Multi taps or single supply for linear PSU ?

[MathWizard]

I have a few multi tap/multi-winding transformers from stuff I've scrapped over the years, and I'm looking at some different CV-CC linear PSU schematics, with and without op-amps (so nothing too modern or flashy), and I'm modeling them.

I don't have exactly the right parts, for most of these design's, but I can get close enough and modify the circuits a little too. But then I could probably replace whole sections of some circuits, with parts from other design's.

Lot's of designs use a single supply rail for everything, so your reference and controls, pass transistor, LEDs, etc. Some use other windings to make nice low noise references, and then tie it to the output.

So what's generally better ?

[ledtester]

I think you're talking about using a separate power supply for the op-amps which is then floated onto the output voltage (the output produced by the PSU you're implementing).

This video has a nice explanation of how such an arrangement works:

Building a Lab Power Supply with a Discrete Linear Regulator
https://youtu.be/_CFIovMkRyg

A schematic of the entire design may be found around 21:40.

Note that the op-amp supply rails are only 10V apart. The alternative is to use the input voltage (from the bridge rectifier) and ground as the op-amp supply rails. For a PSU designed to deliver a high voltage (like 30V+) this might be too much for the op-amp. For instance, for the LM324 the max difference between its V_CC and V_EE is 32V. Moreover, it's maximum output voltage is at least a few volts less than V_CC. Using a separate supply and floating it around the output voltage solves these kinds of problems.

[MathWizard]

Yeah that's 1 I've been looking at, I also have a great trans. and heatsink from a home stereo amp, but really it's a lot higher voltage than I can handle heat wise for now. But I want to use my transformers in a modular sense. In fact, I should put 2-3 trans. all in 1 box, under the bench, and have all the IP/OP and switches/fuzes in that 1 box, then attach what ever PSU I want to what ever taps. That's half the hold up on my PSU, is worrying about what winding's/combo's to use. Another 1/2 would be trying to make it from what I have on hand (which is enough tho)

For now I'm going for anything 15-30V range, 1-2A type design's, and with the present heatsink from the stereo, I was looking at only 20W that I was comfortable with from the TO-220 pass transistors I have.

I did start making a dual LM723 design on the breadboard, then realized I must have fried 1 LM723 years ago. And I was too cheap to buy another 1. So I was modeling using the 1st LM723 reference for a block diagram model of the other 1, but LTSpice can be a pain, and then I went to other designs.....and learning more.

I spend the last 2 days load test my various trans, I have 2 that be great for 15-20V 1A stuff, and 1 that should be fine for 30V.

[coromonadalix]

the simplest design(s) would be fixed regulators like lm7805 lm7808 7812 7815 .... and their negative counter parts lm7905 lm7908  ...  you just need at least 4 volts input voltage higher than the output

a variable one would be lm317 ...

All of them are for low currents, say at least 1 amp,  for beefier currents and voltages, you'll have to do more evolved designs like  lm723, or op amps based psu's if you want to add cv and cc options


sorry  MathWizard post got there before me  loll

[MathWizard]

Well I haven't considered making it with LM317/337, I watched Dave's video where he makes 1 w/ a better version LM317

So what would it actually be like with LM317's in parallel ?

I really want to make this, just for the fun of it, and it would make it easy to poke at and fix too.

heathkit_ip-18_1-15v
https://elektrotanya.com/heathkit_ip-18_1-15v_dc_regulated_power_supply_sm.pdf/download.html

I more or less had a 18V model working with my trans., and was about to build it until I realized the heatsink temperatures. Also my UHF/VHF JFETs are only 25V rated, and 12-15V PSU is not what I was aiming for ATM

I should be making some of this modular anyways too, like the current sense resistors, I can put a few of them on 1 board, and go ahead and put a 2 or 4 pass BJT's and their resistors on a PCB on the heatsink.

[xavier60]

This is another option. It has an 8V control rail regulated from the main unregulated input.
https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/msg3089254/#msg3089254
The design has been working well in one of my bench supplies.

[MathWizard]

Thats pretty cool looking, I have lots of low voltage high current windings. I have lots of op-amps and TL431's.

I need to make the heart of a few of these on proto-board. Another annoying thing is I'm short on long-life pots, and the common sizes in general. For the Heathkit model, the 100R pot in the load current path is a problem, but if I used a few extra 1R 1W resistors on a switch before a 100R 100mW potentiometer, it was ok in models.

I don't mind if I build a few of these as control boards with ribbon cables, what else am I going to do with all my junk ebay parts.

Once I get the hang of it more I want to try some digital control w/ ADC's and MCU's. Or a SMPS, I have lots of salvaged SMPS parts of all types, but I never really tried to make a proper 1 yet.


I have a nice Siglent PSU*, and a few 12V SMPS and a 24V SMPS, so I'm not in dire need (*the 2.5-3.3-5V ch stopped working, but I haven't looked inside yet)