General Purpose Power Supply Design

[Teemo]

Hello,
Here is some good reading about the stability of linear regulators: http://sva.ti.com/assets/en/appnotes/f10.pdf
Regards.

[Thor-Arne]

Great reading. Thanks. 

[TheRevva]

I definitely agree that it's a very well written text.
(My only problem is that it arrived a few days too late for me.  I finally managed to get my head around the phase margin issues a few days ago.  Murphy was an optimist!!!).

I'm now MORE certain than ever that my decision to go with an NPN regulator is the RIGHT one for me.
I want as little as possible in the way of bulk capacitance on the output, and this is almost impossible with an LDO without incurring stability issues.
When coupled with the fact that my supply will experience HUGE differences in the output load (ranging from mA to several Amps), it makes my goal of unconditional stability harder to realise (with an LDO)!
The additional input-to-output losses I'll encounter by rejecting use of an LDO design are largely insignificant to me.  I'll simply make sure the pre-regulator maintains the input supply voltage far enough beyond the desired output voltage.  The NPNseries pass control current (i.e. the current injected into the base of the series pass transistor) is dissipated by the external load rather than having to be internally dissipated.
I totally understand the logic of an LDO regulator, but in MY circumstances, the classic NPN regulator certainly appears to be a better choice.  While the PFET style LDO described later in the document alleviates _some_ of my issues with the general LDO design, I still hold that the NPN style is a 'better fit' for me.

My primary remaining obstacle is to better understand how well the incoming switching noise ('hash') from the SMPS pre-regulator will be suppressed by the linear regulator.  Obviously, I'll include HEAPS of bulk capacitance on the pre-regulator output spread across several capacitors (from several thousand uF down to mere nF), but I know there will still be SOME amount of this 40kHz hash that will still find it's way to the linear regulator input.  I've yet to determine how I should quantify the source noise rejection of the linear section.  (In other words, I have to perform some more 'LTSpice' simulations)

[Jebnor]

Thank you all.  This is fantastic thus far. (I'm on 22March 2012)  I'll be by with something to really contribute when I get to the end.

[Jebnor]

A fantastic day of reading and researching was had by me. 

Thank you all!   

Has amspice posted a 'final' version yet?

[Kirigozo]

Hello,
Here is some good reading about the stability of linear regulators: http://sva.ti.com/assets/en/appnotes/f10.pdf
Regards.

Hi folks,

I've tried to get the app note mentioned in the quote but that domain name is simply not responding to pings even. I've searched TI.com and had no luck either. The actual document title might help me locate it.

TIA

[Thor-Arne]

Here you go...

Document title is "A user's guide to compensating Low-dropout regulators".

[Kirigozo]

Here you go...

Document title is "A user's guide to compensating Low-dropout regulators".

Been looking for something like that for ages!

Thank you very much.
  

[lunemec]

I am working on an update, and I hope to finally be able to build a prototype this weekend. Currently I have a lot of my parts in storage boxes, so I just have to sort that out first. There are some changes needed, and you are right, at very least you have to change the 1N4148 for a schottky. There is also a change needed to th reference input circuit  - A 1N4148 has to be added to a cap on the voltage reference input to the output to drag the reference down when the output voltage is low.

Several other changes, but I will build up some actual hardware first before putting up new circuits.

Hello amspire, I'm looking to build this power supply (analog version) because of problems with LT3080 (https://www.eevblog.com/forum/projects/lt3080-wierdness-dave's-power-supply-(eev224)-gone-mad/msg283556/) I'd like alternatives. Your power supply looks very promissing, but there has been no update for 2 years ...

[rolycat]

Hello amspire, I'm looking to build this power supply (analog version) because of problems with LT3080 (https://www.eevblog.com/forum/projects/lt3080-wierdness-dave's-power-supply-(eev224)-gone-mad/msg283556/) I'd like alternatives. Your power supply looks very promissing, but there has been no update for 2 years ...

You are unlikely to get a reply from amspire, since regrettably he has not visited the forum for over 2 years. This may explain the lack of updates...

[Thor-Arne]

The regulator is basically all there, only the pre-regulator is missing, some tweaking of the feedback caps might be required on the final design.

I played around with it a little, but it stalled since I have enough psu's now.

[commie]

Hello amspire, I'm looking to build this power supply (analog version) because of problems with LT3080 (https://www.eevblog.com/forum/projects/lt3080-wierdness-dave's-power-supply-(eev224)-gone-mad/msg283556/) I'd like alternatives. Your power supply looks very promissing, but there has been no update for 2 years ...

There are some things that you make for yourself and there are some which you buy. Of those you buy, it's an oscilloscope and a power supply plus other stuff. The reason for this is simply cost, performance,guarantee its going to be fully functional and ease of acquisition cannot be beaten.

[lunemec]

Thank you guys for the response.

I'd like to ask if someone has LTSpice simulation of this circuit (ideally with the pre-regulator). I'm interrested if someone actually built this scheme.
I have spare 30W laptop charger (19V, 2.9A) that I'd like to use with this circuit. Kind of mini-lab psu.

Did anyone build this? Could you please attach LTspice file (maybe even board design?)

I'm looking for any PSU schematic, but this looked the most promising ... know of any other good ones (without LT3080 as it seems to have some issues)?

[Thor-Arne]

I've posted the Mk.3 spice files further up in this thread, as far as I know this is the latest ones posted.

[lunemec]

Thor-ane, thank you for reponse, I tried your LTSpice schematic, and it works as expected, but I was looking for something complete - with all the subcircuits as a schematic, not just blocks like in the mk3 scheme....

I looked around the internet for some complete scheme and found this: http://www.electronics-lab.com/project/0-30-vdc-stabilized-power-supply-with-current-control-0-002-3-a/

I put it into LTSpice, but I'm not sure if I messed up the schematic, but there is 0.1V on the output all the time...

What do you think about this schematic, is it any good?
Any ideas about the LTSpice scheme (attached)?

[sorin]

What do you think about this schematic, is it any good?

I'ts not good.

[Thor-Arne]

That one isn't good, it has been discussed several times.

Attached is a pre-regulator that I've tried with the Mk.3 design, works fine.
I did not finish a psu based on this, but it's a good start.

[AccountRemovedPerUsersRequest]

That one isn't good, it has been discussed several times.

Attached is a pre-regulator that I've tried with the Mk.3 design, works fine.
I did not finish a psu based on this, but it's a good start.

I have one "...stabilized PSU..." built - up and running. Could you point to the discussions about it.

Axel.

[Thor-Arne]

I didn't keep the bookmark for the original threads, there's at least two different versions of that design.

A search here on EEVblog should result in some results on this.

[lunemec]

I'd like also some reasons why it is so bad. Sure, it is not much efficient and there are no thermal or short circuit protections, but that can be easily added.

[AccountRemovedPerUsersRequest]

I didn't keep the bookmark for the original threads, there's at least two different versions of that design.

A search here on EEVblog should result in some results on this.

I found at least one thread and noticed that you have had some issues with the design.
I would not say that the design is that bad. I started to build mine using the unmodified version and fixed all the issues I came across. I found out afterwards that many of the issues are discussed in different forums. Some solutions of mine are similar - some are different.
As a result I have flat straight DC 0-30V 0-3A PSU. Works for me.

But I fully agree that the unmodified version is not recommended as is.

Axel.

[lunemec]

Axel, do you have any schematic of your fixed version? Could you maybe post it here?

Thank you, Lukas

[AccountRemovedPerUsersRequest]

Axel, do you have any schematic of your fixed version? Could you maybe post it here?

Thank you, Lukas

Hello,

It seems that there is so much discussion about the circuit already that all of the issues are covered. Since this circuit seems to be doomed I'm not too eager to publish my revision. (and there is not updated paper version anyway).

However, as a summary, comparing to the Electronics Lab version:
- transformer 180VA 38V
- Q3, U1 and U3 regulated to 26V
- U2 regulated to 35V
- U2 = OPA604
- removed R13, R14, Q1 circuit
- much bigger input and output caps
- Q4 as double 2N3773
- "adapter style" -solution in 5V reference. Used the original circuit and ISL21009 based voltage reference
- R7 to 20W
- small tweaks in component values, added few coupling caps to new power rails etc etc
- overheat protection capability (by separate control unit)
(proper heat dispatching must be taken care of - as probably mentioned everywhere)

Axel.