda can. I only use solder paste if I have a stencil. That will change soon though. I will be forced to use paste with a project that has my first QFN footprints/chips in the near future.
I haven't had much luck using hot air with larger SMD chips or anything without a stencil. I actually use my main, large chisel tip on my iron.
First a bit of background. I have the old RadioShack "Digital Soldering Station" still. It's an Atten 60w station that was just rebranded for R.S. It is a Hakko 900 series-ish station with a 2 wire element instead of a Hakko 900 type 4 wire, but it has an identical handle and form factor otherwise. The important bit is that it is a true 60w and uses Hakko 900 series tips.
My main tip is a genuine Hakko 900M-T-3.2D. I got a bunch of them on eBay second hand NOS in the original individually sealed Hakko packaging. That tip makes a big difference for me in a lot of different circumstances like this. I have tried several dozen tips, many were clones and the last dozen or so have been genuine Hakko's. I have several fine pointed tips that I will use if I am forced to, but I use my main chisel tip 95+% of the time. This tip holds an enormous amount of thermal energy compared to all the other tips I've tried. I usually set it to 575F (302C) as my general setting. This is hot enough to work on almost anything. It is also hot enough that intentionally bridging standard 0.1in pins using good (MG Chemicals No Clean leaded 0.8mm) solder is nearly impossible. My low temperature setting with this iron for stuff like cheap perfboards and bridging pins is 505-515F (263-268C). All of my other tips I use regularly require settings that are around +100F (+50C) higher than this tip to achieve the same results.
I credit this tip's thermal mass as the biggest contributing factor when it comes to soldering fine pitch SMD stuff. I also use a block of Sal Ammoniac. That's the traditional name for a block of ammonium chloride. This has traditionally been used as a soldering iron tip cleaner. It will remove any oxidation from a tip. I think of it like soap for a tip. If you rub the tip around on the block with a bit of solder, then clean off the junk with brass turnings and a wet sponge, when you put fresh solder on the tip it will be just like a brand new tip. This is also the condition when real Hakko tips do really well IMO. I always do a cleaning routine like this before soldering a large SMD chip.
I start by applying a lot of Amtech 559 UV flux to the clean footprint. My rule for flux is: "Too much flux = I can't see the footprint any more. AKA There's no such thing as too much flux."
Wow! Thanks for such a detailed reply, Jake. Seriously, I'm very grateful you took the time to give me your experience. I could imagine in my mind's eye exactly what you meant too, as I was reading it. I think perhaps, that I just don't have enough experience at the moment and need to practice more on those Banggood practice boards.
One thing I discovered to my cost it to *not* press the chip's pins using the iron or you will end up with bent, misaligned pins. So the dragging is just for the solder, nothing else! Experience lesson #1.
I am thinking of getting a better iron than my Antex, which has served me well for 20 years or more but is not up to SMD soldering (and certainly not desoldering, not even using soldering wick to clean up afterwards, for which I have to use a much bigger, hotter iron). The range of tips for my iron is small, I'm not even sure they make it any more.
Anyway, I will re-read you post a few times and see what I can extract from it to improve my soldering (although I did get both working so I'm not too disappointed!).
@@RalphBacon
Getting it working is the important bit. Sorry if I'm too wordy. I'm not the brightest and feel like I need to explain too much to get my point across. My software code is no better.
If I were to get a new iron it would probably be a T12 clone unit. Voltlog has some good uploads reviewing it. Those T12 cartridges with the integrated element and tip make a big difference.
I didn't mention it earlier, but on my soldering station I actually made my own PCB's for the part that goes inside the handle. I did this to control exactly how long the heating element extends into the tip. I also tried to offset the ceramic element slightly so that it sits closer to one side of the tip. The Hakko 900 series tip bore is about 1-2mm larger than the ceramic elements that go inside of them. I made sure my iron has maximum thermal contact between the tip and heating element. That made a big difference. The original PCB/element didn't extend to the end of the tips.
I'm about to add a dual iron setup too so I don't have to take things apart to change tips, but that's a whole nother topic.
If your iron has enough thermal mass (or total power if you've got several thousand dollars for something like a JBC) you won't have problems with anything sticking to the pins, solder braid, iron, or otherwise. If you want something nicer than the T12 with it's 70w rating, watch the uploads from Marco Reps about his JBC setup. He only bought a JBC handle/ cartridge and built his own power supply and controller. It's one of the few DIY projects you can watch and see the CC actually using in later projects and uploads. Marco has been sent a sponsored T100 and used it before, but you will see the JBC most often in peripheral use in other content he has uploaded when he's recorded more serious bits. He also references a killer open source soldering station project on dangerous prototypes' website in his uploads about his JBC project.
-Jake
I shall check out all those pointers, Jake, thanks very much and no need for any apology over comment length, it was (and is) very interesting.
You may find this interesting Ralph. I have this STM32_Nucleo F0 board and I have played around with it in the past. I used the STM development software Cube and HAL to program the board because at that time the only Arduino core (wrapper) was for the 103 chip. That Arduino core is known as STM32_Arduino core or "Rogers core". Your last video made me pull out the board and program using the Arduino extensions through Platformio. VSC/Platformio IDE uses STM32duino core. This core supports many more boards Nucleo, disconvery, and chips, F0, F1,F2,F3,F4,F7. The STM32duino was written by ST themselves and is now considered the "official" Arduino core. The official core uses HAL where as "Rogers" core (STM32_Arduino core) uses the Standard Peripheral library. I think in this video you are using the Rogers core which many people still use. No need to use Platformio as the STM32duino core can be installed into the arduino ide . The ST official core became available in May 2017. ST supports the core with a ST employee on Rogers STMduino forum, crazy days !
Interesting, Joey. I read about the Cube and thought I would leave it for another day. Guess what? That day might have arrived. thanks for posting.
Dragsolder the chip with enough flux works really great. Have to use a proper solderingtip though
Thank you for your feedback, Trailtraveller, noted and appreciated!
Check out a program called Zadig. It allows you to specify which USB driver is loaded. An alternative would be ImpulseRC driver loader. The Naze (pronounce Nah-zeh in a
Japanese manner ) bootloader is what was loaded onto early multicopter fight controllers. Windows has a nasty habit of overwriting the USB driver so if stuff suddenly stops working that may be the issue.
I downloaded this (thanks for the heads up, Rob) but it could not install anything as it said the driver was already installed. I could see no way of 'forcing' it to use a particular driver for the STM32 (which has a weird, ie invalid, VID of 0000).
You need to select an appropriate driver from the combo box. The reference below relates to OpenTX used on the Taranis RC transmitter which uses an STM micro. STM is also popular on many other RC related hardware products. open-txu.org/home/undergraduate-courses/fund-of-opentx/using-zadig/.
17:40 It is pretty simple to solder that stuff. Actually tehre are 2 good ways to do so. One is with your hot-air gun, apply the paste generously across the pins, do not worry about the single pins - just lay a small saussage across each side across the pins on the board, place the ic onto it, about where it would sit when soldered in correctly and heat the stuff up. The solder paste will draw to the contacts on the board on its own and with a tweezer you simply make sure the ic goes into the right position by rudly but gentil pushing around the ic to keep it in the position you want it to be afterwards. --- the second way is almost as simple. Take some solid solder and put the solder with some liquid flux onto the board, but without the ic first. When all connections have soem solder on them put the ic on top, heat the stuff up and use a tweezer to push the ic around to fit the right position when the whole stuff becomes liquid. --- In both ways the solder substance as well as the ic want to get into the right position on their own anyways, since there is a certain addiction caused by surface tension of the solder pads and the solder substance. Just make sure there is enough flux in play to lower the surface tension on the pads on the pcb. --- There is also a third method working without reflow oven, which relies on using superglue. You simply glue the ic in the corret position and then apply paste onto the pins like a saussage on each side of the ic and heat the thing up to make the paste become liquid and find it's place. Excess is removed afterwards with a desoldering braid (?) ... [Entlötlitze]. Do not worry to have pins soldered together at the first try ... it doesn't matter. Just solder the thingy into place and remove the excess afterwards...that will do the job.
[edit] So do not worry to get the paste onto the pads. It doesn't matter at all, which also makes stenils pointless for small series done at home. If there is too much, get the excess off afterwards with a soldering iron / heat gun. Not worries about placing solder paste. It pointless fumbeling about effort at the end of the day ...
Yes, I never understood how a stencil would really work for a single hobbyist board, Sebastian. The solder paste sausage has been suggested by others too; I'm waiting for an opportunity to test it out. Like most things, it's trial and error and lots of practice. And when you can do it you have no idea why you couldn't do it before. It's all about creating new neural pathways in this aging brain of mine! Thanks for the info, I shall practice soon, I suspect.
@@RalphBacon Have a nice weekend and lots of fun playing around with the soldering then!
Btw. stm32-family is a nice choice to go for. Well done!
@@RalphBacon Well, there is one thing to use stencils for as a hobbyist, but that is not about soldering. The stencils are great as a template for spray painting things, like inscriptions/symbols on a front panels or for inscriptions on a solder-stop of a pcb and similar stuff. It is just pointless for applying liquid solder paste, since that really is a waste of time.
Great video, Ralph. Just one thing though - the number of primes per 30 sec is not a good indicator of speed, because the amount of time required to check if number N is prime is proportional to N^2. I think it would be better to fix the amount of work the CPU should do (let's say, calculate the first 1000 primes) and measure the time it takes to do it.
I'm sure you're right, Dima, about it being a strange test, but then it is the _same_ test for each processor, so no matter how it determines whether a number is a prime, they are all doing the same work? Funnily enough, the even odder method used in the previous video showed how long it took to generated the first 500 primes. Dang! I should have left it alone.
Sorry Ralph, I was not quite right. The amount of time required to check if a number N is prime is not N^2, it's sqrt(N). My original point still stands - it's not a linear dependency. Try calculating the first 500 primes and the first 1000 primes on the same CPU, the latter will take at least 4 times as long as the former. So, if two CPUs can do 500 and 1000 primes in the same amount of time, you could've thought that one is twice as fast as the other, but it is in fact 4 times as fast.
Oh, and one more thing. Your arduino sketch uses sqrt(num), which does floating point calculations and returns a floating number which then gets converted into an integer. ATMega and STM32F103 don't have hardware support for floating point, but ESP32 does, so it's not a fair comparison. Try removing the sqrt() call and changing line 52 to this:
for (long cnum = 2; cnum*cnum < num; cnum++)
Then your sketch would use only integers. I wonder how much difference it'll make..
OK, I'll try this next time I fire up my Eclipse and let everyone know in a future video, thanks!
Thanks for the tip about the naze32 firmware, I spent a couple of hours trying different ways to upload over USB. It is possible that you may have restored my sanity.
I hope so, although I have no idea why that particular bootloader works and the "official" one doesn't. And yet others have no problems with the USB and the official one. I blame Bill Gates.
@@RalphBacon naze32 didn't work for me. Sanity is back in question again. Serial programming will do for now. I think it depends on what clone you after using.
Oh drat! It could well be down to the actual clone. Well, this subject is not dead yet.
Hey Ralph, fun and interesting comparison. I guess the primes/clock is roughly the same for GD32 and STM32, with the ST one having a bit of an upperhand.
Some suggestions:
1) overclock the STM32, you can make it work at 108Mhz by modifying the PLL clock multiplier which is usually at x9 for 72Mhz but can be set higher.
2) check out those ST-Link V2 programmers, they are really cheap and use a STM32F1xx chip inside and if you open them they include headers so they be reprogramed like any other STM32
Thanks for the great content.
Yes, the GD32 does seem OK and is, apparently, used as the _de facto_ STM32 in Chinese markets. And it will be able to run at 108MHz too, just as you say. I do have a couple of ST-LINK programmers now (after I blew up my previous one, no idea how I did it) but haven't really used them in earnest yet. Thanks for posting, great to hear from you.
I picked up one of those "blue pill's" about six years or so ago. The Ebay listing said it was a super fast arduino! Never have gotten the blasted thing to work. Yesterday I was cleaning in my old backroom shop, long abandoned in favor of my recliner, when I came upon it, in a little static proof envelope that I pushed it in after many disgusting attempts to make it work, so thought I would view some of your videos and see if I could get the damn thing to do something. Still haven't tried but it is laying next to me on my end table right now. As far as micro soldering, I have the perfect tool! My unaided vision is at 20/400 20/400 (my doc says it is much worse then that but that is the top of the chart) With glasses, I do have 20/20 in both eyes. So all I need to do is remove my glasses and add a bright light, and instant microscope so long as it is about 7 inches in front of my nose. For an iron, I picked up some curved needle type tips, I use the backside of the curve for most joints but for tiny SMD's the tip works good so long as I have a good hold with my curved tweezers. I built one of those little SMD wristwatches with the clear case and leather band that are being peddled on Ebay using that, all very tiny soldering some in very close proximity to some melt-able cables that hook the screen to the board. I was very surprised indeed when the blasted thing actually worked when I finished. It had a couple of chips like the one on that board. The way I got them to work was to flood them with solder, then use the iron to flick off till only the links remained. I didn't think it would work but saw someone on a video do that so figured it was worth the try.
I don't really know what 20/400 means but it sounds bad. I've found a source of +6.0 and +5.0 glasses on Amazon so with my contact lenses I can now do SMD soldering with just a pair of bottle bottom glasses. Cool. (No, it doesn't look cool, as my eyes are as large as a watermelon with these on, but at least I can focus real close again).
I'm getting the age related vision problems too. For small work like that, I use the +3.5 magnification reading glasses that my local dollar store sells. They work great.
Yes, I got some of those too. But I'm more short-sighted than that which makes it harder to focus really close up. I'm now using +6.0 reading glasses in one eye, +5.0 in the other (I swapped over the lenses from two pairs). Now I can solder _really_ close up. I can't see a thing further than 6 inches away but for soldering it's even better than my jewellers headset magnifier. All in all, I'm using 3 sets of glasses (over my contact lenses) for computer, middle distance and close work until I get my cataract surgery done. What a pain!
Yet another great video, thank you.
For soldering the chips, definitely look at drag soldering (search UA-cam for David Watts and scroll back a year or two and he has a great video doing it). Alternatively spread some paste across the pads, drop the chip on and heat with your air gun. So long as there is solder mask between the pads then surface tension will pull the solder to just the pads and pull the chip into place. If you get the amount of paste correct there will be no shorts as the solder mask repels the molten solder.
Alternatively take a look at Unexpected Maker's channel and search for the video on his reflow master. More expensive though as that requires a toaster oven.
I'm (slowly) getting the hang of dragging properly, Gavin, (I was using far too little flux paste) but I might get a better soldering iron too (or tip, at least) that has the "well" of solder.
One time I was eating chocolate chip cookies in front of my PC, when I picked up and ate a very crunchy crumb. I'm pretty sure it was one of the SMD parts I was picking out previously, when going through a list of needed components for a PCB, but suddenly vanished out from the tweezers. But don't worry, the parts were conforming to RoHS, so should be fine ;)
I do like the 0603 or even 0402 size, not necessarily when designing a circuit, but for bodges and repairs. They fit in between pins or almost anywhere on a PCB after scratching away the solder mask. And they work even better on one off projects that are quickly done on perfboard, instead of using "huge" THT components.
I'm quite myopic (5 diopters), so I see squat in the distance without glasses, but it's very useful when doing closeup stuff. Over 40 now and definitely recognizing that I can't go as close with my glasses on anymore, but the myopia will stick (luckily?), so no reading glasses for me.
This is probably a true anecdote, and I laughed a lot when I read it! Is it the first step towards Borg assimilation, do you think? I just hope it was a capacitor you swallowed. Resistance, RoHS or not, is futile.
That's CS32 from CKS... GD32 (F103) uses serial flash (like ESP32) to store program. That flash then copies itself at boot to SRAM (yes, it has more SRAM) which allows microcontroller to run at higher speeds with no waitstates (and elaborate accelerators). CS32 - I don't really know how they handle things. However ARM Cortex Core performance isn't actually the most important thing in microcontroller world. It's more interesting if GD32 and CS32 has at least similar ADC, Timers, GPIO, SPI, I2C, DMA performance.
From what I read from Roger Clark's website they are pretty much interchangeable. But as I haven't actually tested all this it's only what he reports.
Yep, and someone have decapped it more than 3 years ago. For those interested in die-shot and interesting analysis visit: zeptobars.com/en/read/GD32F103CBT6-mcm-serial-flash-Giga-Devices
The other parts being so close may make it hard to use a soldering iron but I'd try the "Flood and Suck" approach. Clean the board well, flux the pads, place the part and solder it with the iron, not worrying at all about the bridging, just get more than enough solder on the part. Then use solder removal braid and suck out all the solder from the pins you can. The bridges go away but the surface tension will leave enough solder between the pins and the pads so that you end up with a nice solder job. To be sure, after soldering, use a magnifier to inspect. I use a xacto knife under a magnifier to gently push each pin. If the pin moves, it's not soldered. Rinse and repeat.
I like the idea of pushing the pin to see if it soldered, David. I had a couple of unsoldered pins. But now I have a better solution overall so keep watching, all will be revealed in a future video!
I just got hold of an STM32F401CCU6 WeAct board. I ran the Primes test and it finds 88049 primes in 30s, which is impressive. Programming was straightforward and the USB bootloader worked first time. I think its work looking at both this board and the 100MHz STM32F411CEU6 WeAct board. Both of these boards have Floating point processors so that brings other performance enhancements over the STM32F103 blue pill boards
Ah, yes, a FP processor will undoubtedly assist in speed (when calculations are required). It's quite amazing how quick some of these STM processors are, although the ESP32 seems the fastest and most rounded in capability right now IMHO.
It's obvious it's more performant. STM32F4 is using ARM Cortex-M4F, which adds a FPU compared to STM32F1, which is Cortex-M3 and has no FPU. You might want to have a look at STM32F405, these run at 168MHz and can work as a USB OTG host at USB 2.0 (480Mbit/s) speed
if you dont have a stencil and solder paste, i just slightly tin the contacts on the board with regular soldering iron, then add flux and hot air the chip on works pretty well
Yes, that sounds a good way of doing this. The answer is plenty of flux as that just makes the soldering work so much better.
Great video as usual. Keep up the good work. I don't think I could solder that small. Age gets us all
Age, and a dodgy eye, has already got me, Michael, so I was foolish even attempting this (but now that I've done it, and recorded it, I'm feeling just a little pleased with myself that it eventually worked). But never again with components (specks of dust) that small!
@@RalphBacon with my eyes l don't think I will try lol
Like others have said, I'm a fan of spreading some solder paste over the board and running over it with a soldering iron, effectively tinning the pads without putting solder everywhere. Then start in one corner and work your way around. Also, plenty of practice can be done on old or broken boards- why practice on good components 😀😀😀
I have some practice boards to try this out on, Tim. And, a bit like a surgeon, eventually you have to use good components rather than cadavers!
16:11 - the caps around the LDO are insane and will _cause_ usb problems. The USB specification lists 10uF as max (to lower the inrush current), so a 10uF and a 100nF on each side is just about right.
Indeed. Madness. Think of the inrush current! Perhaps it was a misprint and was supposed to be 33µF? I've never seen such a large cap after an LDO. 10µF with a 10nF in parallel perhaps is standard, as you rightly say.
Ralph take a look at the ADSM301 and the ADSM302 USB microscopes I got the ADSM301 about 6 months ago and I love it, it was around 150$ US, now looks like it's around 180$, for a 20$ or so more, the ADSM302. Before, I was using a magnifying lens and I would get headaches and my eyes would "wig out" :) for a few hours. These have excellent working height unlike the cheap 10-20$ USB microscopes. My next planned upgrade would be to mount it on a articulating stand so I can examine larger boards like PC motherboards and XBOX boards
Oh I did, believe me Sylvain. Long, long hours evaluating them all. But it's too expensive to just go out and buy so I'm starting off with a cheapie that most will probably shudder at but others have said good things about it so it's worth trying it out, just to see how it goes. Buy cheap then buy the tool you really need, hey?
The best way to solder these devices is by "drag soldering" where you wipe plenty of liquid flux over the pins and then drag a ball of solder on the soldering iron tip along all the pins.
The key do success in this method of soldering is the use of liquid flux.
Flux of any kind it would seem, Neil, is the key. I obviously didn't use enough according to all the video links others have posted here!
I tried this again this afternoon, with a decent amount of, admittedly jelly, flux and I had great success. I guess everyone does it differently.
9:45 both stm32 have the same performance. it's just oscillator variations
Not true. Even if we run the GD32 at 72Mhz it is still quicker due to the zero wait state for flash memory. I'll do a further demo in due course on this aspect.
@@RalphBacon
Ralph, I agree with you.
These clock crystals are specified to 100 PPM (or better) or an error of 0.01% (note:PPM Parts Per Million)
The performance difference shows the "Blue Pill" 4.4% faster than the "Maple Mini".
If these clocks were that different, I would expect to see errors on the serial communication due to baud rate issues. (although not guaranteed).
16:12 the guy changed the caps but never thouoght about adding a free wheel diode from out to in to protect the AMS1117 from the surge of current coming from the C6 discharging and going thru the reg everytime power is down, which will surely kill it in no time! The kind of thing I directly notice in a circuit and shun reading the rest.
Is a fly back diode really required across a voltage regulator? The spec sheet makes no mention of this and the internal circuit diagram of the AMS1117 doesn't indicate that it would be helpful. If the flyback is to protect from inductive loads then the place for the diode is across the coil there, surely?
You can solder the LQFP just by putting a thin stripe of solder paste over the area the pins are to solder to. Hot air, wicking action combined with the solder mask will flow the solder to the pins and unless too much solder paste was use should not cause bridging. Use the desolder wick as indicated by David Westernall to clean up any bridging you see under microscopy. The hard part of this method is the chip will tend to float on the solder pads so you need to use some kind of a pick to position and then hold the chip down for a few seconds while the solder cools.
Using a USB microscope to solder under does not work well. It is very difficult to position components and tools without stereo vision.
I wondered how the paste would fare, Carol; but when I use it, it does seem to get everywhere! Practice is the answer I suppose. Perhaps the USB scope will work OK for me as my stereo vision is pretty poor anyway! Yet another cross to bear... Thanks for posting, nice to hear from you.
Drag soldering, with 0.5mm soldering iron tip works fine down to 0.4 mm pitch chips. Micro scpe look at ADM503, good work clearance, many mscopes too close no space to work
That's pretty much what I ended up doing, Helmuth, but I was very unimpressed with the result (as you may have noticed). Others have suggested solder paste and a hot air gun (which I used to desolder, not solder). I might experiment a bit more. Thanks for the suggestion about the microscope.
yup drag solder and use paste flux, or solder paste and hot air/hot plate then touch up with fine tip
I have been meaning to ask you how you were getting along with the YOCTOSUN LED Light Hands Free Headband Illuminated Magnifier that you bought when starting to use SM components
I guess you saw me in the video with them then, Bob? Essential Stormtrooper headgear, I would say.
I could not have done this soldering without them, that's for sure. 20 years ago I would not have needed them.
Another comment here says that they bought a strong pair of reading glasses from one of the Chinese websites (x4 magnification) and they worked OK. I did a test against my x2 readers and my x2 headset; the headset was more powerful so I don't know where the x2 comes from. More like a x4. And they go up to x3.5 (which is a bit like x7).
@@RalphBacon Thanks for the reply. I originally wrote my post when you said that you had used your camera when soldering the SMDs. In fact my post ended by saying something like "I presume, therefore, that the headband was not a success".
Of course, as soon as I saved it you mentioned the headband so I edited it to remove the last sentence. It sounds like it was useful after all. I can feel an illuminated headband in my future !
You tack one or two pins and flood solder the rest. Then use solder wick to take off the excess. Watch Dave EEVblog
Believe me I watched _lots_ of videos before attempting this, Graham, as well as practicing on the Banggood boards. Everyone make it look so easy! Which it probably is, once you have mastered the technique. So I shall practice some more.
I believe the right technique to solder chips like that is to use some of that jelly like flux that comes in a syringe, you tin your iron tip and drag the tip across the pins, the flux does it's magic and the solder just sticks to the pins/pads perfectly. Look at this video from Snazzy Labs where he's building a RPi Zero retro gaming handheld - in particular where he solders the ribbon cable for the screen. I also believe iFixit have done a series on micro soldering too.
I have that flux in a syringe (I used it lots for this video!) so I will search for SnazzyLabs (no link! - I will find it). [Edit] Found it! ua-cam.com/video/YrofIDloHgw/v-deo.html
I got a blue pill about a week ago. I was trying to hook a literal USB to serial link into the uart for programming and of course that would not work. Then I tried to program it using a ttl serial converter today. It does not work for whatever reason, including me blowing the uarts. I soldered a 2K resistor I found on some scrap on. I finally decided I had to use tweezers while soldering to force it into place because you just can't to it with just your soldering iron. Very small stuff.
Blue Pills are great if you have the right bootloader (as I found out).
@@RalphBacon oh later in the story it was a learning experience,. The serial program does work if you (de)press reset at just the time the cube programmer pauses to connect. Also I used the stduino and pc13 generic bootloader and it windows it is pretty much the same experience as an arduino at this point. Linux needs some examination. Does not go into DFU mode as it stands.
You're saying board X is so many times faster than board Y. However, the test you are using doesn't scale linearly for multiple reasons. For one checking if a number is prime is faster for smaller numbers. Furthermore you seem to be printing the number of primes found and not the total numbers that have been checked. The distance between prime numbers is bigger for bigger numbers.
So would it be better to say "Find the first 1000 Prime Numbers" and then print the time it took (I suspect the Nano/UNO might take a while)?
@@RalphBacon Yes, using a fixed number of calculations is easiest. You could pick a low number such that the time is acceptable. If you find that the results are too noisy then just repeat it and take the average. Although I would expect on microcontrollers like these the time is quite predictable. Just be aware that the Arduino framework might do things in the background.
@@RalphBacon Yes, exactly. Pick a number that the UNO can do in a reasonable time (50 to 60 sec ) by trial and error, and then stick with that number for all the tests. The times for the other boards are then inversely proportional to their computation speed.
@@RexxSchneider well and you're depending on oscillator consistency if it is timing itself... Swapping the chip actually would have helped a bit (if you weren't using any of the internal oscillators)
18:00 Lots of Flux (like flooded with paste flux) then it is super easy since the flux will help prevent the pins from bridging. (Louis Rossman has a channel on Apple board level repairs, it is awesome). Here's an older vid where he does a QFN chip smaller than what ours ua-cam.com/video/mr1UVPsExiE/v-deo.html
Yes, he uses so much, James, but it works! But I always clean mine off, I don't trust the "no clean" label one bit. I like his microscope too but beyond my reach, that's for sure! Thanks for posting.
@@RalphBacon The 'official' term is a Rossman unit of flux /s I agree it is excessive as is his iron temps but it is hard to argue with his results! :) I have also fallen in love with using the Amtech flux he uses, it has transformed my ability to perform SMD repairs and builds. I used to think drag soldering tiny adjacent traces was magic, but nope its just the flux :)
For old eyes with up-close work, I bought a few "old man" reading glasses. They have several levels of magnification. I like the 2.0 versions.
You can get them up to +4.0 on places like Banggood and AliExpress. In the shops here they only seem to go up to +3.0. But then you can get really close for SMD soldering. Cheers, David, thanks for posting.
Hi, I don't go below 0805 either I think that's it at 53 without any expensive gear! as for the chips I thought the same way and now I'm playing with a dirt cheap (£6.70) hot plate, all I have to do is get a kindly Arduino to automate it and bingo, Roberts ya relative!, seriously look up "hot plate reflow" see what ya think, I mean there are folks getting good results using an upturned Iron!
Hotplates are quite popular. I think the biggest issue is heat uniformity. It's also more difficult to follow a proper temperature reflow profile as you're applying heat to the back of an insulator (pcb). Most people seem to monitor the surface temp of the hotplate but the part that matters is the chip pins and footprint. I have a "PCB fabrication" playlist saved with all the different hotplate, toaster ovens, and reflow rigs I've come across that other people have uploaded. The coolest cheap and unusual rig that comes to mind is one by Andy Brown. He used a halogen flood light with an enclosed container and μC to make a reflow oven. That one is definitely worth checking out :-)
-Jake
@@UpcycleElectronics Oh my!!!...My name is Andy Brown and I have just bought a Halogen hotplate from Ebay and I plan to use an Arduino to control it. I also want to monitor the temperature by embedding a thermocouple in a PCB in order to give me a reading closer to the actual board temp...this is so spooky please give me the site address to check out..as I haven't done this yet!!!!
@@andymouse
Lol
The impostor/Other Andy's upload:
ua-cam.com/video/6bxpy1Ret6Y/v-deo.html
My playlist with a few references I've come across:
ua-cam.com/play/PLPIwHuVy9EyNsNgV-mOPapYaKjEw2C_kP.html
Probably the most notable upload I've seen about hotplates and reflow:
ua-cam.com/video/075D_tuswVU/v-deo.html
-Jake
@andymouse123 you'll be wanting to watch my video on thermocouples then, right? (Shameless plug ends).
And, regarding your reply to Upcycle Electronics, are you travelling backwards in time? Spooky indeed.
I hear what you are saying about small components and being older. I would recommend a large illuminated magnifying glass. The board is likely to have lead free solder on it also using flux paste will help a lot.
I now have headband magnifying glasses (for jewellers) and a USB microscope too. Works well, all will be revealed in a future video.
Hi Ralph,
I think I'm about your age and I have the same kind of eyesight issues. I have a couple of suggestions.
Drag soldering works surprisingly well. You need to use plenty of a quality flux. I use water soluble organic flux. It works well and cleans up beautifully with soap and water.
I'm not sure when you made this video, but now you can have JLC PCB assemble your boards as well as fabricate them.
I've also had success with solder paste and a toaster oven.
Thanks for the video!
Cheers!
Thanks for the tips! Nice to talk to another 40-year old (stop sniggering at the back there).
@@RalphBacon
I kinda remember 40 :) Or should I say 40 with 17 years experience. ;)
Ha ha! I like that. But I am 40 with only a very few years experience. Nudge, nudge. Stop laughing.
What about the energyconsumption of the asian alternatives ?
No idea off the top of my head, sorry. I guess the datasheet makes good reading for this sort of info.
Ralph ... I can TOTALLY relate to your eyesight/soldering etc. thing and everything else you're talking about. I'm 65 and can't see my hands in front of my face without one or another pairs of reading glasses. HOWEVER ... there is help. First ... Forget about USB microscopes for soldering. What you want (trust me) is an "Optical" stereoscopic microscope. I heard the part about 'you can't afford much', but honestly, these are a game changer, and they are a LOT less expensive than you might think. (USB microscopes are great for other things, but not for soldering.)
I bought an Eakins 'trinocular' (the 'tri' means there is a 3rd tube for mounting a camera), but the 'binocular' is more than adequate if you just want help seeing small stuff while soldering. Get the 'lowest' magnification range, typically 3.5x-45x. (trust me, you don't need any more than this.) You can find more on their AliExpress site: www.aliexpress.com/store/836673?spm=a2g0o.detail.1000002.2.5aeb643f8U2BIV
They have a billion different options on their site, and it can be confusing, but really everything comes down to a few options: 1) trinocular vs binocular, 2) high vs low magnification, 3) type of stand, and 4) which camera if you pick the trinocular option. Eakins is not unique, there are several other suppliers selling very similar equipment, for similar prices. Further, every month or two they have sales. ... Wait for the price drop.
Why the pitch? Recently, I bought about 15,000 0402 resistors and capacitors from LCSC for a total of only a few dollars. Look for yourself; you can purchase 1000 resistors for as little as $0.38CDN/1000. I've never heard of anything SO CHEAP. 38 cents for 1000 of anything. I haven't done much with these yet but importantly, with my microscope, I don't have ANY PROBLEM at all soldering them. Did I mention ... my eyesight SUCKS, and is probably even worse than yours. And, soldering 0603 parts is literally ... EASY and routine.
You didn't mention what kind of soldering iron/solder you are using. I have learned over the years that these make a world of difference. Some time ago, I bought a 'BAKON 950D' soldering iron. This is one of the 'new' type irons, which is VERY responsive to temperature changes. Unlike others however, this sells for about $50, and honestly it is ABSOLUTELY AMAZING! Again, there are other similar options. Solder is a whole other issue, but I use 37/63 lead based solder, 0.3-0.5mm. It is FAR easier to work with than lead-free solder. But if you think lead is a dirty word, then spend a LOT of money and get GOOD lead-free, otherwise you'll find soldering to be tedious and difficult.
I have no commercial interest in supporting the above ... I've simply figured these things out from experience. I hope this helps.
I feel your passion and enthusiasm in your answer, thank you for all the information.
Now, my eyes are slowly improving, to the extent that I can solder 0603 with just magnifying (jeweller's) glasses +3.0 / +4.0 although because my eyes are not the same (they got the replacement lenses for my eyes wrong when they did my cataracts) I still have some difficulty with stereoscopic vision.
I go back to the eye clinic on 10th August, when the consultant will decide whether we can continue eye drops or whether some further laser surgery needs to happen to remove scar tissue from my retinas.
Regarding the USB microscope, I find them very good even for soldering (no 3D vision, admittedly). I did investigate the type you mention (binocular) but eventually decided the cost vs the benefits could not be justified - but then again it sounds like your eyes are (even) worse than mine!
I have a great soldering iron now, the Atten ST-100 which uses Weller tips and heats up in seconds and has great temperature control too. As you say, a decent iron makes a huge difference in soldering success and I never use lead-free, it is the devil's spawn. I use 0.4 - 0.6mm leaded solder and it works well especially with copious amounts of extra flux for SMD components.
Your comment could have been written by me, we are on the same page. At the moment (and until the eye surgeon signs me off) I'll stay with my current setup then consider that binocular scope again, thanks for the information.
@@RalphBacon Thank you for your quick reply. I feel a bit awkward now, since I didn't mean to pry into your medical details. Good to hear things are getting better though.
For the price, I'm thrilled with my little BAKON soldering iron, but clearly there's no point in dumping your perfectly good Atten iron.
There is actually a 'significant' difference between 60/40 and 63/37 solder ... not sure if you and/or others are aware of it. Briefly, 63/37 forms what is called a 'eutectic', which in a nutshell means that at this ratio, the alloy behaves more like a pure element with a very precise melting temperature (183°C in this case), making it ideal for hand-soldering electronics. 60/40 tends to stay semi-liquid (sticky) in an around the higher melting temperature (190°C), making it ideal for plumbing. This can make a lot of difference especially for tiny parts. (Lead-free solders by comparison, (there are many), often suffer from stickiness and higher sloppy melting temperatures, making them difficult to hand solder.)
As for the microscope, it was a big, much considered decision for me. I also considered the less expensive USB option, but I'm very happy that I bought the optical. It made a huge difference. USB microscopes can still be useful for inspection and making videos etc.
Save up your coins and buy a true stereo microscope you will never look back. Get it with a .5x Barlow lens for better lens to board distance & easy access for soldering. For the small parts use .020" or .015" diameter solder (use lead based 63/37, none of this lead free crap) and a soldering iron with a fine conical .016" diameter tip. I use a Metcal iron but others irons with small tips should work fine.
I looked very hard at true stereo microscopes (I would actually want a trinocular one so I could film the stuff, or output the image onto a big screen for recording). But at around £600 ($750) I am going to have to forgo this until I have at least 100,000 subscribers! But you never know, Amazon Prime Day is coming, perhaps they will knock 50% off!
The 330uF might be causing too high an inrush current - perhaps you had a lower ESR cap than R3bers?
I agree. Had similar problem with ESP32 Cam. I added a capacitor to the supply input and it shut down my supply. Had to re-set and reconnect the supply 3 times before it stayed on. All down to inrush of current.
@@sausagehider I had the reverse problem on an esp32 board. Had to fit a 330u cap otherwise it would brown out when doing access point on wifi!
@@EdFrench_uk Yes, that is why I had to fit the cap too. It was when connecting the supply, it was 3rd time before the cap had charged enought so as the surge didn't trip the supply.
@@sausagehider a series resistor might help but harder to retrofit onto existing board
@@EdFrench_uk Wouldn't that drop the voltage. The ESP-32 cam suffers from a lot of noise lines on the image if the voltage is low. A few 100mV and the picture is unusable. I had to use 6 volts from a boost converter to power it enought to get a cleaner picture. Could be down to the onboard LDO not handleing the higher current draw of ESP-32 and camera!
I am impressed, soldering these small parts is not easy, I prefer 0805 parts. I have done some chips with about 0.6mm pitch, they are not easy (I cannot remove shorts with the soldering iron, but has to use solder wick). I often use a USB microscope to check stuff, but it is not easy to solder under.
It looks like you are getting away with comparing the performance with the long data type, both mega and ESP32 use 32 bit values and I supposed the STM32 will also do that.
Yes, the power of a 32-bit processor. Poor old Arduino Uno/Nano never really had a chance to compete.Don't be too impressed with my 0603 soldering, Henrik, I thought it was awful. I will stick to 0805 in future. Perhaps I can stretch that r3bers board up to that size _and_ add in an FTDI or other USB controller too. Then I can really call it a Green Pill and retire on the proceeds. Meanwhile, back on Planet Earth...
A STM32 or similar that is cheap (I.e. cheaper than ESP32), small and works perfectly with the Arduino environment would be nice. The best/easiest way to get money is probably to start a succesfull company and sell it (This can be a cooperation between multiple people).
Since GD is running at a higher frequency it is not a very fair comparison.
To make it more reasonable, you should try overclocking those chips and check what is their maximum frequency.
Chinese manufacturer may have just claimed higher specs but in reality, it may be pretty much the same thing.
Stm32f103 is known to work fine at 128MHz and some even claim 160 and over 200mhz
As I know the limiting factor here is flash speed, if you copy your program into ram you can increase speed even more and make it closer to esp32
You're the second person to use the word "fair", as though these chips need defending! They were just out-of-the-box tests to see what they offered.
Yes, the GD32 _might_ be internally overclocked to appear to give an advantage. But we also know that there are zero wait states on the flash memory for the GD32 whereas the STM32 requires 2-3 wait states - thus the GD32 gets a speed boost.
That's not smoke and mirrors, that's better chip design. Without copying anything anywhere!
But I'm sure the STM32 will continue its unabated reign of Blue Pill excellence, regardless!
@@RalphBacon
Well, I am not some Of STM company fan. In fact, I am even using GD chips myself, because of the better price, ST cost $ 1.25 and GH costs $0.81 on Aliexpress with free shipping. That's a pretty decent difference.
But your tests are extremely primitive and useless if you do not do a proper comparison. I am not defending STM design but proper testing would be way more useful than comparing default clock speed not just as a marketing ploy but as really valuable information.
If you claim that GD made a better design you could compare their performance at the same clock speed and then later you can try overclocking them all to the max to test their capabilities. Measuring power consumption also would be useful.
OK, I'll see how easy it is to set the clock speed on either of these chips and see what I come up with in a future video.
That said, the fact I desoldered one STM32 chip and soldered in another (the GD32) and it ran quite a bit faster is still a valid test, IMHO. You wouldn't ask BMW to disable the turbo when comparing it with a Ford, just because the Ford didn't have one, so the road test would be 'unfair', would you? You get what you get with the manufacturer's design.
We'll have to agree to disagree (not that I _am_ disagreeing exactly!) and I'll do that clock speed setting research asap. I suspect the GD32 will still be faster due to the zero wait states on flash memory, but we'll see...
@@RalphBacon
Thank you.
In fact, I wanted to do such test myself so you will save me some time. Setting frequency is quite easy, you can even change it anytime so to avoid problems with peripherals you can even change it when doing the test and then set it back to print results. or if you set it to some obscure value you need to reconfigure UART and USB speeds again
Well if Ford does not have turbo I do not ask to disable it on BMW but if both have a turbo and one comes with turbo witch on and another with turbo switch off from the factory I would ask to put then moth in the same position ;)
You don't have anything to disagree because I also want to find out if GD is actually faster at the same clock speed than STM
Also I just found out that GD has another advantage that it can use fractional frequency PLL multipliers so it gives you more frequency choices what is pretty important because you can adjust core speed without affecting peripherals. that multiplier usually happens to be the main problem when setting the desired frequency.
according to my eyesight you have shorted B0 to B1. (around 20.27)
I think you are right. Well, my eyes were pretty bad back then so I'm not surprised. But no damage done, all working OK.
BTW. PB0 and PB1 is shortcircuit ( solder bridge)
Thank you so much for pointing this out. I had to look with my video camera zoomed in to spot this. As a result of fixing it I also attached a few more components including the USB socket, and my SMD soldering skills are definitely improving. Great to hear from you.
Pre-solder the pads, run some flux over them, set the chip in place and apply heat. Looks like factory when you're finished.
Yes, sounds like a good way of doing it.
@@RalphBacon Comes out perfect every time! I use the heat gun for most small items like that, otherwise, I use the drag method, which works great if you use a good flux.
0603 are great :) 0402 is much more fun :)
Oh definitely, that will be my next project. I'm building a full blown Arduino Uno on a board just 1cm squared, using 0402 components. Gosh, I can hardly wait.
Could this be due to differences in the fuses on these chips ? Did you compare those?
I didn't dig that deeply (yet). What was interesting is that the GD32 Green Pill used an 8MHz crystal (not 12MHz) so I'm guessing that the multiplier is set to 13.5 but maybe not. I know the zero wait states on the flash in the GD32 give a big improvement over the STM32 (1 or 2 or 3 wait states). It will all come out in the wash!
@@RalphBacon I have a red one, have not even looked at it, might have to check it out now, great video, very thorough
SMD soldering is not for my poor old eyes. Far easier to just order a blue pill. :D
I don't blame you one bit Nick. That's why these modules are available. If we all had to solder them (eg as a kit) I wonder how many would work?
even a cheap binocular microscope is far better than the best usb microscope.
a good binocular microscope and an SMD soldering station (hot air) is the only way I've been able to get good results trying to solder fleas (0603s) to PCBs.
It's amazing that you were able to solder those 0603s with a soldering iron. Great skill.
Pure luck, Dirk. I was depending on several decades of THT soldering to see me through. I shall have to see what binocular microscopes are out there that allow room for a soldering iron.
Hey Ralph, Do you have fixed set of schematics for the GD board. I would like to put it together myself. Keep up the good work and pet the kitty for me
I do have everything for the Green Pill board, Skip, including all the schematics and PCB from r3bers. I'll put the schematics etc up on my GitHub. Stand by... It's there now.
@@RalphBacon I might change all the components to 1206 squeeze it in and pay for a stencil and use paste.
@@RalphBacon Thanks Ralph
I know it's an old video, but if you haven't bought a microscope yet, I highly recommend. I recently bought one off Aliexpress with a 7" LCD and even though it's a cheap piece of crap which overstated it's magnification level by like 1195x and will probably burst into flames at any moment, it's a game changer compared to the magnifying glass on my "helping hands" and my eyesight isn't even that bad (yet).
Yes, I did this some years ago now. In fact, I got a slightly larger one from Andonstar and I use it most days, even if to just read the names of certain chips! Deffo worth having. Oh, also great for checking soldering, with my eyes I need to double check.
If you don't know how to solder SMD chips it doesn't hurt to watch a few tutorials.
Btw. it's actually easy to solder with an iron with a chisel or bent tip just by holding the chip down with tweezers in the correct position, applying a Rossmann of flux(aka a lot), picking up some solder on the tip, and just dragging it over the pins, relying on the flux to keep the pins separate.
All good suggestions! Mind you, given that this video is now 2 years old (at time of this response) I am much improved on my SMD soldering!
An ARM core does not have a division instruction; so pretty odd to use these kind of prime tests for comparison. It is not fair and misleading.
Are we to tailor the sketch then to what each chip does best? Not a level playing field, which is why I changed to this sketch.
I suppose I should really be using a Dhrystone/Whetstone sketch to calculate a proper VAX MIPS rating. As it happens I know the STM32C103 has a VAX MIPS of 48.81. In comparison, an Arduino Mega 2560 has a VAX Mips rating of 7.23. Other than that I don't know what it tests but I bet division must be included in there somewhere, surely?
Anyway, benchmarks are really just a bit of fun and entertainment and give you a a good idea of what performance to expect in Real Life (a bit like car 0-60mph tests). That's all.
The STM32 does have a HW multiplier and division - www.st.com/en/microcontrollers-microprocessors/stm32f103cb.html
The Cortex M3 does have a div instruction. The M0 doesn't have DIV (much like the early ARM designs).
Hi Ralph,
A microscope is undoubtedly the best way to go, especially if you have tired eyes like I do but, they are tremendously helpful even for young eyes.
Once you try one you will see for yourself that everyone that works with PCBs should have one.
All of them have an independent adjustable focus on one of the eyepieces and in my opinion is a must!
Just a note on magnification: You will likely never need magnification over 20X for PCBs! In fact, 7X to 10X seems to be the most common magnification range used for our application.
45X and 90X is just too much for what we do (in my experience).
You mentioned in a response to a comment below that you would want to buy a Trinocular scope so you can use a video camera at the same time.
"Be aware that Trinocular scopes only allow one of the eyepieces to be used while using the video camera"
The only scope that allows you to use a video camera "While" using the scope with both eyepieces at the same time are "Simul Focal" stereo scopes!
Simul Focal scopes aren't really more expensive, there are lots of deals on the net for Simul Focal scopes.
Check out this video for an demo of using a Simul Focal scope as well how to successfully drag solder while using the scope and.....you'll see just how much flux he is using for the process.
ua-cam.com/video/C_eQrbop-J4/v-deo.html
You have received lots of great tips for soldering SMD components.
It's obvious that you aren't willing to spend the money for the "Tools You Need" to do the job (which is kind of ironic since PCBWay is investing in you and your channel).
You're going to have to bite the bullet at some point and "Invest" in the right tools for the job.
I have found that one of these circuit board holders is also a must, this one allows for rotating the board so you can see the back of the board without having to remove it from the holder and it's pretty cheap (worth every penny IMHO).
It helps a lot to be able to steady one hand with the other hand but to do that, the circuit board has to be held in place by something.
This holder works very well under a microscope as well.
www.amazon.com/Aven-17010-Adjustable-Circuit-Holder/dp/B00Q2TTQEE/ref=sr_1_3?keywords=PCB+holder&qid=1563118302&s=gateway&sr=8-3
Anyway, Best of luck to ya Ralph!
Wow! Thanks, Joe, for all that information. In no particular order, that holder you mention at the bottom of your post, I already have one, for a long time. I bought it to, ironically, be able to solder THT components and then flip the board over to insert the next one. I never thought to try it for an SMD board. My "Strange Third Hand" tool is a blessing though. But I'll give it a try.
Regarding my toolset, or lack of it, I'm in the process of buying a pretty simple $40 USB microscope that many have said they're OK, and many have said, Don't do it, you need binocular vision. But, I'm testing the waters to see how far I get. The reasoning is that I tried using my video camera as a USB scope (very enlightening to zoom in) and it "sort of" worked. The monitor is as the wrong angle and position but the principle made me think I have to try it first, before spending anything like what they want for a true stereo binocular scope - a _simul focal_ one at that, thanks for putting me right. That said, the microscope's camera is eye-wateringly expensive.
Anyway, at 1/20th of the cost of a 'proper' tool, this USB 'toy' will be an interesting testing of the waters - and I might be able to sell it on (or give it away) if I decide to go big time. My sponsors don't pay _that_ much so it will be my own money for this!
As a result of another comment telling me I had a short between two pins on my swapped out STM32 board (I had, who knew?) I've just been practicing all the techniques and guess what? Yes, I'm definitely getting better. I've cleaned up the chip, soldered a few more components, even the USB socket, all without _too_ much trouble. Exciting times!
Thanks very much for all the info, it's great that so many (including you) have taken the time and trouble in giving me their slant on things. Most appreciated.
I have magnifying glass with ringlight for desktop use. Pretty ok for SMD if you don't want to buy microscope.
@@RalphBacon
Do you plan to begin showing viewers your soldering process?
Most of what I've seen from you is "The results" which is great IMHO, there are plenty of other videos showing the process.
If you are not planning to show viewers your own soldering process, you could buy a simple 10x to 20x stereo scope just for soldering and inspections off camera, not to mention the fact that a decent scope is great for much more than soldering.
I'm not kidding, a good microscope is THE SINGLE MOST IMPORTANT tool in the shed, especially for tired eyes like we have!!
Here's a good scope on Amazon that would be perfect for just soldering and inspections at $202.95:
www.amazon.com/AmScope-SE400-Z-Professional-Microscope-Magnification/dp/B005C75IVM/ref=sr_1_7?keywords=stereo+microscope&qid=1563135891&s=gateway&sr=8-7
The best arrangement I've tried so far is a 1x objective with 10x eyepieces, the scope I just posted the link to has that very arrangement (along with a set of 20x eyepices for really close up inspections).
That arrangement has the best amount of work space under the objective lens I've seen (check out the specs on that scope in the description)
You can also buy an adapter for most any camera so it's not etched in stone that you have to use a camera made specifically for scopes (If you intend for sure to film your soldering process).
Not enough for me, LimbaZero, I need a lot of light and closeup magnification. We shall see what I end up with!
Hi again Ralph,
I feel I should be more clear about my suggestion to buy a stereo microscope without the camera port.
It was merely an observation that you usually don't film your soldering process.
I was only trying to help you figure this out by voicing an observation.
I too am one of those guys that only spends money on tools I feel I really need.
For what it's worth, my decision to buy a microscope came from a friend who offered to let me borrow his microscope for a few days, I literally could not believe how helpful it was to be able to find the issue and be able to solder while viewing the board through the microscope. It was one of those "WOW" moments! :-)
The issue I was having at the time was a really tiny strand of coil wire that had broken. I couldn't see that wire strand even with my WiFi microscope, but with the borrowed microscope I was able to re-solder the broken wire while looking through the microscope.
Anyway, if you do plan to start showing your soldering process, then by all means, hold out for the Simul Focal stereo microscope.
Again, just trying to help :-)
Joe
Tell me, what kind of plate do u have ontop of the table?
I mean the green plate with those squares.
I didnt found any similar on ebay. nor on internet at all, may be because of wrong search-text.
It's a *self-healing cutting mat* and I have several. A3 size for normal use but my soldering station has a huge A1 size (900mm x 600mm) to protect the bench. You will see it in a future video but looks pretty much the same as the smaller one. Example of A1 size: www.amazon.co.uk/gp/product/B078XCKDHT
Umm in my ECE lab we used microscopes and regular old Weller irons, you start off by lin in ng up the chip, solder two corners, and then go arround getting the rest of the pins, to fix any bridges you just clean your iron tip and wick it off or over into the other pins
That's pretty much how I've ended up doing it now, although microscopes make it harder (no stereoscopic vision) unless you have (expensive) ones!
@@RalphBacon we had the expensive ones 😅
16:17 The 330uF is too large! It is true, that regulators cannot follow varying current demands at the speed required by some modern controlers and therefore they brown out. But 330uF needs to be charged first and that might overload the regulator already at startup. It would be better to have a much smaller one (10uF...22uF) as well as a number of 100nF (ceramic) in parallel, to lower the effective series resistance of the current reservoire on the output of the 1117-3.3. So what died might be the regulator, since the value of 330uF is quite demanding when initially asking to be charged to 3.3V ... that can draw quite some current through the tiny regulator thingy. Perhaps try to exchange that regulator for some replacement of similar type ... any "xx1117-3.3" might do.
I removed the 330µF and used a smaller one, I _think_ it was around the 100µF mark but I'm guessing now. Whatever, it worked just fine and continues to work so I guess that suggested large component was never going to work!
@@RalphBacon Well, it might at least turned off the regulator by its thermal protection when charging the large electrolyte capacitor on its output. Bigger does not always do good...
But some ESD-protection o the 5v power input is missing isn't it? Perhaps something like CG0603MLC-05E(bournes) or a B72500D50A60(epcos) or at least some zener like 1SMB5918BT3G (on-semi)?
To place the chip on, solder with out a lot of care and then use solder rope on the top surface to remove excess solder.
Yes, I don't worry about bridges these days. My next video shows this too!
@@RalphBacon I've worked in electronics for 45 years and didn't know this until I saw it in a UA-cam video the other day. Of course I was of the days before surface mounted chips took over completely. I still like a hole. I like your show, keep sharing, its fun. My HA sytem was designed in the mid 80s and is still running 24/7. It would have been nice to have had a few of these Arduino boards. Mine was designed around a WIZ ITC232 controller which is still available to this day believe it or not. This may be TMI but the cost of energy for that 2700 sq ft home was $39 a month for the first 15 years due to the HA system integrating passive solar and a hydronic heating system. All Good Things, Cliff
We're never too old to learn something new everyday, pretty much, Clifford! I like the idea of $39 a month for energy, I pay the equivalent of $187!!!
Is GD32 programmable using ST-Link? I use Keil. If ST-Link can write a program there I can work in Keil.
I use a standard USB-to-Serial device to program the ST/GD 32 devices.
Couldnt the esp32 be twice as fast. It is a dual core?
Each core runs independently. So, in parallel. Thus, we could have _two_ tasks running at the same time to do twice the amount of work but it would not complete each task any quicker.
@@RalphBacon oh I see how the benchmark was ment to deliver numbers.
I thought is was about total possible throughout. Thanks a lot!
Cheers Ralph, all great info.
Glad you liked it, zydawn, Thanks for posting.
Thanks for all the great videos Ralph! That prime number code is an offense to mathematics in particular and the universe in general though. :)
Well, its sole purpose is just to get the processor to do some work so we can measure it. The previous sketch used the sieve method but ran so fast after I removed the float I couldn't get a measurement! Zero milliseconds. Even the UNO did it in about 3ms. Glad you like the videos, thanks for posting.
Ralph S Bacon I really do enjoy your videos and appreciate the time you take to make them. They have inspired several projects here and helped out in several others. I hope I didn’t offend you and if I did I apologize.
One question: where did you get the gerber files for the nano?
Hi Ralph, the green pill looks like a fun project... just wondering if you thought about hard wiring extra functionality to the board, like maybe a temp sensor, canbus chip or opto-isolated inputs?
I didn't, Tim, but I wish now that I had included an FTDI or similar usb-to-serial converter.
@@RalphBacon there's always next time... to be fair, I wouldn't know where to start having someone else's PCB design!
SOLVED IT
Cost me a few hours but I have a boot-loader working on stm32f103c8t6 blue pill. I can download programs but even better I get serial debugging from USB also. The 10k to 1.5k was all a wild goose chase you can get it working without any hardware changes at all. For those interested here is what to do.
- Use ST-LINK to put boot-loader on blue pill. STLINK flasher and driver are on the ST site for free.
-You have to match the correct boot-loader via LED pin i.e.builtin led on my board is PC13 so I used Generic_boot20_PC13.
-Don't make my mistake move the jumper back to Boot 0 before reset or the boot-loader get erased.
-The official ST core will not work with the boot loader it is not supported. use dan.drown.org/stm32duino/package_STM32duino_index.json.
All working for me now shows as "Maple Serial (COM15)"
I did this countless times on different boards during my R&D for the video. How come it works for you and not me? Life is so unfair. Strop mode starts...
@@RalphBacon Hmm, in that case it must also involve one of the many things I tried along the way that i didn't realise, was important:
I did at one stage follow the manual instructions from RogerClarkMelbourne github.com/rogerclarkmelbourne/Arduino_STM32/wiki/Installation which included:
-Installing Arduino Sam boards but I assumed it was just for Arm tool chain stuff.
-I also ran his maple drivers Install_drivers.bat that you don't do with the json version unless it is automated. But I am pretty sure it got replaced by Zaidg driver install
The only other factor I can think of is hardware, I bought 4 6 months ago so a different batch to new onces and all are stm32 not gd or CW etc.
You know the device is working properly when you power them up and they power up as DFU device then a second later disconnects and become a Serial Port.
Also when you did the zadig driver install when one did you install? I will check which one I had in the end, its working on two win10 machines and all 4 clones now.
I think it would have worked earlier for me but there are two pc13 bootloader files and I think I grabed the GD not generic one.
Good luck.
Have you used a mask and hot air gun or oven for SMD soldering? I use my hot air gun for lots of things, almost daily.
I use a desoldering hot air gun, but for desoldering it's not always necessary. ChipQuick is good for larger components (such as ICs).
This clears a lot of confusion!!
I'm trying to replace STM32F103CBT6 from one of our products due to chip shortages and price hikes. Since I'm getting the PCBs from JLC, I am looking for IC supplies from LCSC. Now, LCSC does have GD32 F 103CBT6 (Cortex M3) in stock right now, however, they have GD32 E 103CBT6 (Cortex M4) at a very good rate and in large quantity.
Although I've ordered 10 pcs of E103 from LCSC so that I can discover myself, but I'm just dying to know these:
1. My board currently has an FT232 on board. Can GD32 E 103 be programmed through the HW bootloader on Serial1 ?
2. What config changes will I need to make in arduino in order to build?
3. What config changes will I need to make in arduino in order to Upload?
4. And the most important one. I use FT232 earlier and it gives me data rates up to 2Mbps. Which is very much required by my application. Can the USB be used to program in the bootloader and also simulate Serial-over-USB while in the application? Will it give me at least this much bandwidth? (Some forums claim that with some buffer upgradations, it should go as high as 8Mbps, is it true on the standard bluepill)?
I'm afraid I just don't have that information to hand; I'm hoping you will either research it or find it out during the course of your experiments. Good luck with your project.
Make sure that you wear breathing protection when soldering small SMD. 😎
I accidentally ordered metric 0402 resistors instead of imperial. Only 100, so it wasn't an expensive lesson.
And you didn't solder them on, Bernd? All you needed was a steady hand an an electron microscope :)
Regarding ventilation, I have a small USB desk fan blowing the fumes away from me but it's not great.
Hello Ralph,
I hope you are doing well. STM chips are manufactured in many different countries like Korea, Malaysia, Philippines, Taiwan, and China. And I am thinking that GD, CKS, and CS are a few of these Chinese manufactures. So they don't have the need to clone them.
I don't think those chips are exactly the same as a genuine STM32, though, Georg. There are some differences (as noted in the comments on this video). The GD32 seemed most like it, and for general use the others will probably work too. What I would do is design a circuit with a genuine STM32 then swap it out for a "clone" and see if it still worked!
@@RalphBacon I've just took a look in ST's application note AN2606 that the STM32f103 supports only SUART boot. The STM32f105/07 support CAN boot, DFU boot, and so on. I think I will compare the data sheets. That may will take a while.
Best regards.
21:47 you don't mention what solder you use for smd components. I assume you use solder paste.
Nope, I don't, Pete. Whilst solder paste works well for (a) fabrication houses and (b) if you have a stencil so you only put the paste where it should go rather than all over the place, I find that standard leaded solder works best (for me). Use copious amounts of flux on the SMD component area (which also helps stick it in position), optionally hold the component with sharp pointed tweezers on the top, and then just touch the joint with some solder on your iron. Soldered in less than a second. You will see a close up of how I do this in two videos' time using my wonderful HDMI microscope, so don't miss it!
The CKS103f is not fully compatible with the STM32F103C8 Overclocking at 128 Mhz didn't work. The library for 7 segment display TM1637Display.h also didn't work. LCD library LiquidCrystal.h didn't work properly and displayed many incorrect characters. LiquidCrystal_I2C.h did work. I tested bij swapping a number of cks blue pills with stm blue pills in a soldered circuit.
Hmm. I wonder what the underlying issue is. Registers? Hard-coded pin definitions? Something else..? I guess we shall find out as we progress. Thanks for posting your findings.
Great video Ralph- Q: does the STM Maple Mini require toggling the small yellow bridge pins after programming like the Blue Pill does? I think moving that bridge each time would drive me crazy!
No, the Maple Mini has extra circuitry so that the USB is used to upload the sketch in exactly the same way as an Arduino Uno or Nano. No jumper changes required!
Very informative video
You are most welcome MUHAMMAD YAWAR IFRAHEEM, I'm glad you like the video. Nice to hear from you.
Actually, this video is more useful today because of the chip shortage. Right now on Canadian Amazon the originals can go for over $100. Next day delivery is $22. If you can wait a few days, you can get 3-5 for around $30. The clones are no-brainers right now and can easily be returned to Amazon if needed.
The clones are not too bad, although I would not use one without prototyping the entire project with one to ensure compatibility. If that gets the thumbs up then it's good to go!
I've bought 3 stm32 boards. On 2 of them, when I selects boot0 to 1, the blink sketch continues working and can't program via serial. I needed to buy an stlink.
Or just a cheaper FTDI USB to Serial adapter which allows ICSP programming and Serial Output too. See my video #155 on a custom PCB I built for mine.
@@RalphBacon I've got an ftdi. But 2 of my 3 bluepills can't be programmed via serial. Boot0 jumper does nothing. Check this video: ua-cam.com/video/z-AR7Xl8pA0/v-deo.html
I have the exact same problem with a board I ordered from hiletgo on amazon. Interestingly the ic on the board is a knockoff cks32, not an stm32...
@@MarkEcker One of the boards with the problem is cks32. The other is stm32.
my standard st-link cannot connect to cs32f103
tell me what am I doing wrong?
There are only four wires to connect but you must connect them correctly, of course. VCC and GND are self-explanatory, the SWCLK and SWDIO go to CLK and IO on the Blue Pill end connections. Make sure you use the correct voltage pin on the ST-LINK. I used 3v3 and connected it to the pin marked "3" on my STM32F103 device.
Seems logical that the Naze32 bootloader would work well since the racing drone community has been using the STM32Fxxx CPU for quite some time now.
Indeed, sounds good. My main issue was that my STM32s were fakes (clones, not perfect ones). As soon as I bought from a reputable supplier my bootloader issues disappeared.
why not change the Resistors and Caps to 0805? I always use 0805
I tried this and they didn't fit on the pads, Jens.They are quite a bit bigger than the 0603 components.
I am still looking for a way to add the necesary files & packages to my IDE (without being connected to the internet!) to get it working. Just the way I do with my libraries which is very very easy. Is this possible? Just downloading some packages on a USB stick and copy them into some IDE folders. I still hope so.
Yes, just copy from the USB stick to the "libraries" folder in the Arduino IDE (if that's what you're using). They will be picked up automatically.
@@RalphBacon Thanks Ralph ☺️ but are you sure you do not mean the hardware folder? because that's what I tried and that did not work 🙁
Creality replaced the STM32F103RET6 with the GD32F303RET6 in the Ender 3 Pro. Is there a possibility of extracting the source code from the board? I am not knowledgeable about this topic. Anyway, I like the explanations in your video and the way you present yourself. Keep it up and who knows, maybe someone or I will find a way to figure this issue out.
You can read the contents of the μC (the uploader does this to verify the upload) unless the fuse bit has been set to prevent this. How you do this on the STM32F103 escapes me but must use avrdude command line.
Whether the code is going to help you (it might be very processor specific) is another question - and it's in hex.
Great video again.
Pls ignore trolls....
Thank you for sharing.
Glad you enjoyed it, Jedan, and yes, I will ignore trolls! Thanks for posting.
II gather your board is open source, you mentioned that LCPCB would supply boards fitted with components, any idea how much.
JLCPCB will supply SMD components on a board (one side only). You pay for the components as priced on LCSC.com (where they/you source the items from) plus the assembly fee (which is on their website). For small boards with few SMD components it's possibly not worth it. But a fully populated board with lots of closely packed SMD components it could be a life (or project) saver.
Well presented and engaging, thank you. :3
Glad you enjoyed it!
What about ADC quality of the clones? STM32 has a good ADC where ESP32 works not so exact and consistently.
I've not examined the quality, Olaf. I do know that at higher resolutions there may be some "noise" but the only way to find out whether that's a problem is to try it out.
hi there hay just wondering on avage what micro..... do you tend to use the most.. and one more i tend to used the uno in most in my projects (i just like it and pretty much has just the right amount of goodness for my projects) but what i do find a real pain in the ass is the fact that a lot of things tend to like 3.3 volts and not 5.0 volts if you hit this problem often what do you do to get around it i been using resistor as voltage dividers which seem to work ok but a real pain what's the answers lol.
Good question! If Wifi is required then an ESP8266 or ESP32 is simplest to use, although you can get some hybrid boards with an Arduino Uno and an ESP8266 WiFi module all on the UNO shaped board - but I've not used one.
The Wemos D1 Mini ESP8266s are good and reliable and you can plug things together easily all on 3v3. That's what I used for my Home Alone project.
For non-demanding projects I will use a Nano or a Arduino Mini which are basically Arduino UNOs in a smaller form factor. UNOs are great for developing but far too big for actual projects IMHO.
If no WiFi is required but a faster UNO-compatible board is needed then the STM32 Blue (or even Black) Pill is great value and so very fast. There's even a way to update or remove the bootloader now.
To interface 5v to 3v3, a resistor divider works (usually) but is fiddly; I have used bi-directional 4-way or 8-way level shifters in the past and they work really well but need a lot of connections on a breadboard but are a no-brainer when used with a PCB. I mention these in a video when I did on RFID Reader/Writers: ua-cam.com/video/g3a0X30Le-8/v-deo.html
Wait, I am confused. You run code which counts how many primes you have found, but finding the next prime should take more computational power than finding the previous one. Then how come the number of primes the ratio (primes stm32 found)/(primes arduino found) is much greater than (clock speed stm32)/(clock speed arduino)?
I give up, what's the answer? Forgive me, Beacon, this was just a rough and ready way of showing the STM32 was a gazillion times faster than a standard Arduino. I need to use the Dhrystone method in future which has something of a scientific basis!
Ralph S Bacon I just realized the reason, I’m dumb and forgot that arduino runs 8 bit and not 32 bit, if you take that into account everything makes sense :)
I have a chip CS32f103, stm32f103 all is well CS32=stm32???
Nope. The CS32 is a clone just like the GD32 or CKS32: www.cnx-software.com/2019/02/10/cs32-mcu-stm32-clone-bluepill-board/