I love watching you work through these design challenges. You're a great engineer! It's also a refreshing change to watch non sponsored content on YT. Sponsored videos have become almost unwatchable to me in recent months.
15:00 you're already paying for through hole and double sided load for those crappy dupont connectors. Use a good latching smd connector instead. Even if you have to notch the board to allow the wire to fit inside the enclosure it will be worth it. Those dupont connectors are unbearably crappy and are a likely source of your issues in a relatively high vibration environment like a lathe.
They are crappy, but they are almost certainly not the source of the glitching. He very likely has ringing on the clock and data pins due to fast edges being generated by the source and no termination.
23:52 One of the coolest aspects of this project is that there is essentially no curtain. Turning a potential customer service hassle due to a supply chain issue into a couple of interesting videos is a 200 IQ play.
Customers are much happier when they understand the problems of the supplier. When I have a problem, I work with my customers to fix the issues, involving them in it as much as possible. In that way several good things happen. First they feel valued and heard. Second, I get to pick their brains for solutions that I may have missed. Third, they tell their buddies that I care.
James. I had the flicker problem and fixed it by replacing the pull-up resistors on your board with lower value. I THINK is was 10K and I replaced them with 4.7K and problem solved. Not sure about the values, but I made them 1/2 the value of what it was. I replaced the resistors in the opposite end, on your board that sits on the controller. No more issues. If you decide at some time to upgrade to a bigger lathe, then I can provide pictures how I installed the ELS on my Birmingham 12X36 Lathe (Generic 12X36). Thanks for what you are doing for us. Very much appreciated.
James, you are a gem of a man. Your talents, design skills, concern for your customers and integrity are what make you so successful. Thanks for everything. Paul
I am very happy with my electronic lead screw. I did have noise on mine when first set it up. I did various screening. But eventually the problem was tracked down to not grounding the main control board to the chassis of the lathe. As Soon as I did that everything seems to be very stable. don't think I needed to have screened anything.😅
Watching James succumb to the chilly logic of efficient productization is both a great example of how engineering tradeoffs work and vaguely disheartening…
I am sure you have thought about this, but looking at the setup, with the metal box attached to the lathe, unless the screen on the cable does not connect to the die cast box you are introducing a ground loop. So either isolate the box from the lathe or not terminating the cable screen on one end or the other would help. I did watch your video years ago but have forgot if you covered this or not, if you did sorry for being a know it all commentator !
Hi, In my experience errors there are usually caused by ringing on the signal transitions. The easiest way to fix that is usually a bit of series resistance in each signal lead at the signal source end. That slows the edge down. Resistance value is not usually critical, anything from 22 to 100 ohms will probably work. I assume the DIO signal is bidirectional so a bit of resistance at each end may help. I gather the clock and strobe are sourced in something at the other end of the cable, so the easy fix isnt all on the display board. Could be in a cable adaptor.
@@feynthefallen yes, but ringing on DIO is probably least important. Well, at the master that is, the peripheral still needs a clean clock. I speculate cleaning up clock and strobe is likely sufficient. Source termination is the relevant buzzword for all this.
When I did surface mount soldering for work, we had a pair of soldering tweezers; those made removing and reworking small surface mount parts really easy.
You should also consider putting decoupling caps on the power pins on the ic. It would help to kill all excessive noise. Maybe a 10uf tantalum plus a .1uf ceramic.
Yes, add ~10uf decoupling. Add ~47R series resistance to the data/clock lines. For mechanical simplicity, have a look at the Wurth SMT threaded nuts. LCSC has Chinese-knockoff nuts for assembly too. That way you could remove the need for the printed spacer and locknuts.
@@hoggif These look like they are supposed to suppress power supply noise. On that note, an inductor in the power and ground lines might not be the worst idea. My guess is that the data lines need signal cleaning capacitors. The frequency on those signals is low enough that a 2.2nF from each signal to ground would de-noise the signal without degrading it. Only way to be sure would be to put a scope on them and scan for transients of course.
If you want to save a few more pennies and have a higher yield consider changing the LEDs to SMD. With that printed spacer if you extend it a little in that area you will get the light guide effect to stop bleeding from one to the other that might result otherwise. Hand assembling that many LEDs there's going to be some reversed. I'm guessing you're planning on low pass filtering the SPI lines for the noise? Be interesting to see how that works. Be even more interesting to see if you could diagnose the noise first, scope it or some such to find the source.
@@routercnc9517 Good question, but low pass is in the design of the mind of the passer. In some circuits, low pass rolls off at 500Hz, in others the roll off window is up in megahertz regions. It all has to do with the overall band width of the signals involved.
@@maikeydii thats a good point. i dont see a need for that button and if everything else needs / is getting a facelift it might save a bunch to ditch it
@@maikeydii Going back to the beginning, James talks about that button. It's a standard panel with numerous applications. Probably more expensive to eliminate the useless button than to keep it.
Once I needed to go long distance with 3v3 and of course I had the same capacitance problems you had at the start of this series, so I used a pair of level shifters and bumped it up to 12v. As a plus it turned out to be surprisingly noise resistant and easy to filter, getting a nice square wave out of the 3v3 side of the level shifter.
Not a bad idea. Take it all the way up to the industry standard 24 volts, then when you roll back down any induced noise is reduced by a factor of 5 or 6.
Nice to see you revisiting this project James. I’ve integrated it into my little lathe and it’s been working like a champ for ages. Quietly hoping this is just the start of a number of improvements. Thanks for this project. If implemented well, it’s a fantastic addition to any lathe and makes a big difference to its usability and efficiency.
Hi James. Impressive. I'm always prepared to support people who seek to improve their products. I already have your earlier display kit but have been unable to install it up to now due to some heart trouble. That's now being managed. So, when you have all this sorted I'm buying. I've sent a 'Super Thanks'. Cheers.
I’m going to be honest. As a low voltage electrician that specialized in controls. These are my favorite videos. You seem like a guy who would be a fun friend between me and my sparky friends. Thanks for your videos. They are always fun.
For anybody wondering why things cost a lot more that the basic parts cost this video is a good illustration of how much unseen work goes in to getting a product right. Whilst you're improving the product and effectively cost reducing it, I'd question why you fit switch 7 as it appears not to be used. A few cents saved, or do you use the boards for other things which make use of it?
It's all the non reoccurring engineering time, that is where the major initial cost arises. If you make 10 of a product, then the NRE per product is very high, but if you make millions of that product the NRE falls very low.
To combat the noise, you could write your own I2C driver to use the microcontroller push-pull (totem-pole) outputs for transmission, and revert to open-drain configuration when receiving. The lower impedance output should reduce the likelihood of noise on the line. Bit banging I2C is very simple, and the waveforms are very clearly explained, so it's not in any way difficult.
The timing of this video was very convenient. I finished installing my ELS yesterday and noticed that one of the LEDs was not working and now I know why. It has either slipped the quality control or more likely, failed during shipping or when I powered it up. I also cut the wings on the spacer to make the display assembly fit the cutout in the headstock where the old tachometer LCD used to be mounted.
Great video James! I really enjoy this sort. I read through all of the comments, 158 at the time of posting this. Many thoughts/suggestions to ponder. Rolling your own PCB makes for a trip down the rabbit hole with all the possibilities it opens up. Looking forward to seeing what changes you implement, both for the noise/glitch and other comments/suggestions. Thanks again.
Next he will be adding the Z axis DRO directly into the board so that a threading operation will involve putting in the thread pitch, the start point and the end point. They all the user has to do is advance the tool depth and hit one button to get the next pass.
i gotta say it... i see the "ough" in a word, and i instantly have the thought to have a thorough plough through the several ways i can pronounce it... and i like the "oo"... i get "clue for two..." cluff sounds so weird to me. :D
When using solder wick for those little guys, douse it in flux and the solder will wick even better. You did a good job with the SMT's. For cleanup work, if your pads have too much solder on them to look "pretty", you can wipe the tip of the iron off very clean and go back and re-heat the joint. The small amount of solder that transfers to the tip helps to take the excess away so you get the right amount of solder on the SMT's. It takes a few trips back and forth, but eventually you take the excess away.
I really love these "mixed media" projects. I hope that in the next video you'll explain a bit where this bit error comes from and how changing impedance help to fix this.
The ELS has been one of my favorite projects, I purchased the kit when it first came available and have yet to install the kit. I have been upgrading my lathe so much that I nearly forgot about the ELS project until now.
Glad to see you're doing your own design! I recently got a kit that had a bad board, keystrokes were missed 50% of the time, the screen would go all crazy if it was flexed ever so slightly (like a hard button press), and the LEDs and displays were all dim. Fortunately they're cheap and the one I got for $2 or $3 to replace it worked perfectly. Love my ELS and am really glad you are pushing it even further!
It's amazing in these times to find someone who really cares what they supply. If you where a company like ford, GM, Apple. etc they would just wash it under the carpet. Then they would blame the buyer for the fault so they keep their ever expanding profits, and to buy more (pollies) to ensure they never loss. will await for the new update product and thanks.
Another labour saving trick when panelising boards that may be possible here is to gang together all the input connectors via the panel edges (obviously putting the traces on the unscored side. Then you can test the whole panel at once with a single connection.
SPI isn't really intended as a board-to-board interconnect, though of course it can work for that reasonably well. For noise resistance you really want something with differential signalling like CAN, RS-485, or Ethernet. Probably CAN for a project like this. The obvious disadvantage is an increased BoM cost, and more design needed since you have to have transceivers and some way to translate the protocol at each end. You could probably run SPI over differential drivers, and thereby only need transceivers at the ends (no protocol translation). You also probably want to try to eliminate any through-hole parts you can to keep the cost down. An all-SMD assembly is usually far cheaper.
Thank you, James, for this video series, I really enjoy it a lot. Maybe you need to add more "Bypass Capacitors" to the panel's voltage rails as close as possible to the main chip, these capacitors will filter any power glitches, remember that you are using long cables. Maybe 10uF and 100nF capacitors in parallel should be fine for you.
While this is not a product I can use today I really apricate the process and how difficult it can be to produce a product at a high quality. Thank you for sharing.
I bought those years ago for working on RC helicopters, and they're hands-down the best hex drivers I've ever used. I'm sure there are some fancy German ones that are also really good, but these have stood up to everything I've thrown at them.
If the connector on the back is such a problem with multiple manufacturers, not just one "not feeling like it", maybe it'd be an idea to adapt the box it goes into to have more room, so the pins can be just bent around to the side? Not sure if the pins are soft or fragile, but if they allow bending, maybe that'd be an easy "fix". Depending on the mallability of the pins, maybe the yellow plastic thing can be snipped off and the pins bent around 180 to face backwards into the housing. If that makes it better than just ~90 off "outside". If they're long enough without the plastic and soft enough, maybe they can even be bent so they end up roughly on the backside as actually intended.
I think it's about time I buy one of these for my lathe! There are times where I just want to slow down the carriage speed and swapping out gears to do that is unpleasant. Being able to cut any thread pitch at the push of a button has gotten more like something I want too.
Interesting detail about the source schematic having the inputs wired out of numerical order... that could be like a trap street on a map, or it could equally be for some backwards compatibility reason.
James, I started binging on the ELS series a few days ago. Last year I bought a G9972Z which appears to be essentially identical to yours other than the "11x26" exaggerated capacity. It was a replacement for my "real" lathe, a Jet 1024 that had back gears, feeds in both directions, cross feeds in and out and a full gearbox with gear changes only for hybrid metric threading. Watching this series gives me hope that the G9972Z can almost become a real lathe. It's time for me to start collecting parts! Thank you! BTW- You might look into Rosco theatrical lighting gels for the filter. They're very thin (3-4 mil thickness) and made of a stable polycarbonate. Once upon a time I made my living as a photographer back in the days of real film and liquid processing chemicals. I would put dark green Rosco gel over red LED displays on the darkroom equipment to prevent damage to color film while handling it prior to processing. The displays were still very visible but didn't pass much light into the darkroom. Without the green filters the red LEDs lit the room enough to trash color film and fog B&W film. I know that's not important to this application but the contrast between the dark portions of the displays and the illuminated LED segments was striking. Just mentioning it because someone might be interested in experimenting with filter colors other than red. I've watched you for your machining videos for a long time. Your depth of knowledge of electronics, programming, machining, manufacturing and the multitude of skills required for this project is most impressive. The equipment around you suggests you've been through more rodeos than most.
I don’t recall how I found your channel but enjoy every video, even if I don’t entirely understand what you are doing 😊. You clear and concise process and directions are always appreciated.
If you are going through the trouble of making your own board you should consider using a differential transceiver to send the data from the panel to the main board. I had a ton of issues with noisy lines causing flickering screen, and ended up modifying a PCA9615 which is for I2C but can work for SPI as well
Every wire, going in or out of the display board, should be protected from noise. Loading the lines down with a resistor is a poor way of doing it. The power lines, should have two ceramic caps after the connector, directly across power and ground. First a 0.1uF, then a 2.2 to 10uF capacitor right after it. These capacitors, do not replace the 0.1uF that should be as close as possible to your display chip across power and ground. All signal lines, should have a low pass RC network on them, immediately after the connector pins. Typical values 0.01uF to 0.1uF, and 100 ohms to 2.2K ohms. Layout is important here. There should be a ground plane directly under these components, so one end of the capacitor can go directly to ground with the shortest possible path. The best scenario for selecting these values is run the lathe at its maximum RPM and monitor the signals with an oscilloscope. If you see the square waves being rounded off, you went too far on the RC filter. Here are some search terms, in Google image, search for this "input circuit RC noise filter", and in Google search for this, "Digikey Low Pass/High Pass Filter Calculator". P.S., absolutely amazing work you have done here, using the 3D printers to make a professional electronic device. One question though, on the original display boards with the pins sticking straight up. Could you bend them over 90 degrees without breaking the pins and have room for the connector?
I just wanted to drop in and say that I just got a Bambu X1 Carbon with AMS and I'm thrilled with it. I only mention this because back in the day, I think you and I were the only people on Thingiverse with Makerfarm Pegasus printers. I still have mine, but like the Ship of Thesus, I'm not sure if it's the same printer after all these years.
I have been playing with your code and kit... Those connectors are very tight in the through holes. I ordered some spare boards from ebay, and they were just junk. Good to see you are working on this. Great project!
In regards to your button clearance - I think it would've been better to edit the spacer a second time. You now have two revisions of 2 parts in the wild- one is easily identifiable (your spacer lost its wings) while the other (your buttons) needs to be measured with calipers. A great exercise to solve the issue though. Great work!
That is easy to fix, just print an extra nub or change the shape of one end of the button carrier section. Then you can tell them apart by feel in the dark.
I finished my ELS recently and even managed to upload a video series on my channel. I spent along time reducing electrical noise to make it function and it worked in the end but could not get rid of the occasional display glitch. I spent more time trying to fix what I thought was more electrical noise. In the end I realised it still functioned fine so I left it. Good to know I’m not going mad and it can be improved in the hardware. If the new display is offered stand alone I may consider buying one just to remove the issue. Great work thank you and I realise just how much time goes into this sort of thing!
One of the LED's died on the board that came in your kit. I didn't bother fixing it I just ordered a board from Amazon to replace it, they are so cheap. Now I know how to fix it.
I have enjoyed this video series so much. Working on my own ELS with some improvements and I really appreciate how much detail you have shared from your design process. For what it's worth, one option to eliminate noise issues would be to use a different bus type (CAN comes to mind) as an intermediary between your main board and your display board. CAN is incredibly noise tolerant albeit relatively complex to implement in hardware compared to SPI or I2C. But I imagine there's a simpler fix than that...looking forward to seeing what you come up with!
In Chris Chatelain's comment, he speaks to doing a level translation to 12 volts on those signal lines. This has the effect of quashing any noise by a factor of 3 or 4.
Another great video James. I like how your persistence to achieve perfection yields a final result that is a work of art. This was also insight into the process of manufacturing a consumer product.
Good for you, taking this on. ❤ You might check to see if there are even bond wires on the substrate of the failed LEDs. That’s assuming that there’s a substrate in there somewhere. 😂 Yes, I do electronics for a living. I feel the pain. A panel knife will really help separating the boards cleanly.
Fundamental problem is that the TM1638 TTL logic serial communication is not appropriate for your usecase. It's would probably not pass any EMC certification. If you think about the possibility of harm to the equipment and operator if for example a button read is wrong and changes something critical. I recommend doing a FMEA. And if it's not already so I would make the control panel "write only" while the spindle is moving. I would be tempted to use an I2C solution, it would probably be better but not 100% if you keep the length under 30cm. It has some built-in communication error detection. If I felt fansy I use the PCA9615 differential I2C transiver. There is always the option of going for FDCAN but it may be overkill.
Excellent! So many little details, most worthy of a facepalm. You demonstrate - in an excellent and accessible way - the engineering refinement cycle. You hate having to go around the loop, but the result is improved each time you do.
I have been thinking about building my own lathe and possibly a mill. I was thinking its alot of work to build it to thread screws. Then thought build it without. Then i was thinking what the point in that, may aswell thread also but how. This couldbr the answer , i want it to be atleast 6" chuck, and 48" capacity with two steady rest, with 6"-8" capacity, make it from welded steel, using fixtures, so the components can be precisely aligned and hopefull adjusted by using a file or adding shim stock. Use 4"x1" flat bar, and 1x1" bar. Precisely machined for the ways. Maybe cast a compound slide, about 10" bottom travel, and 8"-10" upper. With the chuck mounted high enough to have a 12"+ chuck, use a 3kw brushless dc motor, with speed and direction control, use a wide serpentine belt, it should offer a very quiet operation, use tapered truck wheel bearings for the spindle and either electronic Regen braking, the rotation converted to heat, or a light duty motorcycle disc brake, spring applied, with the shutoff lever. Having reverse ,neutral, brake, neutral, forward detents. And a decent size leadscrews, a carrage with double nuts to allow some jamming action to adjust clearance, just a idea, im guessing a els needs a decent size motor to push thr carrage!
Thanks, James for making the ELS series of videos. The project has completely transformed my import lathe. I did have the glitches while I was fabricating my setup, but they have (mostly) disappeared since my final install.
Having seen those glitches, and seeing your schematic, I'd say try putting some small capacitors, just a couple of nF, between your data lines and ground. Full disclosure, I'm not an electronics engineer, just a gifted amateur, but that is what I'd try first. I'd keep those pullups rather high too, because trying to make them smaller, you are only going to put extra strain on your signal pins. I think the only reason why changing the resistor values even mitigated the problem was that you swamped something in either the source or the sink of the signal.
As well as the decoupling caps on the IC (positioned as close to directly across the IC power pins as possible), if you're concerned about noise glitches then consider a cap from the high-impedance digital input of the IC down to 0V on each of the lines - again positioned as close to the IC pin as possible. Exact value should be determined looking with a scope to have as high as you can whilst still obtaining valid logic levels on the narrowest of pulses. The caps will form an RC with the series impedance of the long wire. If you want to have further control and better manage noise then add a series resistor of your own as well.
Should really take the opportunity to move to SMD LEDs, would probably save a dollar or two per board. Also, MELFs are a pain for SMT lies, standard SMD diodes are geerally preferred.
I've been using the ELS now for a while and it works perfectly. It would sure be great if you would continue with the electronics and add the ability to start and stop without using the half nut.
It's been interesting to watch you develop this setup.Thanks for sharing all these details. for my sins I'm current playing with building my own using just an arduino nano. it's been fun to work my way through issues and refine my own setup. :-)
It would be best to have a hot air station for surface mount soldering. They can be very cheap or costly, like most tools. If you spend $200-300, you can get a relatively inexpensive model that's of reasonably good quality. Look for Quick 861DW or Atten ST-862D (about 100 bucks cheaper than quick). Of course, you can get much higher quality units from JBC and the other usual suspects that manufacture pro soldering gear, but those are going to run you a good thousand dollars or more.
Just curious if you’ve considered getting a CO2 laser, like 50W to 100W and laser cutting the red lenses? If nothing else, it would be another neat toy! And I’m sure you could find a use for laser cut Acrylic somewhere.
I am upgrading an old craftsman 109 lathe that i got a while back. The original spindle is very flimsy and prevents it from being able to handle any real work on a lathe designed and marketed for real work. Ive been going back and forth whether to keep the change gears, or buy an electronic lead screw from you ( leaning the way of the ELS ). Is it possible to buy the board and components from you and solder/assemble everything myself to keep my cost down? Like most start up shops or hobby shops, the budget is tight. Love your videos and you are absolutely an inspiration on par with the legends. I.e. ToT, AvE, Abom, MrPete, and the rest of the greats, in my opinion. Thank you for doing what you do.
This is one of my all time favorite series on UA-cam. Thanks for all the time I've spent enjoying them.
Same. When I get short on stuff to watch, I just come back to this playlist and run through it again.
Thanks. For the ELS display upgrade.
I love watching you work through these design challenges. You're a great engineer!
It's also a refreshing change to watch non sponsored content on YT. Sponsored videos have become almost unwatchable to me in recent months.
15:00 you're already paying for through hole and double sided load for those crappy dupont connectors.
Use a good latching smd connector instead. Even if you have to notch the board to allow the wire to fit inside the enclosure it will be worth it.
Those dupont connectors are unbearably crappy and are a likely source of your issues in a relatively high vibration environment like a lathe.
Hell, JST is available in SMD right angle. 7 segment displays are available in SMD also, he could eliminate the handwork entirely.
They are crappy, but they are almost certainly not the source of the glitching. He very likely has ringing on the clock and data pins due to fast edges being generated by the source and no termination.
23:52 One of the coolest aspects of this project is that there is essentially no curtain. Turning a potential customer service hassle due to a supply chain issue into a couple of interesting videos is a 200 IQ play.
Customers are much happier when they understand the problems of the supplier. When I have a problem, I work with my customers to fix the issues, involving them in it as much as possible. In that way several good things happen. First they feel valued and heard. Second, I get to pick their brains for solutions that I may have missed. Third, they tell their buddies that I care.
James, I really like that you share the trials and tribulations.
very inspirational for a one man product shop.
James. I had the flicker problem and fixed it by replacing the pull-up resistors on your board with lower value. I THINK is was 10K and I replaced them with 4.7K and problem solved. Not sure about the values, but I made them 1/2 the value of what it was.
I replaced the resistors in the opposite end, on your board that sits on the controller. No more issues.
If you decide at some time to upgrade to a bigger lathe, then I can provide pictures how I installed the ELS on my Birmingham 12X36 Lathe (Generic 12X36).
Thanks for what you are doing for us. Very much appreciated.
James, you are a gem of a man. Your talents, design skills, concern for your customers and integrity are what make you so successful.
Thanks for everything.
Paul
As always, you hit it on the head. The process IS everything. You’re great James and your integrity shines through in every video you make! Bravo.
I am very happy with my electronic lead screw. I did have noise on mine when first set it up. I did various screening. But eventually the problem was tracked down to not grounding the main control board to the chassis of the lathe. As Soon as I did that everything seems to be very stable. don't think I needed to have screened anything.😅
The ELS is a great project and I'm glad you're selling through hundreds of kits on a regular basis!
Watching James succumb to the chilly logic of efficient productization is both a great example of how engineering tradeoffs work and vaguely disheartening…
I am sure you have thought about this, but looking at the setup, with the metal box attached to the lathe, unless the screen on the cable does not connect to the die cast box you are introducing a ground loop. So either isolate the box from the lathe or not terminating the cable screen on one end or the other would help.
I did watch your video years ago but have forgot if you covered this or not, if you did sorry for being a know it all commentator !
Hi, In my experience errors there are usually caused by ringing on the signal transitions. The easiest way to fix that is usually a bit of series resistance in each signal lead at the signal source end. That slows the edge down. Resistance value is not usually critical, anything from 22 to 100 ohms will probably work. I assume the DIO signal is bidirectional so a bit of resistance at each end may help. I gather the clock and strobe are sourced in something at the other end of the cable, so the easy fix isnt all on the display board. Could be in a cable adaptor.
DIO definitely is bidirectional if you are using buttons, yes.
I hope he sees this, was going to say the same thing. He needs to get a scope on it to see what's happening tbh.
@@feynthefallen
yes, but ringing on DIO is probably least important. Well, at the master that is, the peripheral still needs a clean clock. I speculate cleaning up clock and strobe is likely sufficient. Source termination is the relevant buzzword for all this.
@@ferrumignis Exactly - without knowing what is causing the problem, it's just shooting in the dark.
Wonderful example of the iterative design process.
Yep, make the best pass that you can, put it out and see what happens, improve and go on.
When I did surface mount soldering for work, we had a pair of soldering tweezers; those made removing and reworking small surface mount parts really easy.
That's something to add to my wish list.
@@Clough42 The hot tweezers do work well.
You should also consider putting decoupling caps on the power pins on the ic. It would help to kill all excessive noise. Maybe a 10uf tantalum plus a .1uf ceramic.
I wonder if C1 and C2 on board are decoupling caps as on the reference design?
Yes, add ~10uf decoupling. Add ~47R series resistance to the data/clock lines.
For mechanical simplicity, have a look at the Wurth SMT threaded nuts. LCSC has Chinese-knockoff nuts for assembly too. That way you could remove the need for the printed spacer and locknuts.
@@hoggif These look like they are supposed to suppress power supply noise. On that note, an inductor in the power and ground lines might not be the worst idea. My guess is that the data lines need signal cleaning capacitors. The frequency on those signals is low enough that a 2.2nF from each signal to ground would de-noise the signal without degrading it. Only way to be sure would be to put a scope on them and scan for transients of course.
If you want to save a few more pennies and have a higher yield consider changing the LEDs to SMD. With that printed spacer if you extend it a little in that area you will get the light guide effect to stop bleeding from one to the other that might result otherwise. Hand assembling that many LEDs there's going to be some reversed.
I'm guessing you're planning on low pass filtering the SPI lines for the noise? Be interesting to see how that works. Be even more interesting to see if you could diagnose the noise first, scope it or some such to find the source.
I’m no electronics expert but is there a risk of an LP filter introducing a phase delay on a line which might need critical timing?
@@routercnc9517 Good question, but low pass is in the design of the mind of the passer. In some circuits, low pass rolls off at 500Hz, in others the roll off window is up in megahertz regions. It all has to do with the overall band width of the signals involved.
Another cent or two saved could be achieved by ditching the extra, apparently unused switch between the + and - buttons.
@@maikeydii thats a good point. i dont see a need for that button and if everything else needs / is getting a facelift it might save a bunch to ditch it
@@maikeydii Going back to the beginning, James talks about that button. It's a standard panel with numerous applications. Probably more expensive to eliminate the useless button than to keep it.
Once I needed to go long distance with 3v3 and of course I had the same capacitance problems you had at the start of this series, so I used a pair of level shifters and bumped it up to 12v. As a plus it turned out to be surprisingly noise resistant and easy to filter, getting a nice square wave out of the 3v3 side of the level shifter.
Not a bad idea. Take it all the way up to the industry standard 24 volts, then when you roll back down any induced noise is reduced by a factor of 5 or 6.
I could have sworn the last time I watched this channel it was in English. But this video is clearly in Greek.
Very interesting to see the productuon process optimization and simplification that develops over months and years.
Nice to see you revisiting this project James. I’ve integrated it into my little lathe and it’s been working like a champ for ages. Quietly hoping this is just the start of a number of improvements. Thanks for this project. If implemented well, it’s a fantastic addition to any lathe and makes a big difference to its usability and efficiency.
I bet you could make improvements and submit a pull request ;)
James, I love your attention to detail. Nice work! Great video
Hi James. Impressive. I'm always prepared to support people who seek to improve their products. I already have your earlier display kit but have been unable to install it up to now due to some heart trouble. That's now being managed. So, when you have all this sorted I'm buying. I've sent a 'Super Thanks'. Cheers.
For the lip between the f panel and box, add four plastic washers to captivate the four screws.
I’m going to be honest. As a low voltage electrician that specialized in controls. These are my favorite videos. You seem like a guy who would be a fun friend between me and my sparky friends. Thanks for your videos. They are always fun.
For anybody wondering why things cost a lot more that the basic parts cost this video is a good illustration of how much unseen work goes in to getting a product right.
Whilst you're improving the product and effectively cost reducing it, I'd question why you fit switch 7 as it appears not to be used. A few cents saved, or do you use the boards for other things which make use of it?
It's all the non reoccurring engineering time, that is where the major initial cost arises. If you make 10 of a product, then the NRE per product is very high, but if you make millions of that product the NRE falls very low.
Thanks for all the attention to detail James!
To combat the noise, you could write your own I2C driver to use the microcontroller push-pull (totem-pole) outputs for transmission, and revert to open-drain configuration when receiving. The lower impedance output should reduce the likelihood of noise on the line. Bit banging I2C is very simple, and the waveforms are very clearly explained, so it's not in any way difficult.
It uses SPI
Oh no, it’s going to need another video? Good thing we lobe watching those things 😂
You are an inspiration!
The timing of this video was very convenient. I finished installing my ELS yesterday and noticed that one of the LEDs was not working and now I know why. It has either slipped the quality control or more likely, failed during shipping or when I powered it up. I also cut the wings on the spacer to make the display assembly fit the cutout in the headstock where the old tachometer LCD used to be mounted.
Great video James! I really enjoy this sort.
I read through all of the comments, 158 at the time of posting this. Many thoughts/suggestions to ponder. Rolling your own PCB makes for a trip down the rabbit hole with all the possibilities it opens up. Looking forward to seeing what changes you implement, both for the noise/glitch and other comments/suggestions.
Thanks again.
Next he will be adding the Z axis DRO directly into the board so that a threading operation will involve putting in the thread pitch, the start point and the end point. They all the user has to do is advance the tool depth and hit one button to get the next pass.
Sorry that happened. Great job checking the quality. You the man! Way to lookout for your folks and have standards! Appreciate the extra effort.
i gotta say it...
i see the "ough" in a word, and i instantly have the thought to have a thorough plough through the several ways i can pronounce it...
and i like the "oo"...
i get "clue for two..."
cluff sounds so weird to me.
:D
A master class in rapid prototyping. Hat doffed.
When using solder wick for those little guys, douse it in flux and the solder will wick even better.
You did a good job with the SMT's.
For cleanup work, if your pads have too much solder on them to look "pretty", you can wipe the tip of the iron off very clean and go back and re-heat the joint. The small amount of solder that transfers to the tip helps to take the excess away so you get the right amount of solder on the SMT's. It takes a few trips back and forth, but eventually you take the excess away.
I really love these "mixed media" projects. I hope that in the next video you'll explain a bit where this bit error comes from and how changing impedance help to fix this.
Adding decoupling capacitors on the ic power pins will help out with filtering unwanted noise.
Not is the noise is on the signal lines, which seems likely.
The ELS has been one of my favorite projects, I purchased the kit when it first came available and have yet to install the kit. I have been upgrading my lathe so much that I nearly forgot about the ELS project until now.
Glad to see you're doing your own design! I recently got a kit that had a bad board, keystrokes were missed 50% of the time, the screen would go all crazy if it was flexed ever so slightly (like a hard button press), and the LEDs and displays were all dim. Fortunately they're cheap and the one I got for $2 or $3 to replace it worked perfectly. Love my ELS and am really glad you are pushing it even further!
It's amazing in these times to find someone who really cares what they supply. If you where a company like ford, GM, Apple. etc they would just wash it under the carpet. Then they would blame the buyer for the fault so they keep their ever expanding profits, and to buy more (pollies) to ensure they never loss.
will await for the new update product and thanks.
I applaud you for being the perfectionist! Your work and videos are always top notch! Thanks again James!
Another labour saving trick when panelising boards that may be possible here is to gang together all the input connectors via the panel edges (obviously putting the traces on the unscored side. Then you can test the whole panel at once with a single connection.
SPI isn't really intended as a board-to-board interconnect, though of course it can work for that reasonably well. For noise resistance you really want something with differential signalling like CAN, RS-485, or Ethernet. Probably CAN for a project like this. The obvious disadvantage is an increased BoM cost, and more design needed since you have to have transceivers and some way to translate the protocol at each end. You could probably run SPI over differential drivers, and thereby only need transceivers at the ends (no protocol translation).
You also probably want to try to eliminate any through-hole parts you can to keep the cost down. An all-SMD assembly is usually far cheaper.
Well the final solution is to move it into a bidirectional fiber optic system where the data is just light. Then no spikes.
Thank you, James, for this video series, I really enjoy it a lot. Maybe you need to add more "Bypass Capacitors" to the panel's voltage rails as close as possible to the main chip, these capacitors will filter any power glitches, remember that you are using long cables. Maybe 10uF and 100nF capacitors in parallel should be fine for you.
While this is not a product I can use today I really apricate the process and how difficult it can be to produce a product at a high quality. Thank you for sharing.
Those ground drill rod tipped hex drivers are DYNAMITE!
I bought those years ago for working on RC helicopters, and they're hands-down the best hex drivers I've ever used. I'm sure there are some fancy German ones that are also really good, but these have stood up to everything I've thrown at them.
If the connector on the back is such a problem with multiple manufacturers, not just one "not feeling like it", maybe it'd be an idea to adapt the box it goes into to have more room, so the pins can be just bent around to the side? Not sure if the pins are soft or fragile, but if they allow bending, maybe that'd be an easy "fix".
Depending on the mallability of the pins, maybe the yellow plastic thing can be snipped off and the pins bent around 180 to face backwards into the housing. If that makes it better than just ~90 off "outside". If they're long enough without the plastic and soft enough, maybe they can even be bent so they end up roughly on the backside as actually intended.
love seeing how you work through problems. keep it up!
I think it's about time I buy one of these for my lathe! There are times where I just want to slow down the carriage speed and swapping out gears to do that is unpleasant. Being able to cut any thread pitch at the push of a button has gotten more like something I want too.
Interesting detail about the source schematic having the inputs wired out of numerical order... that could be like a trap street on a map, or it could equally be for some backwards compatibility reason.
James, I started binging on the ELS series a few days ago. Last year I bought a G9972Z which appears to be essentially identical to yours other than the "11x26" exaggerated capacity. It was a replacement for my "real" lathe, a Jet 1024 that had back gears, feeds in both directions, cross feeds in and out and a full gearbox with gear changes only for hybrid metric threading. Watching this series gives me hope that the G9972Z can almost become a real lathe. It's time for me to start collecting parts! Thank you!
BTW- You might look into Rosco theatrical lighting gels for the filter. They're very thin (3-4 mil thickness) and made of a stable polycarbonate. Once upon a time I made my living as a photographer back in the days of real film and liquid processing chemicals. I would put dark green Rosco gel over red LED displays on the darkroom equipment to prevent damage to color film while handling it prior to processing. The displays were still very visible but didn't pass much light into the darkroom. Without the green filters the red LEDs lit the room enough to trash color film and fog B&W film. I know that's not important to this application but the contrast between the dark portions of the displays and the illuminated LED segments was striking. Just mentioning it because someone might be interested in experimenting with filter colors other than red.
I've watched you for your machining videos for a long time. Your depth of knowledge of electronics, programming, machining, manufacturing and the multitude of skills required for this project is most impressive. The equipment around you suggests you've been through more rodeos than most.
I don’t recall how I found your channel but enjoy every video, even if I don’t entirely understand what you are doing 😊. You clear and concise process and directions are always appreciated.
If you are going through the trouble of making your own board you should consider using a differential transceiver to send the data from the panel to the main board. I had a ton of issues with noisy lines causing flickering screen, and ended up modifying a PCA9615 which is for I2C but can work for SPI as well
I've wanted this since I knew it existed since my old POS geartrains are so noisy. Can't wait for the final version.
Every wire, going in or out of the display board, should be protected from noise. Loading the lines down with a resistor is a poor way of doing it. The power lines, should have two ceramic caps after the connector, directly across power and ground. First a 0.1uF, then a 2.2 to 10uF capacitor right after it. These capacitors, do not replace the 0.1uF that should be as close as possible to your display chip across power and ground.
All signal lines, should have a low pass RC network on them, immediately after the connector pins. Typical values 0.01uF to 0.1uF, and 100 ohms to 2.2K ohms. Layout is important here. There should be a ground plane directly under these components, so one end of the capacitor can go directly to ground with the shortest possible path.
The best scenario for selecting these values is run the lathe at its maximum RPM and monitor the signals with an oscilloscope. If you see the square waves being rounded off, you went too far on the RC filter. Here are some search terms, in Google image, search for this "input circuit RC noise filter", and in Google search for this, "Digikey Low Pass/High Pass Filter Calculator".
P.S., absolutely amazing work you have done here, using the 3D printers to make a professional electronic device.
One question though, on the original display boards with the pins sticking straight up. Could you bend them over 90 degrees without breaking the pins and have room for the connector?
I just wanted to drop in and say that I just got a Bambu X1 Carbon with AMS and I'm thrilled with it. I only mention this because back in the day, I think you and I were the only people on Thingiverse with Makerfarm Pegasus printers. I still have mine, but like the Ship of Thesus, I'm not sure if it's the same printer after all these years.
I have been playing with your code and kit... Those connectors are very tight in the through holes. I ordered some spare boards from ebay, and they were just junk. Good to see you are working on this. Great project!
In regards to your button clearance - I think it would've been better to edit the spacer a second time. You now have two revisions of 2 parts in the wild- one is easily identifiable (your spacer lost its wings) while the other (your buttons) needs to be measured with calipers. A great exercise to solve the issue though. Great work!
Maybe print the new, shorter buttons in a different color to easily identify them?
That is easy to fix, just print an extra nub or change the shape of one end of the button carrier section. Then you can tell them apart by feel in the dark.
I finished my ELS recently and even managed to upload a video series on my channel. I spent along time reducing electrical noise to make it function and it worked in the end but could not get rid of the occasional display glitch. I spent more time trying to fix what I thought was more electrical noise. In the end I realised it still functioned fine so I left it. Good to know I’m not going mad and it can be improved in the hardware. If the new display is offered stand alone I may consider buying one just to remove the issue. Great work thank you and I realise just how much time goes into this sort of thing!
A dot of flux on your SolderWick can really help with getting solder up quickly, which helps to prevent damage to the pads.
these modules are usually used for development hence the pin direction and the push fit aids the assembly
It's all about the processes AND design for assembly. I can't imagine making mine in any volume and selling them 🤣
One of the LED's died on the board that came in your kit. I didn't bother fixing it I just ordered a board from Amazon to replace it, they are so cheap. Now I know how to fix it.
I have enjoyed this video series so much. Working on my own ELS with some improvements and I really appreciate how much detail you have shared from your design process.
For what it's worth, one option to eliminate noise issues would be to use a different bus type (CAN comes to mind) as an intermediary between your main board and your display board. CAN is incredibly noise tolerant albeit relatively complex to implement in hardware compared to SPI or I2C. But I imagine there's a simpler fix than that...looking forward to seeing what you come up with!
In Chris Chatelain's comment, he speaks to doing a level translation to 12 volts on those signal lines. This has the effect of quashing any noise by a factor of 3 or 4.
Thanks!
Another great video James. I like how your persistence to achieve perfection yields a final result that is a work of art. This was also insight into the process of manufacturing a consumer product.
A small 1-10nF, ceramic cap towards ground on each of the inputs should do the trick.
Tune the capacitance to where you get an acceptable slew rate.
If we're dealing with external EMI, this will help.
Everybody and their brother had Dynamite hardened Allen drivers. I reground my fair share of those a few years back. Good product.
Now you need an electronic cross slide screw yet so you can program a taper or radius via the values of the drop scales
Good for you, taking this on. ❤
You might check to see if there are even bond wires on the substrate of the failed LEDs. That’s assuming that there’s a substrate in there somewhere. 😂
Yes, I do electronics for a living. I feel the pain.
A panel knife will really help separating the boards cleanly.
Fundamental problem is that the TM1638 TTL logic serial communication is not appropriate for your usecase.
It's would probably not pass any EMC certification.
If you think about the possibility of harm to the equipment and operator if for example a button read is wrong and changes something critical.
I recommend doing a FMEA. And if it's not already so I would make the control panel "write only" while the spindle is moving.
I would be tempted to use an I2C solution, it would probably be better but not 100% if you keep the length under 30cm. It has some built-in communication error detection. If I felt fansy I use the PCA9615 differential I2C transiver.
There is always the option of going for FDCAN but it may be overkill.
very nice James
Excellent! So many little details, most worthy of a facepalm. You demonstrate - in an excellent and accessible way - the engineering refinement cycle. You hate having to go around the loop, but the result is improved each time you do.
I have been thinking about building my own lathe and possibly a mill. I was thinking its alot of work to build it to thread screws. Then thought build it without. Then i was thinking what the point in that, may aswell thread also but how. This couldbr the answer , i want it to be atleast 6" chuck, and 48" capacity with two steady rest, with 6"-8" capacity, make it from welded steel, using fixtures, so the components can be precisely aligned and hopefull adjusted by using a file or adding shim stock. Use 4"x1" flat bar, and 1x1" bar. Precisely machined for the ways. Maybe cast a compound slide, about 10" bottom travel, and 8"-10" upper. With the chuck mounted high enough to have a 12"+ chuck, use a 3kw brushless dc motor, with speed and direction control, use a wide serpentine belt, it should offer a very quiet operation, use tapered truck wheel bearings for the spindle and either electronic Regen braking, the rotation converted to heat, or a light duty motorcycle disc brake, spring applied, with the shutoff lever. Having reverse ,neutral, brake, neutral, forward detents. And a decent size leadscrews, a carrage with double nuts to allow some jamming action to adjust clearance, just a idea, im guessing a els needs a decent size motor to push thr carrage!
Better shielding for the cable to prevent those glitches from the motor EMF
Thanks, James for making the ELS series of videos.
The project has completely transformed my import lathe.
I did have the glitches while I was fabricating my setup, but they have (mostly) disappeared since my final install.
shielded cabling and ferrite beads can help a lot with EM noise. But maybe not worth the cost if you're able to fix it on the board.
I've had mine for awhile but haven't had time to install it on my lathe yet. It will probably be a winter project. The QC is top notch.
you should scope the signal while operating and see what the noise looks like... you might need ferrite cores somewhere or some sort of noise filter..
Having seen those glitches, and seeing your schematic, I'd say try putting some small capacitors, just a couple of nF, between your data lines and ground. Full disclosure, I'm not an electronics engineer, just a gifted amateur, but that is what I'd try first. I'd keep those pullups rather high too, because trying to make them smaller, you are only going to put extra strain on your signal pins. I think the only reason why changing the resistor values even mitigated the problem was that you swamped something in either the source or the sink of the signal.
Possible. The C2000 MCU GPIOs can sink 4mA. I think the EMI is on-board, though, and self-inflicted.
That was fascinating, thanks!
If you order those breakout pins seperately and solder them yourself underneath, you'll avoid significant 2-side assembly cost.
As well as the decoupling caps on the IC (positioned as close to directly across the IC power pins as possible), if you're concerned about noise glitches then consider a cap from the high-impedance digital input of the IC down to 0V on each of the lines - again positioned as close to the IC pin as possible.
Exact value should be determined looking with a scope to have as high as you can whilst still obtaining valid logic levels on the narrowest of pulses. The caps will form an RC with the series impedance of the long wire. If you want to have further control and better manage noise then add a series resistor of your own as well.
Should really take the opportunity to move to SMD LEDs, would probably save a dollar or two per board. Also, MELFs are a pain for SMT lies, standard SMD diodes are geerally preferred.
Mine lasted 1 day had to buy a new one off ebay. Love the ELS and would add it to any Lathe I got if I was able too.
I will definitely buy one, so love your work
Excellent video!
God I enjoy product development. Real nice to see those changes to simplify production!
Iterate iterate iterate (-8
I've been using the ELS now for a while and it works perfectly. It would sure be great if you would continue with the electronics and add the ability to start and stop without using the half nut.
It's been interesting to watch you develop this setup.Thanks for sharing all these details. for my sins I'm current playing with building my own using just an arduino nano. it's been fun to work my way through issues and refine my own setup. :-)
It would be best to have a hot air station for surface mount soldering. They can be very cheap or costly, like most tools. If you spend $200-300, you can get a relatively inexpensive model that's of reasonably good quality. Look for Quick 861DW or Atten ST-862D (about 100 bucks cheaper than quick). Of course, you can get much higher quality units from JBC and the other usual suspects that manufacture pro soldering gear, but those are going to run you a good thousand dollars or more.
Just curious if you’ve considered getting a CO2 laser, like 50W to 100W and laser cutting the red lenses? If nothing else, it would be another neat toy! And I’m sure you could find a use for laser cut Acrylic somewhere.
I did the touch screen version . And i love it
for electronics : EEVBlog forum
shielded cable or opto isolators 817 or schmitt trigger IC to filter
I am upgrading an old craftsman 109 lathe that i got a while back. The original spindle is very flimsy and prevents it from being able to handle any real work on a lathe designed and marketed for real work.
Ive been going back and forth whether to keep the change gears, or buy an electronic lead screw from you ( leaning the way of the ELS ). Is it possible to buy the board and components from you and solder/assemble everything myself to keep my cost down? Like most start up shops or hobby shops, the budget is tight. Love your videos and you are absolutely an inspiration on par with the legends. I.e. ToT, AvE, Abom, MrPete, and the rest of the greats, in my opinion. Thank you for doing what you do.
Great Update James, Ive setup the ELS 18mths ago on a SIEG SC6 and works great
I just love my ELS, use it daily!
Awesome! I just bought a new lathe, it an 330x600mm HBM. I would love to install this electronic leadscrew! Please let us know when it is for sale.