Really interesting project. I am doing a PhD on power electronics and I have previously studied how to improve buck converters. There is a topology called "active clamp ZVS buck Converter" that besides using synchronous rectification with mosfets allows to have ZVS in the switching of the mosfets, which greatly reduces the switching losses. In addition, due to its behavior, it allows placing several modules in parallel to increase the total power without additional control, since it distributes the current automatically.
@@paugasolina5048 its funny you call someone a no lifer when you are on this thread chugging haterade. Must just be self reflecting or something. At least his post relates to the video.
FYI, many STM32 MCUs have a PWM mode with complimentary outputs and dead state insertion -- specifically for switched mode power supplies and brushless DC motors. I didn't really understand the connection until watching this video. Thanks!
@@PerchEagle If I understand correctly, the STM32 can generate the complimentary outputs using a single output compare channel (the 328P seems to need two, with the same compare value, but opposite output polarity). Also, the STM32 can insert dead time (neither output enabled) automatically. I think the 328P would need two timers with different TOP/BOTTOM values (or just different initial counter values?) to produce dead time.
Yes, you do. For example, the STM32G474RE available on evaluation kits has a peripheral called HRTIM. A timer capable of generate a pwm with 4GHz clock. It has multiple outputs and features that makes a synchronous buck converter easy to implement. There's also some APIs sponsored by a ST's partner called BIRICHA with professional tools used for design a 2p2z digital controller using this peripheral. Yes, z transforms and transfer functions simulated and measured all in the ecosystem. Have fun.
Sounds like How your GPUs/VGAs' Chipset and VRAM module voltage are supplied. A mosfet driver, Low and High Side Mosfets, A Capacitor, and an inductor/Coil. Interesting!
@@maxhouseman3129 yeah, i saw modern GPUs like 20 series use some kind integrated mosfet-driver because tons of Vcore phases they use and they don't use much space like older GPUs.
@@farizfadillah7557 they call those powerstages. Crazy beefy chips. One of them can easily suppy 20-30 amps without overheating and the high end ones rated for very high currents can push 50 amps. If a GPU pulls 300W at roughly 1V you have 300 amps going trough the VRM. The VRMs on modern GPUs are really the most advanced dc to dc converters out there. They convert 12V to something as low as a single volt while retaining 90+ percent efficient and delivering a few hundred watts.
Great scott, sir. I know it's less likely that you read this. But I love how much clearer you explain everything steps/components now. When I was first year hs, I can't understand a thing in your videos, it's full of maths and very fast. It's just great to look at but nothing to learn from. Only your same level could understand. But since they are your same level, they couldn't also learn anything. Now that I'm getting college that I understand those videos, but still they are very fast and unclear functions of components. Now you are very much more interactive. Much more clear in explaination. And much longer videos but still direct and great written videos. I actually learn many things now. Thank you so much!
A left handed engineer is a dangerous weapon indeed! Left handed people think differently and are typically more holistic thinkers and creative types. This is a generalization of course, but left handed people tend to be more creative and not have the tendency of falling into STEM fields. So when we see a left handed engineer, it is truly something special because they bring a unique and different approach to problem solving. Again, this is a generalization but there is something to it. I have a friend who is a great programmer and who is also left handed. He always has a different and unique way of approaching problem solving that is different but also efficient and elegant. I really enjoyed this video!
Thanks Scott! Now I'm 'very glad' I picked a 'Synchronous Converter' for supplying my 150W pure sine inverter with the correct voltage when using two of my EGO (56VDC) batteries ... Makes for a nice, small, emergency backup and camping power unit. 10AH@56V ... 560WH. 5VDC, 12VDC and 120VAC as needed.🙂 Nice job on making one!
Years ago, a very determined man by the name of Walt Jung was on the rampage to innovate a voltage regulator that could do its job and handle higher currents that’s the typical 0.8 amps that we are familiar with. He wrote some great articles and he produced some schematics that are just as valuable!! I suggest that you would be interested in playing with his designs, or else looking at them if you already haven’t. I believe you would probably take away much more meaning and valuable info than I did. Thanks for sharing again!!
7:32....oh right, I've seen that setup before in laptop power supplies, but never really understood how they worked without a diode - this makes sense now. They're also susceptible to killing the load if the top FET shorts and sends VCC to the load directly.....
I dont have a more new ideas for you but I can totally say that ur are the most detailed videos I have ever seen. I keep trying to learn from ur videos.. thx for encouraging everyone.
After adding a feedback system. I hope you can go forward and talk about isolated dc to dc converters. The bridge configurations. Zvs. Zcs and phase shifted pwm are very interesting topics and I would love to see what you come up with as a real circuit because I could only get the design due to lack of a workshop of any sort. But yes please a feedback system first. You have a way of making these topics simple to understand to start out
Damn that's some great efficiency numbers I saw today from a DC to Dc converters. Yesterday in my viva, I was asked what can be the efficiency of a DC to DC Buck Converter and books said about 80%. I think this can improve power draw of many project that are about portability. Great Video as always!
Efficiency is largely a function of size, load, and conversion factor. If you can make a switching converter larger, you can use a lower loss inductor. If the load is lower than all the magnetic and conduction losses are lower, and if the conversion is a small step than the converter can act more like a wire.
Oh yeah these days you can quite easily build DC converters that operated well above 90%. The part that is hard is when you need to balance other design aspects such as size, cost, EMI, and if you want high efficiency at a specific operating point versus across a wide range of input and output voltage and current range.
What I enjoyed the most, was the fact that you thought off and catered for the "new to electronics audience" thank you for it and akthough this is is a bit afvanced for me at this stage, I will definately tune back into this video.
Sometimes, the AC resistance of the inductor can be a tricky bit, as, at short duty cycles, the apparent frequency content of the inductor current can be high enough to cause several % loss. And, of course, PCB layout can affect performance due to parasitics that reduce switching performance ....and so on, as I'm almost sicher that you know. Nice video. BTW, the ripple frequency and amplitude when in DCM, lets you do some calculations of parasitic capacitance and ACR. "Left as a proof by the student" ;-)
About the inductor, I tried some experiments with switch mode chips, and found I could not just wing it with the inductor. I had some ferrite toroids and wound them up. The uH was good but if you have the wrong core material - not so good. So anyone attempting this sort of project, look into the recommended toroid for the circuit.
I wish you talked about ground bounce too, at least a bit. It's quite an important thing when it comes to making compliant products that have switchmode converters. Great video though!
Also a neat observation you might like, if you look at the generalized/abstract schematics for buck and boost converters (using switch symbols instead of practical implementations, you will see that they are in fact the same. Just the port where power input and output are flipped around. You can make a buck converter, feed the voltage into the output, and use the input of the buck converter as the output to turn it into a boost converter! This does not take into account any practical aspects of the controller or MOSFET drive but it should be possible!
I recommend using a bootstrap IC driver for your MOSFETs, particularly where you have a high-side FET. I like the IRS2186 but the stupid thing is out of stock because of the silicon shortage. You could make your own but the tricky bit is how to control a high side driver with low-side logic. Internally these ICs might use an optocoupler (at the cost of propagation delay), others use a high voltage emitter follower (also limited in propagation and bandwidth due to miller effects). The fastest ones appear to use a transformer akin to a GDT or capacitive coupling into the inputs of a SR latch on the high side. It might make a good video topic!
U are the best electronics youtube i have ever seen. learnt alot , i really have an intrest in electronic components, like i open every electronic devices that are not in use and extract pcb. after i complete my high school in 5 months i am gonna do Electronics and Communication Engineering(ECE), then after than i want to build, invent, create things that can potentially change our lives ;)
This is awesome timing. I am actually building this same circuit this week, except for the controller I am using a UC3843 IC and with either type II compensation network around it for current mode control. I am building this circuit primarily to test my new Picoscope 5444D and it’s frequency response analyzer tool to actually plot the bode plot and measure phase margin.
Know the strength and weakness of the different technologies and select accordingly. Easy to say but development can be quite fast so the correct answer might change. Bipolar transistors might be the best choice sometimes and got it strengths.
Great video as always 👍😀 One little trick when soldering SMDs: put the required amount of tin on the pad, and 'cut it off with the solding-tip' and then drive it with solding tip to both the pad and the compont-lead. It is faster this way, and you have more control of the amount of solder on the compont/pad. And do the same on the first pad too. You can adjust the amount of tin, by cutting the length of the tin, or using a thinner tin on SMD😀 Thanks for sharing your great walkthrough of the electronic wonders of the world 👍😀
@@juanmontes8905 How would you use the reference as a controller? Yeah, if you don’t compensate the Buck it will oscillate depending on the ESR of the inductor and output capacitor.
I'm voting for a software-implemented feedback loop here. Much more cool. Much more hackable. Much more programmable. In fact, if he solders just another halfbridge module, he can get a complete pure sinewave DC-AC with just a software mod :D Or just a multichannel DC supply that can both charge and discharge a battery :)
The more I learn about switchmode power supplies, the more they just look like fancy motor drivers - that's clearly a half-H-bridge. Now if only I could put that knowledge to use 😁
I was just about to build my synchronous buck converter, but after watching your video i think i'd have to make some adjustements like adding some decoupling capacitors and diodes to discharge mosfet's gate capacitor ! Your videos are always GREAT Scott thanks !
Thanks for explaining! I've done a few integrated buck conversion designs in my circuits, and I'm really only familiar with the diode type topology, that explanation opened my eyes
Biggest deference is that mother boards use Multiphase Bucks. Rather than having to add a ton of output capacitors, they can lower the ripple by using several inductors that take turns charging.
Scott, this is one of the best videos about DC-DC converters i've seen!!! Such a good explanation with graphs and examples. I Appreciate your great work put in this video! Will be interested in next topic: efficient BOOST converters and are they possible?
That's easy. Diy if you want the most bang for the buck, buy if you want it to just work. Diy requires a bunch of trial and error to get everything working good. Plus, you have to be extremely comfortable with tearing a lot of stuff apart if you get a clog/jam, etc. That goes for both diy and buy though. For context, I built a modified version of the HevORT for about $600-$700. There's nothing even close to how good and fast it is if you wanted to buy something at the same price.
@@androiduberalles not exactly the kind of budget 3d printers u can buy theses, days totally defeat the purpose of diy-ing a budget one, but for larger or complex printers diy-ing makes a lot of sense, imo just but a decent printer like the ender 3 and then u can diy a very decent 3d printer cuz u can easily print most of the parts
I only recently learned this from UA-cam, but your soldering exploits would be made much easier with a generous application of flux. I've been using some colophony I got on ebay from the Ukraine, works a treat and leaves my room smelling forest fresh! But seriously, the Louis Rossman school of "Would you like some solder with your flux" cannot be understated. Solder becomes so much more cooperative. There is no such thing as too much, but there is certainly not enough.
I don't think he's having a hard time though. If you buy actually high quality solder it's got at least 2% flux in it anyway, and if you work hot and quick it easily suffices for the initial joint, easily! And it's good flux too, not just dirty tree sap. Then if you need to rework, you do add flux just then. When you do repair like Rossmann, obviously flux should pretty much flow in your veins. Germans have a funny colophony flux brand, "Löthonig". Soldering honey. Can't get over it. Don't like it though, too smokey, too much wood in it. And like why, Chinese flux is OK.
@@SianaGearz if you can do the joint in less than a second, the flux core is good enough. Sometimes. This video helped me a lot ua-cam.com/video/tfIwHuGzUEk/v-deo.html Doesn't matter what brand of flux you use, as you say, the cheap stuff does the job just fine. But it shouldn't have peaks of solder poking out of the joints, the surface tension should be working with you. Of course I wouldn't be surprised if he's just doing it to mess with those of us who have mentioned it before. The German sense of humor is a mighty force to be reckoned with.
@@bur1t0 Yeah that vid is good. But i mean, if you have preheated the joint and applied solder, then you basically don't need to worry, if you trust your solder. Even if you dwell a little too long and it has drawn a whisker, as long as it's not going where it could be dangerous, it's fine, the underlying metal surfaces got fluxed and there will be a functional joint. That being said i do usually flux extra and make my joints look actually neat. But for a proof of concept garbage prototype, putting too much attention to it isn't actually all that time effective. Like you might as well leave it till after you know it works, and after you know it works, you better damn leave the thing alone, why fix what isn't broken. "A German joke is no laughing matter" - Mark Twain.
Awesome SMPS chart. Love the quantitative data charts with graphical (schematic) information condensed well. About all I can see is having some scope graphs to further enhance. :-) Great information and details as always, thanks for sharing!
The terminology Synchronous Converter is misleading as it suggests a new converter type. The design shown is still a Buck converter but with synchronous rectification, which is commonly referred to as a "Synchronous Buck". Synchronous rectification, can in concept, be applied to any use of a diode. Its commonly done on most switching mode power supply topologies or even simple mains frequency bridge rectifiers, although some applications are much easier than others. I'm not sure on the edge rates used, but be aware the long tracks between the MOSFET and the input cap and top MOSFET will eventually cause issues as substantial voltage can be developed across that short length when the switching time is short. These 3 parts should really be place right next to each other. Nice video though, it introduces a very useful concept.
Something very interesting is that synchronous buck converters and synchronous boost converters use exactly the same power components, the only difference is in the control circuits. With the manually controlled circuit you designed you should be able to switch it to a boost converter by putting power into the output pins and putting your load on the input pins. You just need to be careful of the duty cycle.
Wow. Just Wow. The signal on your Silly Scope is fascinating! And it is not what you wanted? Oh, man! It is so colorful! BTW, Merry Christmas and Happy New Year, Everyone! One Day I will get it, Scott. With your help. Thanks!
C’est exactement la topologie des étages d’alimentation des GPU, CPU et RAM avec un transistor MOS côté haut et un côté bas. Hautement efficient et capable de délivrer des dizaines d’ampères avec une chauffe contenue.
Really helpful video, my final year project is based on this synchronous converter this video made me feel good as I was losing hope In my project thanks sir
The MPS 2037 has almost everything you want and is dirt cheap, you can get MPS modules are like 1 dollar each + 1 dollar shipping for me a 3-4A synchronous converter from MPS.
You know, I know Ive looked into using software to control a buck converter because dedicated ICs are just getting so small and its hard to troubleshoot a QFN package. I did it successfully once with a PIC18F but now I wonder if I can "upgrade" the design by using a STM32. So its nice to see that someone else attempted such a circuit successfully :) I know other people have reservations about using software to control a buck loop (what if it fails, etc) but with packages getting smaller and more expensive, using a micro makes more sense to me. Just recently, Ive been looking at designs for a 22-12V IN, 5V out, 5A out (or more) buck converter. Most are out of stock, super expensive ($7+!) or super small. So its nice to have this method as a "backup". Thanks for reigniting my interest a bit in this :)
I got 82% measured efficiency with my blocking oascillator (joule theif)......took me 2 hours to wind the torus...... I used used 22 awg and did like 200-250 turns around the torus in 5 layers of wraps wound bifilar. I'm currently running an LED light string on a bunch of batteried that would otherwise get thrown out. Most people go for low inductance and relatively thin wire. I found that higher inductances make the joule thief more stable and efficient and lower winding reistance helps as well. But I ca currently harvest sources producing voltages as low as 0.5 V and step it up to 95 V and it doesn't need a sepparate voltage source to opperate. It''ll run from the 0.5 V source itself. Just got in an NTE102 transistor as well, it's only rated for 100 mA Ic-e but I should be able to pull current from sources as low as 0.15-0.3 V and step it up to ~12 V. Anyways, nice boost converter.
Only a matter of days ago I watched a different video about making switch mode voltage converters. In that video it was demonstrated that a PCB like this purple one with narrow tracks is not the way to go. Instead use large areas of copper to make all connections.
Let's Get Started and I Will See You NEXT Time! (I so love that; You are awesome, Scott). :-) YES! I am Learning. I just wish you were my teacher, when I was in school...in the 1970's. I really have learned a lot from you. I only watch quality videos about electronics. Also, in my day, even teachers didn't know the future. They could speculate but the technology was just not there yet. Now that I am old, am trying to catch up but...I can make an LED ring light up in a spectacular manner, with arduino and that is cool. I am learning about Mosfets, too. Hard to believe I built all the iRobot B21R and CoWorker Robots. I didn't design them, I just built them (I designed much of the wiring in the B21R). Anyway. Thank You, Scott. You are so good. (And I worked with MANY awesome MIT Grads!)
Small correction; when the inductor current in a syngronus buck converter is negative, it acts as a boost converter, and there are no extra power losses asociated with that! In fact, switching losses go down, becouse the bridge is said to be soft switching. This is when the inductor current helps discharging the parasitic capasitors in the half bridge. So the converter becomes more efficient when the riple current crosses zero ;)
P channel are usually inefficient (and high Gate Charge - Qg). At 80KHz rise time can be a issue depending what mosfet part. A gate driver as simple as totem pole e.g. is definitely a must in many cases.
Great stuff! Ok, so a little challenge for you - a buck converter that takes 3.3...5v and delivers a clean +/-12...+/-15v suitable for audio range op amps. I was very frustrated recently when I found none of the ones I had were remotely suitable for making a split rail, wound up having to use a resistor divider for 0v.
Hi Scott nice video, i think in 4:30 you can use the N channel only in the low side mean connect the S to GND otherwise you need a gate driver. for the same reasons in your video about mosfet drivers...
I like this cute little project :D It's amazing how much you can get from such a simple setup, really. If you're gonna improve for efficiency, note that the input capacitor always takes at least half of the input current, it's probably the most overlooked lossy component here. Its losses are I^2 times its ESR, which might be 100mΩ, far more than the mosfet. In fact, everything here has more resistive losses than the mosfet :D A PCB trace that is 6 times long than is wide, is heating more than the mosfet. Also, a gapped ferrite inductor would be much better. Just a few turns around a ferrite core. It would be interesting to see just how much you could get from this circuit If you'd want to, of course :D I design these things for living and I'd be happy to help :D
hey just read you design these things, i had a question if you dont mind giving your input in the vesc electric vehicle community we are looking for solutions to stepping voltages ranging from 36v to 120v down to 12v or 5v, and we were wanting to design our own purpose built modules/boards. any ideas as to what IC's we could use to base our designs around? thank you!
@@auden_builds well I don't mind, but the "which IC" example depends on the specifics of your needs. If you need something low power (standby supply) and/or very flexible, the NCP1251 flyback controller is a good starting point. If you want something fairly high power and customizable / programmable, I'd use a half bridge driver like IR2184 and a microcontroller with a HRTIM peripheral. If you need insane transient response, there's also FPGA devices with integrated analog front end (ADC/DAC) and also some opamps. Also, if you need something really compact or efficient, consider using GaNFETs. Those are some next level thingies.
Is it maybe an idea to make a windmill out of a car alternator? I played with the idea of feeding the rotor coil with a voltage bonded to the speed of the mill, so the higher the speed, the higher the current. If you then rectify the output and send it to an adjustable mppt converter you should be fine. Set the mppt converter for example to 12v and it will draw more current if the voltage wants to increase. My opinion is that home users should invest in everything renewable but solar(or maybe a little bit). Because too many households have too much solar power Anyways, I like switching converters as much as you do, please continue with producing videos so that linear regulators will loose their status in the future
The same synchronous switching trick can (and often is in high efficiency designs) used in the bridge rectifier portion of AC to DC power supplies to achieve higher efficiency there too. As noted here, such circuits need to be carefully designed to avoid "shoot-through" conditions where you effectively short the incoming power lines together if your switching FETs are on at the wrong times!
Zero-crossing circuit with RC delay, powered with a bootstrap circuit. That gets passed into a set of Schmitt triggers and driver circuit. It essentially becomes a phase-fired dimmer PSU. Alternatively there exists chips that can do that as well, of can be software-controlled using an MCU
Two main problems with these circuits: 1. Extremely large current inrush when circuit is first turned on (this can drain the voltage source which can affect other circuits on same power line or cause fuse to blow, etc., 2) When you turn off the power to the circuit, it can generate voltage spikes on output - e.g. output is 1.3V and then you turn off power and output spikes to 30V if input was at 30V.
you see a significant increase in efficiency especially when you are on low duty cycle, that means that the diode will conduct most of the time, which is the case at your mainboards DC/DC converter which has to go down to 0.9 V or some thing (low duty cycle) whilst providing 30 A. Thats why the buck of your CPU is always a synchronous one. Also you can put synchronous converts in parallel (phase shifted) to pread the load while only controlling one of them. Thats what they also do on your PCs moutherboard. Vendors referr to this as the Motherboards "phases". ( number of parallel synchronous buck stages). Also, did you realize you did not really had to choose wheather you are building a synchronous bock or boost ? Because the schematic is just the same. its just that input and output are flipped. You can actually connect one voltage rail to both sides and depneing on your duty cycle energy will flow from high to low voltage or vice versa! Thats why this topology is often referred as the DC transformer! ;)
That: "Stay creative, and I will see you next time!" goes crazier every time. I love it!
😁
Haha yeah
Keep it, otherwise I'm gonna miss it.
so am i
@@greatscottlabwhy jlcpcb needs $6 to ship and a pcb costs $2
Really interesting project.
I am doing a PhD on power electronics and I have previously studied how to improve buck converters.
There is a topology called "active clamp ZVS buck Converter" that besides using synchronous rectification with mosfets allows to have ZVS in the switching of the mosfets, which greatly reduces the switching losses. In addition, due to its behavior, it allows placing several modules in parallel to increase the total power without additional control, since it distributes the current automatically.
noone cares no lifer
Electronics aren’t real dude
that's awesome! Would be great to read the research you write about this.
Interesting
@@paugasolina5048 its funny you call someone a no lifer when you are on this thread chugging haterade. Must just be self reflecting or something. At least his post relates to the video.
FYI, many STM32 MCUs have a PWM mode with complimentary outputs and dead state insertion -- specifically for switched mode power supplies and brushless DC motors. I didn't really understand the connection until watching this video. Thanks!
Ooooh makes a lot of sense! I always wondered why every timer had 2 output pins
If you and Batu Gunduz meaning the 2 output units for each timer then the ATmega328p has the same scheme.
@@PerchEagle If I understand correctly, the STM32 can generate the complimentary outputs using a single output compare channel (the 328P seems to need two, with the same compare value, but opposite output polarity). Also, the STM32 can insert dead time (neither output enabled) automatically. I think the 328P would need two timers with different TOP/BOTTOM values (or just different initial counter values?) to produce dead time.
@@markday3145 ok, let me check the stm32 datasheet for this point.
Yes, you do. For example, the STM32G474RE available on evaluation kits has a peripheral called HRTIM. A timer capable of generate a pwm with 4GHz clock. It has multiple outputs and features that makes a synchronous buck converter easy to implement. There's also some APIs sponsored by a ST's partner called BIRICHA with professional tools used for design a 2p2z digital controller using this peripheral. Yes, z transforms and transfer functions simulated and measured all in the ecosystem. Have fun.
Sounds like How your GPUs/VGAs' Chipset and VRAM module voltage are supplied. A mosfet driver, Low and High Side Mosfets, A Capacitor, and an inductor/Coil. Interesting!
Oh, you're right!
@@maxhouseman3129 yeah, i saw modern GPUs like 20 series use some kind integrated mosfet-driver because tons of Vcore phases they use and they don't use much space like older GPUs.
Yeh most efficient way to convert power. There are powerstages that reach like 95% efficiency under normal operation.
@@farizfadillah7557 they call those powerstages. Crazy beefy chips. One of them can easily suppy 20-30 amps without overheating and the high end ones rated for very high currents can push 50 amps. If a GPU pulls 300W at roughly 1V you have 300 amps going trough the VRM. The VRMs on modern GPUs are really the most advanced dc to dc converters out there. They convert 12V to something as low as a single volt while retaining 90+ percent efficient and delivering a few hundred watts.
I will say this too. But you already say it
Great scott, sir. I know it's less likely that you read this. But I love how much clearer you explain everything steps/components now.
When I was first year hs, I can't understand a thing in your videos, it's full of maths and very fast. It's just great to look at but nothing to learn from. Only your same level could understand. But since they are your same level, they couldn't also learn anything. Now that I'm getting college that I understand those videos, but still they are very fast and unclear functions of components.
Now you are very much more interactive. Much more clear in explaination. And much longer videos but still direct and great written videos. I actually learn many things now. Thank you so much!
A left handed engineer is a dangerous weapon indeed! Left handed people think differently and are typically more holistic thinkers and creative types. This is a generalization of course, but left handed people tend to be more creative and not have the tendency of falling into STEM fields. So when we see a left handed engineer, it is truly something special because they bring a unique and different approach to problem solving. Again, this is a generalization but there is something to it. I have a friend who is a great programmer and who is also left handed. He always has a different and unique way of approaching problem solving that is different but also efficient and elegant. I really enjoyed this video!
Thanks Scott!
Now I'm 'very glad' I picked a 'Synchronous Converter' for supplying my 150W pure sine inverter with the correct voltage when using two of my EGO (56VDC) batteries ...
Makes for a nice, small, emergency backup and camping power unit. 10AH@56V ... 560WH. 5VDC, 12VDC and 120VAC as needed.🙂
Nice job on making one!
Your videos refreshed my 4 years of Electronic study in 2 hours. Thank you very much
Years ago, a very determined man by the name of Walt Jung was on the rampage to innovate a voltage regulator that could do its job and handle higher currents that’s the typical 0.8 amps that we are familiar with. He wrote some great articles and he produced some schematics that are just as valuable!! I suggest that you would be interested in playing with his designs, or else looking at them if you already haven’t. I believe you would probably take away much more meaning and valuable info than I did. Thanks for sharing again!!
Your calming tone and drawing prowess is Bob Ross lvl. And that is rare.
7:32....oh right, I've seen that setup before in laptop power supplies, but never really understood how they worked without a diode - this makes sense now. They're also susceptible to killing the load if the top FET shorts and sends VCC to the load directly.....
I dont have a more new ideas for you but I can totally say that ur are the most detailed videos I have ever seen. I keep trying to learn from ur videos.. thx for encouraging everyone.
After adding a feedback system. I hope you can go forward and talk about isolated dc to dc converters. The bridge configurations. Zvs. Zcs and phase shifted pwm are very interesting topics and I would love to see what you come up with as a real circuit because I could only get the design due to lack of a workshop of any sort. But yes please a feedback system first. You have a way of making these topics simple to understand to start out
Damn that's some great efficiency numbers I saw today from a DC to Dc converters.
Yesterday in my viva, I was asked what can be the efficiency of a DC to DC Buck Converter and books said about 80%.
I think this can improve power draw of many project that are about portability.
Great Video as always!
Thanks mate :-)
Ok
Efficiency is largely a function of size, load, and conversion factor. If you can make a switching converter larger, you can use a lower loss inductor. If the load is lower than all the magnetic and conduction losses are lower, and if the conversion is a small step than the converter can act more like a wire.
Oh yeah these days you can quite easily build DC converters that operated well above 90%. The part that is hard is when you need to balance other design aspects such as size, cost, EMI, and if you want high efficiency at a specific operating point versus across a wide range of input and output voltage and current range.
What I enjoyed the most, was the fact that you thought off and catered for the "new to electronics audience" thank you for it and akthough this is is a bit afvanced for me at this stage, I will definately tune back into this video.
Sometimes, the AC resistance of the inductor can be a tricky bit, as, at short duty cycles, the apparent frequency content of the inductor current can be high enough to cause several % loss. And, of course, PCB layout can affect performance due to parasitics that reduce switching performance ....and so on, as I'm almost sicher that you know. Nice video. BTW, the ripple frequency and amplitude when in DCM, lets you do some calculations of parasitic capacitance and ACR. "Left as a proof by the student" ;-)
Cheers🍷
Yea
Those are all the fine details. Maybe one day I can get to such detailed topics :-)
About the inductor, I tried some experiments with switch mode chips, and found I could not just wing it with the inductor. I had some ferrite toroids and wound them up. The uH was good but if you have the wrong core material - not so good. So anyone attempting this sort of project, look into the recommended toroid for the circuit.
Dude, how can I get to know so much? Any books you would recommend on power electronics or courses?
I wish you talked about ground bounce too, at least a bit. It's quite an important thing when it comes to making compliant products that have switchmode converters.
Great video though!
I can put the topic on my to do list :-)
@@greatscottlab Please do. Ground bounce is a real pain in the ass - especially in mixed-signal designs.
Also a neat observation you might like, if you look at the generalized/abstract schematics for buck and boost converters (using switch symbols instead of practical implementations, you will see that they are in fact the same. Just the port where power input and output are flipped around. You can make a buck converter, feed the voltage into the output, and use the input of the buck converter as the output to turn it into a boost converter! This does not take into account any practical aspects of the controller or MOSFET drive but it should be possible!
man, i love the drawings that you do. they're so... precise
I recommend using a bootstrap IC driver for your MOSFETs, particularly where you have a high-side FET. I like the IRS2186 but the stupid thing is out of stock because of the silicon shortage. You could make your own but the tricky bit is how to control a high side driver with low-side logic. Internally these ICs might use an optocoupler (at the cost of propagation delay), others use a high voltage emitter follower (also limited in propagation and bandwidth due to miller effects). The fastest ones appear to use a transformer akin to a GDT or capacitive coupling into the inputs of a SR latch on the high side. It might make a good video topic!
A mechanism I'm experimenting with involves a flyback with two feedbacks...
One for the switcher, one for the Ideal Diode - I mean Synchronous Switch
U are the best electronics youtube i have ever seen. learnt alot , i really have an intrest in electronic components, like i open every electronic devices that are not in use and extract pcb. after i complete my high school in 5 months
i am gonna do Electronics and Communication Engineering(ECE), then after than i want to build, invent, create things that can potentially change our lives ;)
I am doing this for my senior design project with more control applications we had a peak efficiency of about 95.5%
This is awesome timing. I am actually building this same circuit this week, except for the controller I am using a UC3843 IC and with either type II compensation network around it for current mode control. I am building this circuit primarily to test my new Picoscope 5444D and it’s frequency response analyzer tool to actually plot the bode plot and measure phase margin.
How to increase the efficiency of most circuits 101:
Replace bipolar transistor with a FET and some complicated logic that emulates the bipolar.
Know the strength and weakness of the different technologies and select accordingly. Easy to say but development can be quite fast so the correct answer might change.
Bipolar transistors might be the best choice sometimes and got it strengths.
Thank you! Really love power supply projects, explains the basics very well.
Glad it was helpful!
In terms of electrical, you are on a whole other level, I was left in the dust.
Great video as always 👍😀
One little trick when soldering SMDs: put the required amount of tin on the pad, and 'cut it off with the solding-tip' and then drive it with solding tip to both the pad and the compont-lead. It is faster this way, and you have more control of the amount of solder on the compont/pad. And do the same on the first pad too. You can adjust the amount of tin, by cutting the length of the tin, or using a thinner tin on SMD😀
Thanks for sharing your great walkthrough of the electronic wonders of the world 👍😀
I'm self taught and when I seen your soldering it just made me smile 😆😁
Nothing special to do on a sunday evening: new video from GreatScott!
❤️
I am looking forward to learn how to add feedback system for this supper simple converter. That would be perfect. 👍
Fingers crossed!
I think a tl431 could be a good and cheap option, but I'm not sure if it will oscillate by itself
@@juanmontes8905 How would you use the reference as a controller? Yeah, if you don’t compensate the Buck it will oscillate depending on the ESR of the inductor and output capacitor.
I'm voting for a software-implemented feedback loop here.
Much more cool. Much more hackable. Much more programmable.
In fact, if he solders just another halfbridge module, he can get a complete pure sinewave DC-AC with just a software mod :D
Or just a multichannel DC supply that can both charge and discharge a battery :)
@@dedamarsovac the Rp2040 can be a good option, but you will need a independent power supply for the uc
I strongly suggest you to get 0.5mm tin. Your smd soldering will look much nicer and you can control the amount much more precisely.
He's way too arrogant to accept comments on how poor is the quality of his solder joints 😆
Good video bro I am from Kerala India 🇮🇳 I waiting for next video 🙏
Thanks :-)
The more I learn about switchmode power supplies, the more they just look like fancy motor drivers - that's clearly a half-H-bridge.
Now if only I could put that knowledge to use 😁
I was just about to build my synchronous buck converter, but after watching your video i think i'd have to make some adjustements like adding some decoupling capacitors and diodes to discharge mosfet's gate capacitor ! Your videos are always GREAT Scott thanks !
Btw adding a proper heatsink will for sure increase the effeciency as it will dessipate heat off of the mosfets !
Your older intro music is still the best, it was timeless
Relay like you explanations and the written diagrams you do that go with them. I'm out of the mechanical space and you make this so clear
Tks Will
Glad to help
I love that GreatScott finally uses ferrules in his projects! Upgrade due to last video :D
You explained this so clearly that even a theory dunderhead like me understood it ! Thank you 👍
He really puts a lot of effort in vids
Thank you :-)
Thanks for explaining! I've done a few integrated buck conversion designs in my circuits, and I'm really only familiar with the diode type topology, that explanation opened my eyes
Glad it was helpful!
I think these converters are used in motherboards to powering cpu's. Great tutorial as always thanks Great Scott
Yes, they're used on that, sometimes even is a single IC with both MOSFETs inside of the same package with the control logic
@@juanmontes8905 thanks for additional information
Biggest deference is that mother boards use Multiphase Bucks. Rather than having to add a ton of output capacitors, they can lower the ripple by using several inductors that take turns charging.
A bad day in electronics nothing happens a good day works as expected, an epic day the smoke comes out of Fred. Great video as always.
I am following this channel for long time but first time I saw the GreatScott!
Scott, this is one of the best videos about DC-DC converters i've seen!!! Such a good explanation with graphs and examples. I Appreciate your great work put in this video! Will be interested in next topic: efficient BOOST converters and are they possible?
FET Gate Driver chip always save the day 😁. Happened to me while designing MPPT based interleaved buck converter
Nice, another video. But im still waiting for The 3d printer diy or buy :D, but keep up The great work!
Maybe one day!
@@greatscottlab I still hope :D
That's easy. Diy if you want the most bang for the buck, buy if you want it to just work. Diy requires a bunch of trial and error to get everything working good. Plus, you have to be extremely comfortable with tearing a lot of stuff apart if you get a clog/jam, etc. That goes for both diy and buy though.
For context, I built a modified version of the HevORT for about $600-$700. There's nothing even close to how good and fast it is if you wanted to buy something at the same price.
@@androiduberalles thats true
@@androiduberalles not exactly the kind of budget 3d printers u can buy theses, days totally defeat the purpose of diy-ing a budget one, but for larger or complex printers diy-ing makes a lot of sense, imo just but a decent printer like the ender 3 and then u can diy a very decent 3d printer cuz u can easily print most of the parts
Your explanations are getting even better! I am amazed
Very good information on the synchronous rectifier part, it helped me to understand how it works a little bit better.
Thanks for sharing.
the Purple and Green color scheme obviously makes it better out of the gate.
Everything else is just gravy
I only recently learned this from UA-cam, but your soldering exploits would be made much easier with a generous application of flux. I've been using some colophony I got on ebay from the Ukraine, works a treat and leaves my room smelling forest fresh! But seriously, the Louis Rossman school of "Would you like some solder with your flux" cannot be understated. Solder becomes so much more cooperative. There is no such thing as too much, but there is certainly not enough.
I don't think he's having a hard time though. If you buy actually high quality solder it's got at least 2% flux in it anyway, and if you work hot and quick it easily suffices for the initial joint, easily! And it's good flux too, not just dirty tree sap. Then if you need to rework, you do add flux just then. When you do repair like Rossmann, obviously flux should pretty much flow in your veins.
Germans have a funny colophony flux brand, "Löthonig". Soldering honey. Can't get over it. Don't like it though, too smokey, too much wood in it. And like why, Chinese flux is OK.
@@SianaGearz if you can do the joint in less than a second, the flux core is good enough. Sometimes. This video helped me a lot ua-cam.com/video/tfIwHuGzUEk/v-deo.html
Doesn't matter what brand of flux you use, as you say, the cheap stuff does the job just fine. But it shouldn't have peaks of solder poking out of the joints, the surface tension should be working with you. Of course I wouldn't be surprised if he's just doing it to mess with those of us who have mentioned it before. The German sense of humor is a mighty force to be reckoned with.
@@bur1t0 Yeah that vid is good. But i mean, if you have preheated the joint and applied solder, then you basically don't need to worry, if you trust your solder. Even if you dwell a little too long and it has drawn a whisker, as long as it's not going where it could be dangerous, it's fine, the underlying metal surfaces got fluxed and there will be a functional joint. That being said i do usually flux extra and make my joints look actually neat. But for a proof of concept garbage prototype, putting too much attention to it isn't actually all that time effective. Like you might as well leave it till after you know it works, and after you know it works, you better damn leave the thing alone, why fix what isn't broken.
"A German joke is no laughing matter" - Mark Twain.
Awesome SMPS chart. Love the quantitative data charts with graphical (schematic) information condensed well. About all I can see is having some scope graphs to further enhance. :-) Great information and details as always, thanks for sharing!
Neglect-able = Negligible?
Glad you enjoyed it!
Where can we download the chart?
www.kynixsemiconductor.com/upload/image/20180523/SMPSChart_20180523.pdf
Those purple PCBs are always so pretty
The terminology Synchronous Converter is misleading as it suggests a new converter type. The design shown is still a Buck converter but with synchronous rectification, which is commonly referred to as a "Synchronous Buck". Synchronous rectification, can in concept, be applied to any use of a diode. Its commonly done on most switching mode power supply topologies or even simple mains frequency bridge rectifiers, although some applications are much easier than others.
I'm not sure on the edge rates used, but be aware the long tracks between the MOSFET and the input cap and top MOSFET will eventually cause issues as substantial voltage can be developed across that short length when the switching time is short. These 3 parts should really be place right next to each other.
Nice video though, it introduces a very useful concept.
It'd be interesting to see a basic buck converter topology with an "ideal diode" (MOSFET + Current Mirror) in place of the normal diode.
Ya just took electronics in bifocal in 11th and your videos help me in the electronics
Something very interesting is that synchronous buck converters and synchronous boost converters use exactly the same power components, the only difference is in the control circuits. With the manually controlled circuit you designed you should be able to switch it to a boost converter by putting power into the output pins and putting your load on the input pins. You just need to be careful of the duty cycle.
interesting, did not notice that on my own before :)
Wow. Just Wow. The signal on your Silly Scope is fascinating! And it is not what you wanted? Oh, man! It is so colorful! BTW, Merry Christmas and Happy New Year, Everyone! One Day I will get it, Scott. With your help. Thanks!
Great video as always. I’d love to see you explain an isolated dc-dc topology.
Ok i get :D Great Scott! is what Doc always says in "Back to the Future" ...nice ! :D
Great series! I would really love to see how to implement an isolated switching converter. The magnetics can be a bit intimidating.
Very nice tutorial about synchronous vs asynchronous switching power supplies.
Glad you liked it!
If only JLC had a matt green solder mask it would be my number one shop....
I always forget that MOSFETs can replace diodes. Good video, as always.
Next stop is LLC! Thank you for the video!
Yes, we'd like to see him suffer.
Thanks for watching!
Thanks from Texas Scott.
C’est exactement la topologie des étages d’alimentation des GPU, CPU et RAM avec un transistor MOS côté haut et un côté bas. Hautement efficient et capable de délivrer des dizaines d’ampères avec une chauffe contenue.
Just be careful that your "1" is not confused for a "7". (Shorten the top of your 1 or put a line under it.) :-) @6:40
Great video and interesting thanks for sharing. The black art of switch mode power supplies. Back in the day we had a lot of small explosions.
„everyone can have a bad day“ that’s true mine was today and blue up a condensator in my hand on my new amplifier
Really helpful video, my final year project is based on this synchronous converter this video made me feel good as I was losing hope In my project thanks sir
Great little example video of how a buck converter works
The MPS 2037 has almost everything you want and is dirt cheap, you can get MPS modules are like 1 dollar each + 1 dollar shipping for me a 3-4A synchronous converter from MPS.
Thanks for the tip :-)
@@greatscottlab sorry, I meant 2307
Great video,Scott.hello from Belarus
You know, I know Ive looked into using software to control a buck converter because dedicated ICs are just getting so small and its hard to troubleshoot a QFN package. I did it successfully once with a PIC18F but now I wonder if I can "upgrade" the design by using a STM32. So its nice to see that someone else attempted such a circuit successfully :) I know other people have reservations about using software to control a buck loop (what if it fails, etc) but with packages getting smaller and more expensive, using a micro makes more sense to me.
Just recently, Ive been looking at designs for a 22-12V IN, 5V out, 5A out (or more) buck converter. Most are out of stock, super expensive ($7+!) or super small. So its nice to have this method as a "backup".
Thanks for reigniting my interest a bit in this :)
Good circuit and teaching of synchronous buck converter.
I got 82% measured efficiency with my blocking oascillator (joule theif)......took me 2 hours to wind the torus...... I used used 22 awg and did like 200-250 turns around the torus in 5 layers of wraps wound bifilar. I'm currently running an LED light string on a bunch of batteried that would otherwise get thrown out.
Most people go for low inductance and relatively thin wire. I found that higher inductances make the joule thief more stable and efficient and lower winding reistance helps as well. But I ca currently harvest sources producing voltages as low as 0.5 V and step it up to 95 V and it doesn't need a sepparate voltage source to opperate. It''ll run from the 0.5 V source itself.
Just got in an NTE102 transistor as well, it's only rated for 100 mA Ic-e but I should be able to pull current from sources as low as 0.15-0.3 V and step it up to ~12 V.
Anyways, nice boost converter.
Only a matter of days ago I watched a different video about making switch mode voltage converters. In that video it was demonstrated that a PCB like this purple one with narrow tracks is not the way to go. Instead use large areas of copper to make all connections.
Let's Get Started and I Will See You NEXT Time! (I so love that; You are awesome, Scott). :-)
YES! I am Learning. I just wish you were my teacher, when I was in school...in the 1970's.
I really have learned a lot from you. I only watch quality videos about electronics. Also, in my day, even teachers didn't know the future. They could speculate but the technology was just not there yet. Now that I am old, am trying to catch up but...I can make an LED ring light up in a spectacular manner, with arduino and that is cool. I am learning about Mosfets, too. Hard to believe I built all the iRobot B21R and CoWorker Robots. I didn't design them, I just built them (I designed much of the wiring in the B21R). Anyway. Thank You, Scott. You are so good. (And I worked with MANY awesome MIT Grads!)
Small correction; when the inductor current in a syngronus buck converter is negative, it acts as a boost converter, and there are no extra power losses asociated with that! In fact, switching losses go down, becouse the bridge is said to be soft switching. This is when the inductor current helps discharging the parasitic capasitors in the half bridge. So the converter becomes more efficient when the riple current crosses zero ;)
There is extra power loss from the extra circulating current in the inductor's DC resistance.
Oh. PS. Love the visual (drawings) on the schematics. You are a great (Scott) teacher..uh, Scott. ;-)
I was blown away learning (Vishay?) SiC series of Buck converter controllers with integrated switches easily have 99% efficiency.
Always looking for higher efficiency. This video is great, Scott.
Glad it was helpful!
P channel are usually inefficient (and high Gate Charge - Qg). At 80KHz rise time can be a issue depending what mosfet part. A gate driver as simple as totem pole e.g. is definitely a must in many cases.
Great stuff! Ok, so a little challenge for you - a buck converter that takes 3.3...5v and delivers a clean +/-12...+/-15v suitable for audio range op amps. I was very frustrated recently when I found none of the ones I had were remotely suitable for making a split rail, wound up having to use a resistor divider for 0v.
awesome video. I was building a boost converter a few days ago and actually wondered how to avoid the power loss from the diode.
Hi Scott nice video, i think in 4:30 you can use the N channel only in the low side mean connect the S to GND otherwise you need a gate driver. for the same reasons in your video about mosfet drivers...
I always imagined great scott to be a 50 yr old retiree
Great project! I feel like someone need to buy you a breadboard 😉
I like this cute little project :D
It's amazing how much you can get from such a simple setup, really.
If you're gonna improve for efficiency, note that the input capacitor always takes at least half of the input current, it's probably the most overlooked lossy component here. Its losses are I^2 times its ESR, which might be 100mΩ, far more than the mosfet.
In fact, everything here has more resistive losses than the mosfet :D
A PCB trace that is 6 times long than is wide, is heating more than the mosfet.
Also, a gapped ferrite inductor would be much better. Just a few turns around a ferrite core.
It would be interesting to see just how much you could get from this circuit
If you'd want to, of course :D
I design these things for living and I'd be happy to help :D
hey just read you design these things, i had a question if you dont mind giving your input
in the vesc electric vehicle community we are looking for solutions to stepping voltages ranging from 36v to 120v down to 12v or 5v, and we were wanting to design our own purpose built modules/boards. any ideas as to what IC's we could use to base our designs around?
thank you!
@@auden_builds well I don't mind, but the "which IC" example depends on the specifics of your needs.
If you need something low power (standby supply) and/or very flexible, the NCP1251 flyback controller is a good starting point.
If you want something fairly high power and customizable / programmable, I'd use a half bridge driver like IR2184 and a microcontroller with a HRTIM peripheral.
If you need insane transient response, there's also FPGA devices with integrated analog front end (ADC/DAC) and also some opamps.
Also, if you need something really compact or efficient, consider using GaNFETs. Those are some next level thingies.
I love all your videos. Thanks so much for doing these!
Next video, synchronised buck boost converter. Love from India
Synchronous ones also have very low dropout voltage.
4:40 always 🥲
Cellulose CAD, my 2nd favorite.
Is it maybe an idea to make a windmill out of a car alternator? I played with the idea of feeding the rotor coil with a voltage bonded to the speed of the mill, so the higher the speed, the higher the current. If you then rectify the output and send it to an adjustable mppt converter you should be fine. Set the mppt converter for example to 12v and it will draw more current if the voltage wants to increase.
My opinion is that home users should invest in everything renewable but solar(or maybe a little bit). Because too many households have too much solar power
Anyways, I like switching converters as much as you do, please continue with producing videos so that linear regulators will loose their status in the future
The same synchronous switching trick can (and often is in high efficiency designs) used in the bridge rectifier portion of AC to DC power supplies to achieve higher efficiency there too. As noted here, such circuits need to be carefully designed to avoid "shoot-through" conditions where you effectively short the incoming power lines together if your switching FETs are on at the wrong times!
Zero-crossing circuit with RC delay, powered with a bootstrap circuit. That gets passed into a set of Schmitt triggers and driver circuit. It essentially becomes a phase-fired dimmer PSU.
Alternatively there exists chips that can do that as well, of can be software-controlled using an MCU
1:00 "But before I do that..."
Ngl, I was going to fast forward already XD
Two main problems with these circuits: 1. Extremely large current inrush when circuit is first turned on (this can drain the voltage source which can affect other circuits on same power line or cause fuse to blow, etc., 2) When you turn off the power to the circuit, it can generate voltage spikes on output - e.g. output is 1.3V and then you turn off power and output spikes to 30V if input was at 30V.
you see a significant increase in efficiency especially when you are on low duty cycle, that means that the diode will conduct most of the time, which is the case at your mainboards DC/DC converter which has to go down to 0.9 V or some thing (low duty cycle) whilst providing 30 A.
Thats why the buck of your CPU is always a synchronous one.
Also you can put synchronous converts in parallel (phase shifted) to pread the load while only controlling one of them.
Thats what they also do on your PCs moutherboard. Vendors referr to this as the Motherboards "phases". ( number of parallel synchronous buck stages).
Also, did you realize you did not really had to choose wheather you are building a synchronous bock or boost ?
Because the schematic is just the same. its just that input and output are flipped.
You can actually connect one voltage rail to both sides and depneing on your duty cycle energy will flow from high to low voltage or vice versa!
Thats why this topology is often referred as the DC transformer! ;)
Thanks sir, you made vedio very close to my request but please 1 time make a vedio on how to make highest efficiency diy lab bench power supply