Lab Power Supply Design / Build

[MJL21193]

A well used and indispensable tool for the shop is a reliable power supply. Some of the features I would like to incorporate into the design:
- Adjustable voltage. As planned, 0 - 30VDC
- Isolated dual supplies, independently adjustable that can be used on their own or put in series to form either a split 30VDC supply or a single 60VDC. When put in parallel, the 2 supplies can give double the current.
- Adjustable current limit for both supplies.
- Built in analogue voltmeter and ammeter on both supplies.
- Built in current limited receptacle. Current limiter in this case is a regular incandescent light bulb in series with the mans. This is for first powering up external power supplies. The bulb will have a bypass switch.

I may think of more along the way
I actually started this project a while ago (familiar story...) but stalled. I have a supply that I built a couple years ago and have been using that. It has some short comings:

Picture 1108.jpg

With its lid off. One cheap digital ammeter one one leg of the split supply.

So,
Start with the circuit:

LAB SUPPLY SCH.png

One channel shown, both channels are identical. I didn't know for sure if it with work as planned. It worked well in simulation but that doesn't necessarily mean it will in reality. As I only have a minimum of actual knowledge, it is yet another cobbled together circuit, drawing upon several sources - a "patchwork" power supply I guess.
Today I built a prototype and ran it.

[MJL21193]

It looks a bit complicated and it is but functionality first, simplicity second.
First thing I discovered today after I ran the prototype: I don't have enough current at the base(s) of the pass transistors. Current to the load during testing was maxing out at ~800mA. To increase this I think that R1 and R2 need to be a lower value, to pass more current as needed.
Voltage adjustment is incredibly linear and smooth. For the prototype I'm using a multiturn trimpot and I think I get panel mounted multiturn pots for the final version.

NPX_845.JPG

In the background (among the mess) are the meters I'll use.

[MJL21193]

On second thought...it's probably best to limit the amount of current that is drawn from the doubler supply. Ripple will undoubtedly go up if I increase the draw.
I feel I may need to either switch the pass transistors to darlington or add a "driver" stage to increase the gain. I think I'll go with a driver, since I don't have any power darlingtons here:

LAB SUPPLY SCH.1.png

It shouldn't be too difficult to add it to the prototype board for a trial run.
Also, 2 pass transistors is probably overkill here and I may eliminate one and with it, both of the emitter resistors.

[MJL21193]

Adding the driver fixed the problem. There is enough gain to easily drive the load. I used MJE15030 for this because I have many of these and no other use for them at the moment.
The revised schematic:

LAB SUPPLY SCH.png

I have deleted the second pass transistor. I think one MJL21194 can easily handle any load this supply will see. Heatsinked and current limited to a max 2A, it should be fine.
Voltage regulation is outstanding. Searching for multiturn pots on Digikey had me looking at units from $20 right up to $150! Who would pay this much for a 5 turn pot? NOT me, I'll either use a single turn or create a shaft attachment for a multiturn trimpot.
I haven't really tested out the current limit yet. I know it functions but I need to see if it has the variability I want. One concern is Q6 - it may need to be changed to a larger device. When the supply is in current limit, Q6 sinks the overload current from the base of Q1. Milliamps only but maybe too much dissipation for a TO-92.

[MJL21193]

Some fiddling with the current limit and I've found that R12 needs to be halved, from 6K down to ~3K (I used 2.7K). Some explanation (as best I can):
R12, D5 and R17 provide voltage for the base of Q6, through pot R18. Rotating pot R18 counter clockwise (less current) increases the voltage at the base of Q6. This in conjunction with the voltage drop across current sense resistor R16, turns on Q6 sinking current from the base of Q1, thus reducing the voltage. Current at the output stays constant as the voltage drops.
By reducing R12, I increased the "threshold" voltage (for want of a better term) giving better variability near the end (max current) of the current adjust pots travel. To fine tune this I will put a 500R trimpot in series with R12.

[MJL21193]

As usual, I was wrong. Reducing R12 was the wrong move and I should have seen it. Sufficient to say that 6K for R12 is still a bit too low. The supply goes into current limit before pot R18 is fully counter clockwise. Ideally, it should hit "0" before that happens, Full scale operation of the pot is what I'm after.
Making R12 8.2K plus a trimpot in series should do the trick. I found that 10K was too high, limiting the low end of the current limit to ~300mA at 30V output.

[MJL21193]

Ok, I have it sorted out:

LAB SUPPLY SCH.png

Quite a bit of monkeying around and I have it where it should be. Current limit goes from a low of 150mA @ 30V to 2.1A at maximum setting - excellent!
At lower voltages, it is possible to set the current limit even lower, to 100mA. I am pleased with this result.
I have made a couple other changes: Added a trimpot in series with the adjust resistor for the LM317. It allows the reference voltage to be set to exactly 15VDC. This will in turn limit the output voltage to exactly 30VDC. I also changed the grounding location for the LM317 to the rectifier side of the sense resistor.
Next is to do a new board layout.

[MJL21193]

Board layout:

LAB SUPPLY BRD.png

Small - 10.5cm x 9.5cm.
I'll need 2 of these, one for each supply.

[ostripper]

Ahh , "men of action" at work... only 30V ?? I need a dual 0-50V setup instead of my crude trafo/bridge/ELcap setup. Your looks similar to the Nico "ELD" ps (below). I am trying an IC-less regulated design with a current sourced Zener (you hate zeners - but 317's have 4 of them).
OS

[MJL21193]

Originally Posted by ostripper

(you hate zeners ...

I got over that

Hey OS,
30 volts is fine for 99% of the testing and prototyping I'll do. If more voltage is needed (like for a completed amp ready for serious testing), I would build the supply that powers it. I think it is a bit wasteful to construct a supply with high voltage and current capability unless it's getting used constantly.