Dual Polarity Linear PSU design questions beginner

[tannhauser-gate]

Hi !

Was wondering if I could get some advice concerning my train of thought and wouldn't mind some math or equations to see how some of you professionals would do the design especially on the filtering caps side. Am thinking there is theory & math but also checking it on the scope to adapt if necssary and not over engineer (I am by nature a perfectionist which is not always a good trait to have when designing something) :

I've started designing my 12V, -12V dual rail PSU. Objective, it has to be low noise/low peak to peak ripple for audio grade purposes .

I will be using
* 80VA 2x 15VAC toroid (230vAC input) --> thought this would be ok as I will end up around 20v before rectifying *1.44 , will only pull max 1,5A and probably a bit lower
* 4 STPS2H100 schottky diodes to rectify (low noise, low forward voltage drop)
* The filtering stage pre voltage regulator, from what I read a general guideline is 2200uF/1A
* the thing is I want low ripple, I see schematics where they put 3x3300uF in parallel, or just 3300uF --> I don't want to over engineer here
* I have 2x 3300uF per rail @50v high quality audio psu caps from Cornell Dublier --> I now worry about inrush current but also when powering-off, would like to decharge them safely
* was thinking soft-start ? --> what's the best practice nowadays cause there are multiple options from what I see, my requirement is that any resistor gets disconnected once charged over short period of time as psu is already small with its 1.0A-1,5A
* same question for the discharge, was thinking mosfet based IRLZ44N & 10v Zener diode (got 5 of em lying around) with a 100K resistor to safely discharge ? Would only turn on when voltage drops below 10v thx to Zener so it wouldn't consume power beforehand
* I read about capacitor multipliers, would this be an option instead of putting 3 or 2x 3300uF for ripple reduction ?
* I see in some schematics big caps are also put after the voltage regulator --> My idea was to start small, measure peak to peak ripple with an oscilloscope or get an approx value with the rms AC measurement & some math of my brymen 869 & adapt , I basically want under <10mV of ripple . Can I use the same mosfet discharge circuit for the caps post voltage regulation as well somehow ?

* a VR LM317T & LM337T can reduce ripple significantly by almost 40dB and with a bypass cap even more from what I read in practical electronics for inventors ? Overkill or not ? Guess i'll first measure with oscilloscope or Brymen 869 ?
* 5K trimmer resistors
* i also sometimes see small non polarized caps being used in the filter stage going from 0,1uF to 1uF , does this make sense for audio grade PSU as I will be <25KHz audible stuff ? and I understand that these are for high switching noise coming from schottky's potentially & AC ?
* diodes 1N4003 to protect Vout to Vin of my VR's

As I got most part I have another dilemma, if i want to load test this circuit this is never going to work on a breadboard, I got prototype soldering boards, is thick soldertraces enough to carry the current ? or can i get special boards that can take more current (up to 1,5A)

tx so much

--- Updated Feb 13, 2025 ---

added my schematic here too to support my post above

--- Updated Feb 13, 2025 ---

3 caps 50v 3300uF in Parellel is pretty lethal Joules wise if it discharges while prototyping (if i decide i need more after measuring), that's why I was thinking of a discharge circuit

 

[betwixt]

D2 and D3 don't do anything. There may be some advantage to add them across the output to ground to protect the regulator if the ground comes adrift.

Your biggest issue is using half wave rectification, you are only using half the input waveform so ripple will be higher. Consider connecting the rectifiers in bridge configuration so both supplies are fed by both sides of the transformer secondary.

If you are worried about "3 caps 50v 3300uF in Parellel is pretty lethal Joules wise", consider adding some over current protection before the regulator.

Be careful to read the specification of LM337 and LM317 with regard to the values of R2 and R3, from there calculate the value of the variable resistors.

The non-polarised caps are not essential but there is no harm in fitting them. It may help with stability and interference rejection.

Forget 'push in ' breadboard but you should be able to get away with soldered boards. The current isn't going to be very high so no need to take special precautions. The important factor is to keep the wiring between the regulators and the capacitors short. There is an amplifier inside the regulator to magnify the difference between actual output and programmed voltage, it can go unstable if the wiring is too long.

Keep the ground connection as short and thick as practical. There may be some advantage to using star grounds to reduce ripple. That means one common ground point where all the ground connections meet instead of them being wired in a chain.

Brian.

 

[BradtheRad]

While building my home-brew power supply I followed advice to install a 'bleeder' resistor to discharge the caps. In series I included an incandescent bulb to serve as an On light. The incandescent type has no voltage threshold hence the caps discharge to 0V.

--- Updated Feb 14, 2025 ---

C8 C2 are upside-down. I believe C6 also.

 

[betwixt]

I didn't spot that - you are quite right Brad.

Brian.

 

[KlausST]

Hi,

looking at the schematic .. I wonder why you don´t use other reliable schematics as reference?
Every semiconductor manufacturer provides them. They take effort to give good and complete information .. all you have to do is: read them.

wouldn't mind some math or equations to see how some of you professionals would do the design

I´m a professinal .. for decades now. And that´s wow I´d do it:
TI is one of the manufacturers, On the LM317 web page they list 15 related cocuments, including many application notes.
Indeed the datasheet alone tells you all you need to know to use the LM317 correctly.
And in the about 10 application notes you get additionaly informations on practical use. Exactly for what you ask in this forum.

Additionally there is an online design tool "webench" .. where you simply input your data and get back a schematic including all part values.

Am thinking there is theory & math but also checking it on the scope

For sure. And available and shown many thousand times in the internet.
You talk about power supply capacitors. Did you look at wikipedia? Did you do an internet search?
There are even online calculators you may use. For free. For you.

From your text I have little evidence what effort you really did.

I will end up around 20v before rectifying *1.44 ,

Before? are you sure?

would like to decharge them safely

Why?

was thinking soft-start ?

Why? What benefit do you expect? How do you expect it to work?

what's the best practice nowadays

An SMPS. Nowadays one does not want those heavy, expensive, lots of heat dissipating power supplies.
LM317 is technology from 50 years ago. Technology went on since then.

a VR LM317T & LM337T can reduce ripple significantly by almost 40dB and with a bypass cap even more from what I read in practical electronics for inventors

The dataheet specifies the ripple rejection. I´d rather rely on the manufacturer´s information than on any random second source informations from internet.

As I got most part I have another dilemma, if i want to load test this circuit this is never going to work on a breadboard, I got prototype soldering boards, is thick soldertraces enough to carry the current ? or can i get special boards that can take more current (up to 1,5A)

Again ... there are online calculators for Amperes vs trace thickness and width and other parameters.



There are so many mistakes in your schematic .... If you wire it the way you have drawn ... there is a good chance for an explosion.

Some issues are alredy mentioned by others

D2, D3: I guess they should be part of a full wave ractifier. If so wire: D2_A --> T1_3; D3_C --> T1_5

C2, C6, C8: Wrong polarity. Turn them around.
You already marked the signals as "negative", thus you also need to put the capacitor´s negative side to the signals.

Missing GND-junctions at: R1, R2, C7, C8, C9, C10. We don´t know whether this is intentional or not. Please clarify.

Missing a lot of part values. Please give complete informations.

You put pots to make the output voltage adjustable. But I can´t find what´s your expected output voltage range.


***
From my experience:
In opposite to single power supplies dual power suppliees can suffer from problems when one side powers up first (or you have unequal output load conditions).
In this case - during power up - a short negative spike (caused by load current / circuitry) may occur. In one application this made the "slower side" not to power up at all.
I tested several voltage regulators. Even with same part numbers different brands acted differently - without mentioning this in the datasheets.
To be compatible to all brands I installed a "reverse voltage" protection diode at the outputs (Cathode to positive terminal) . After this I never faced these problems again.

***

Klaus

 

[tannhauser-gate]

Thank you so much guys for the feedback and very honoured Klaus that you had a look at my post, I often read your feedbacks.

I'll have some homework to do based on all the feedback and will get back in due time.

I did use a reference schematic from Practical Elecontrics for Inventors 2nd edition. The caps are reversed on the schematic too !!! Contacted a buddy of mine who has the last version and guess what, the caps were put the right way on his schematic ! A big lessons learned for me that even schematics in reputable books can be bad, and that I have to learn to read schematics more thoroughly.

 

[betwixt]

Apart from the caps being reversed on the negative supply, it does closely follow the manufacturers advice. Note that your schematic has the rectifiers wired differently to the image in the previous post.

Brian.

 

[Easy peasy]

The transformer sec wdgs are under utilised - better to have a full bridge on the outside wdgs and connect the caps to the centre tap !