My First Power Supply

[rpetitho]

Hi all!

Long time viewer of the EEVblog and reader of the site, first time poster. I've been working on my first power supply and decided to post what I came up with for version 1. I was feeling confident and sent the PCB to the fab shop so we will see how it works.  I'll let you all know!  You may find some extra cap slots on the board. I put in a few extra to experiment a little when the boards show up. 

A little about myself. I have a background in civil engineering, but then I decided to become a pilot. So I haven't done engineering work in about 10 years. Getting into electronics has been lots of fun and I hope to keep learning. Let me know if I bust any forum protocols, not intentional.

-Bob

[queennikki1972]

If it works out I'd be interested in a printed pcb. My current LM338 board doesn't have current adjustment as Im a new hobbyist.

[MarkF]

As a comparison, I did a similar design some time ago (never built it though).
The only real difference is that my voltage can not go below the minimum of the LM317/LM338 (i.e. 1.25V).
Let us know how yours works.

[Zero999]

As a comparison, I did a similar design some time ago (never built it though).
The only real difference is that my voltage can not go below the minimum of the LM317/LM338 (i.e. 1.25V).
Let us know how yours works.

Did you simulate it? It requires some frequency compensation, otherwise it'll oscillate.

All it needs to go below 1.25V is a negative rail.

[MarkF]

As a comparison, I did a similar design some time ago (never built it though).
The only real difference is that my voltage can not go below the minimum of the LM317/LM338 (i.e. 1.25V).
Let us know how yours works.

Did you simulate it? It requires some frequency compensation, otherwise it'll oscillate.

All it needs to go below 1.25V is a negative rail.

I did nothing except layout the circuit for another discussion some time ago as a starting point.
I did not care if it went to zero.  It was worth not having the extra circuitry for the negative rail.
"I don't simulate.  I build the real thing and test."

Show me what you think it needs? (for others if not for me)


I actually have four of these small HP Power Supplies (three 25V and one 50V):
   

[Zero999]

"I don't simulate.  I build the real thing and test."

Show me what you think it needs? (for others if not for me)

The problem is the constant current loop has too much phase shift and gain. I think that 100nF capacitor is in the wrong place. It would be better between pin 1 and 2 of the LM358 and might need to be a larger value. I would simulate the AC/transient response, tweak the capacitor value, build it and tweak it again, if necessary.

[MarkF]

"I don't simulate.  I build the real thing and test."

Show me what you think it needs? (for others if not for me)

The problem is the constant current loop has too much phase shift and gain. I think that 100nF capacitor is in the wrong place. It would be better between pin 1 and 2 of the LM358 and might need to be a larger value. I would simulate the AC/transient response, tweak the capacitor value, build it and tweak it again, if necessary.

I have been trying to see how the circuit works and found I needed to re-arrange it to distinguish each function.
(I don't mean to offend.)

I believe the datasheet suggests that C_ADJ be 10uF to improve ripple (C2 or C11 in your circuit).

It appears the left LM358 is being used as a comparator instead of an amplifier.

And the right LM358 is to provide a minimum load to the LM338.

You are suggesting that a feedback capacitor (indicated in red) be added to act more like an integrator to slow the leading edge?
I can not figured out which circuit you are referring to?   

[MarkF]

I can't help wonder if the LM338 output voltage adjustment should be as follows:

[rpetitho]

I can't help wonder if the LM338 output voltage adjustment should be as follows:

Per the data sheet your change is correct. The reason I put the the cap between the 500ohm resistor and the 1k variable resistor was to increase response time for the constant current circuit to bring down the voltage on ADJ pin (not having to drain the cap), but still have some ripple rejection from the input when the constant current circuit is not engaged. The 100nF attached to the ADJ pin was there to have a place on the board to do some tests.

I ordered the PCBs from JLCPCB, but they just emailed me and told me they can't do snail mail to Hawaii and asked how much more I would like to pay....A bit annoying, so I think I'll pay the premium and support a domestic producer.