EEVblog #222 - Lab Power Supply Design - Part 2

[Nadrealis]

What type of package did you use for the LT 3080 in your final design?

[EEVblog]

@Nadrealis TO-220-5

[pcbdev]

I like it that your explanations are not for dummy-beginners who never had to do anything with electronics. This level is perfect :)

[PetrFM]

This design is becoming more and more complicated.

Simple charge pump like ICL 7660 and you have negative rail for all opamps. This give you the opportunity to use cheap LM317 have real zero voltage output etc.

Noise is not such problem as opapms usually have high supply voltage ripple rejection ratio ...

[EEVblog]

@PetrFM The LM317 neg supply would actually make it more complicated. Not only the 7660, but the output divider will have a varying current based on output voltage, and that throws out your current sensing at low values. The LT3080 and LM334 is simpler IMO.

[MatrixOfDynamism]

@EEVblog Once we have learnt simple stuff like linear circuit design (RLC), and stuff about transistors and other simle components and OpAmp configurations, how do we move onto being able to use such exotic ICs? ICs with datasheets that have so many words that look so unfamiliar? I really want to know that. There are soo many ICs how do we know which one is best. since our criteria only defines some of the things, and there can be soo many ICs that fit that criteria.

[Marzec309]

What rail to rail op-amp are you planning to use?

[CampKohler]

Dave, are you going to enter Agilent's PS contest and try to win that juicy USD11K digital supply?

[EEVblog]

@CampKohler Never heard of it...

[spelunkerd]

As a relative beginner in this field, I've listed these two videos as favourites, for when I have more time to learn the details. What would be nice would be a practical view of the box and circuit board, so I can have some idea of what I need to put it all together. This is definitely one of those projects were effort in will reward the student later.

Thanks for going through this. Is there a part III?

[EEVblog]

@spelunkerd Yes, Part 3 will be testing the circuit on the breadboard to see how it performs. The real build will come later, I have an idea in mind, and will do another Part 4 when that is complete (that will take some time).

[Nickafyit]

Excellent! Great job explaining everything, Dave ! You are truly a very good EE.

I am learning quite a bit.

[DanFrederiksen]

I meant IRGP4069

[heroineworshipper]

7:50 The Vset resistor has to be reduced if it's shared like that. Real trap for young players.

[compaq1275]

Is the circuit good to build as shown in the last few seconds of the video Dave? Or are you going to pull another surprise in the next video :P I wish you would do more of these tutorial videos by the way. You're a great engineer and you do a great job teaching.

[EEVblog]

@compaq1275 No surprises, I'll build it as shown. Although the final built product (as opposed to breadboard) will likely use a current sense amp instead of the opamp based diff amp.

[fiatluxstop]

Very nice design. I really like these design to execution projects, and your explanations of the traps for young players really help out for those of us that don't know about the foolishness of the datasheets. But I have to know, do you have a Clint Eastwood shirt for Lab Power Supply part III? :-D

[techgood]

what can i say, as always wonderful video, good teaching , thank you very much Sir - I am thinking of building a 0-30v 0-30A power supply, should i build or buy?

[EEVblog]

@techgood Building is more fun, but 0-30V supplies are dirt cheap these days. If buying at least replace the controls with good quality 10 turn pots. Why do you need 30A in a general bench supply? 30V/30V is 900W!

[PoirierMike]

I like it. I'm learning a lot.

It's fun and useful to take time to design.

More of these type of show, Dave!

[tommy9411]

these kind of videos are awesome dave

[sabamacx]

At 35:26, the two resistors in the amplifier feedback loop "can be anything". Why not set them to something high, say 1M Ohm, to keep current low?

[EEVblog]

@sabamacx You can, but not arbitrarily high like 1M. The discharge resistor can't be that high otherwise it'd take forever for the constant current to kick in.

[sabamacx]

Magnificent video Dave!

[TheCrazyInventor]

Very good explanation. Nice job. I hope you'll do a video on designing a PC for the supply as well. I'd love to hear about the PCB design considerations.

[EEVblog]

@TheCrazyInventor Yep, there will eventually be a PCB for it.

[DanFrederiksen]

interesting videos. how about a follow up where you do a 30V 10A switchmode supply. enter the 21st century and such. linear is the dog's breakfast : )

the idea that switchmode can't be smooth is an old wives' tale.

maybe two stage unisolated with IGRP4069 IGBT on the first stage and IRFP4368 super efficient FET for extreme coolness. maybe even do something ugly like only chop the half sine so there is only one voltage drop so it's super efficient and cheap. unless it's against your religion : )

[TheMcMonster]

I agree, there are tons of "normal", boring power supply designs and I'd be happy to see a video about non-standard or very high power supplies, say something based on 2N3055 BJTs.

[izimsi]

@TheMcMonster Aren't modern 2n3055 100% fake and having shitty specs, making them unusable?

[EEVblog]

@TheMcMonster I have little interest in very high power supplies.

[TheMcMonster]

@EEVblog

Ok, then maybe a video about symmetrical or negative voltage power supplies somewhere in the future?

But thanks anyway, I love your "lets design a..." videos and this one is great, I'm going to build this myself.

[EEVblog]

@TheMcMonster I don't know, maybe. There is a specific reason why I'm building this "simple" design, and you'll see why in the future (not the next video though). And AFAIK there aren't that many PSU circuits out there that show you how to control traditional voltage and current regs simply from a micro with PWM, hence this design.

[ivanblogs]

@DanFrederiksen I'm not sure Dave's YouTube subscribers are the target market for such a circuit walk-thru.

[hla27b]

I think I'm gonna build one myself as I watch the next one. Thanks Dave.

[Kankki1]

One thing I wonder, why does the offset voltage become worse, because of a larger die?

Anyway really great video, as always!

[EEVblog]

@Kankki1 Good question, and one for the IC process engineers out there.

[vk6hdx]

Lets build it up! Great series dave. Can't wait for the next video.

[GeorgeGraves]

I'm afraid of datasheeets. So I had to force myself to watch this. I'm really glad I did. Thanks for taking the time to walk us through the data sheet.  I fell like I learned a lot just from that!

[williefleete]

i got some quad LM324 op amps. would those work ok for this sort of thing. cause i might give this a go, my current power supply uses a regulator made from a 741 a series pass and some diodes for reference , crap regulation but it lasted longer than the 5 amp LM338's cause the transistor is pretty beefy. in fact the transformer blew out before the transistor in one case

[EEVblog]

@williefleete Sure, but likely not great at the low end on the diff amp due to the offset voltage. You could trim errors out though with pots, but that's rather medieval. I'll be using a slightly better TLC2272 in the build.

[milos1993milos]

14:30 would be better to use resistor ladder D/A converter instead of PWM one ?

its cheap, and if you dont have I/O you can always use port expander

[EEVblog]

@milos1993milos PWM is simpler though. You need a 10bit one to get 10mV resolution in 10V output.

[humblesnurp]

Good work!

[First2ner]

man I love this long videos