EEVblog #225 - Lab Power Supply Design Part 4 - PWM Control

[maqsoodu23]

Dave, I have seen many of your tutorials/reviews . They have been really good. This PWM stuff however, blew my mind away! This was seriously impressive :O

I am going to build this PWM set up and test it out. Any recommendations on the op amp?

[gabrielfinol]

Love it!! you are awesome!!



[Microman171]

Playing along at home. Using a TL072 and a N-MOSFET, I made up a linear regulator in place of the LT3080. Then I did a little bit of math. Since the whole system is a series circuit, the regulator is a resistor to drop the voltage. Also since it's a series circuit, the current is the same everywhere. I set the supply for 1v, and used a 1 ohm load. Using a 30v supply, I needed to drop 29v at 1A. So I was wasting 29W of heat to supply 1W to the load..! That's about 3.5% efficient..!

[EEVblog]

@Microman171 Yes, that's how all linear PSU work.

[quicknuclearscience]

do you have any photos or schematics of this on your websites

[EEVblog]

@quicknuclearscience Not yet, coming soon...

[BadgerInverted]

nice set of videos! what would you recommend using to power your power supply? ;)

[SparkysWidgets]

Finally had a chance to catch up on the power supply vids! Keep up the good work, love learning from them!

[anonimoan]

Shouldn't the resolution be Vref / ((2^n)-1)?

[Goncalgood]

Can't wait for the rest! :D

[tejonBiker]

unitary KRC low-pass second order filters, that was my grave in my school last cycle. What progam you use to simulate?

[lyrez62]

I like it !

[chupascz]

4) use digital potentiometer like MAX5481 (1024 step) - SPI

[ultimatefang007]

I have a question. At 36:40 you turned the yellow triangular wave in to a continuous one. What did you do there?

[alexanderet98]

@ultimatefang007

Hi moved the proabe from the rc filtered PWM-pulse to the outputpin of the constant power source. Hi also explains the may be unpredicted flat curve on the output as a resoult of the capacitance in the powersupplys output...

[ultimatefang007]

@alexanderet98 Thanks man. I realized that too later on.

[SajjadBro]

Hi dave i am new to your EEVblog website. Don't know how to ask you something there so asking here. I have found a free Book on website "Op-Amp for Everyone". Is it good to follow? and if not please suggest a good one. Do reply.

[EEVblog]

@SajjadBro Haven't seen it, sorry. Too many books, I can only recommend The Art of Electronics.

[Dimlowuk]

Can't wait for the next one. great work

[jpelczar]

@EEVBlog: remember to use the snake oil on the high quality pot ;)

[jcndizo]

the video goes blank at 30 minutes and 55 seconds.

[Gameboygenius]

I'm noticing some wobble on the scoped signal when you added the second filter stage. Mains interference due to increased impedance?

[Marzec309]

Thanks for sharing your wealth of knowledge with us :) I've been in need for a CV supply for a while now and this will do nicely. I've drawn up a PCB layout with a tweak or two of my own. If interested I will share it with you. They are expresspcb files. I thought it turned out nice.

[SeanB88]

Very cool Dave!! Can't wait for the rest of it. In terms of Op-Amps, you mentioned needing a rail-to-rail amp...I've recently sampled the MAX4040 and MAX4043 (single and quad versions) and they are really nice little amps that go really really close to the rails. The only draw back in the maximum supply being 6volts. I thought of them for your project however, since you are only dealing with 5volts or less in the MCU portion of the PSU.

[EEVblog]

@SeanB88 In this exact case, only 0V input and output is required, not true rail-to-rail. There are countless opamps that'll do the job, buying almost based on price and availability.

[mathtek1]

Amazing stuff. The impact from these series will reverberate throughout the EE community for many years to come.

[slateraptor]

22:01 LC...oops. =)

[EEVblog]

@slateraptor Those Oops moments happen fairly often. My brain did know what it was thinking :->

[Kankki1]

Really informative as always!

Although feels like you could just make it a lot simpler by just spending 4 bucks on a couple DACs :D R2R ladder perhaps a cheap way out? Requires a ton of IO though..

[laharl2k]

@Kankki1 the good thing about going R-2R ladder would be that you could go have as much precision as IO pins your micro has, for example a atmega8 hase 22 io pins(23 if you use the reset one) say you use 2 for up/down, 2 for select/cancel, and 1 or 2 for the the lcd.

that lives you with 17/16 pins worth of resolution, aka 0,00076%/0,0015% and you wouldn't need a RC filter, hence wouldn't have extra noise, and with 17bits you could go high voltage, say 100V and still have 0,76mV steps!

[Kankki1]

@laharl2k Yep, although you'd need to halve it if you want current control :)

[marcan42]

@laharl2k R-2R ladders depend greatly on the matching between resistors to achieve smooth output. With 1% resistors you can expect to get 5 to 7 bits before the output inaccuracies overwhelm your precision. The output response becomes "jaggy" and actually steps backwards for some (many) steps, depending on how many bits change.

To get your 0.0015% output resolution (not just accuracy), you'd also need 0.0015% tolerance resistors (you'd probably need better than 0.0005% in practice).

[laharl2k]

@marcan42 silly me, totally forgot about that, and dave just explained it some videos ago XD

though i didn't know those details about the accuracy beyond 7 bits, ill have to investigate a little further about that :-P

[EEVblog]

@Kankki1 Even an R2R is more expensive, and more board space than a PWM solution.

[SeanBZA]

@EEVblog Cheaper than running the whole micro on a precision source is to just run a 4050 or 4049 with a single stage driving the other 5 in parallel. Cheap and only the PWM load on the regulator, which generally is pretty poor in current regulation. Isolates the noisy micro from the analogue, just use the 4049/50 ground to join digital and analogue grounds. Will give a nice square output with good and consistent rise and fall times to the filter, and is dirt cheap.

[Spoif]

@EEVblog Interesting video and a bit complicated. I'd tend to go with the R2R network also despite IO pin consumption.

[EEVblog]

@Spoif You can't get easily or cheaply get even 10bit resolution with R2R. The resistor tolerances will kill it.

[ntjbjhq]

Wow Dave, this is an amazing series of videos. I don't even want to think about the time commitment here on your side, but WOW. THIS is the exact type of EE videos missing from the web. No one else dives this deep combining so many important aspects on anything. A simple project from the start, morphing into some serious experience sharing and teaching. This stuff is gold.

You are going to spoil me, I'll never be able to watch another "what is a linear reg" type video again!!

THANK YOU!

[netcore2k]

I really like these tutorial like videos. Very informative. I'd definitely like to see more like this.

[DanFrederiksen]

where is part 5 and 6? I can't see them anywhere : )

[DanFrederiksen]

good videos.

I wonder what pros use to edit video. like the guy that does the editing for ray william johnson's show. I can't believe they use amateur junk like premiere for instance. it has to be a very fast and smooth process to do serious work.

[xasdrubalex]

O HI BODE PLOT.

[Immersitech]

Dave, This series is terrific ... as others have said I am also learning a lot and very much enjoy seeing the design through prototyping. Can't wait for future installments and actual build!

[SeanBZA]

Buffer is because the regulator is a current source, and the filter works better with a high impedance load, preferably a very high value resistor to ground. If direct connection there will be an offset that makes the output voltage rise above the set value, and this will vary according to the voltage set. The buffer sinks the current whilst providing a very high load to the filter. RC filters will always have ripple on the output, cascading them just gets it below the noise floor.

[TheLaidukas]

Next time, please do not use blue color, on my screen I can barely see it, or some times don't see it at all. Thanks

[error079]

Yes. Blue and Red high contrast does not compress well so please stay away from those if possible

[WhiterockFTP]

@EEVBlog Can u explain me what a buffer does :)?

[colt4547]

Very nice video series. I really enjoyed the the first three (all of your videos really) videos on linear supply design but this video has taught me alot. I thought I understood filters pretty good from your other vids but this one has set me straight. Its a good thing im not your neighbor for your sake. you would probably have to get a restraining order to make me leave. Thanks Dave for all of your effort and time in making these videos. People like you sharing knowledge in an entertaining man

[lawrencelee92]

Watching this video brings back memories of filter design in one of my electrical engineering units at uni.

[SajjadBro]

Dave i wish i could have an electronics Teacher like you.

[frac]

Great video (even with the Sony Vegas censorship). The video returns after a few minutes, so you don't miss too much.

[hla27b]

Go Dave!!!

[kiyotewolf]

Interesting how it took two stages of RC constant low band filter to get the PWM level out, with enough filtering. EEVBlog class!

[srhenderson186]

Hmm... this vid makes me want to change my power / energy saver settings. I kept wiggling my mouse... hitting the shift keys... maybe it's a full-screen mode bug... sigh.

[Zwank36]

30:53 , no video?

[superdau]

Ahhh.... I live in the wronf timzone. It's almost four in the morning here and there's a 40 minute EEVblog again. But .... can't ... resist.... (at least this time it's friday ;) )

[EEVblog]

@shiftplusone80 Thanks. Sony vegas screwed up again...

[PesimisTosbaa]

Why dont you use an active filter?

[EEVblog]

@PesimisTosbaa No need for an active filter.

[electrodacus]

:) 30:55

[EEVblog]

@electrodacus DAMN sony vegas. Will have to re-render and re-upload :-(

[electrodacus]

@EEVblog I do not think that are that many seconds to worth re-uploading.

[JumperOneTV]

@EEVblog Have you tried Adobe Premiere? If not, they've got a free trial version for download. IMO much better than Vegas!

[abpccpba]

Thanks; well done you are on a roll. That LT Spice is nice software.

[Afrotechmods]

I love the to be continued

[gustavovalsan]

@Afrotechmods Afrotechmods sighting =)

[Jprojects1]

Wow, even though you are updating these so often, it is so hard to wait for the next one. I am learning a ton from you. Thanks.

[Fatkuh]

I really like it!

[farbe123]

Nice

[StoneFlange]

I'm the first to watch? :D

[GADELHAS82]

Great!!!