need help with 30v 4A power supply.

I need to build a power supply. Probably CLC configured.

It is for a SS integrated amp. Should I have 2 seperate power transformers per channel?

or common transformer and seperate bridge rectifiers?

anyways.... it will be feeding a source follower class A output stage.

Need help with capacitance values, inductor values and size of power transformer.

initially thinking 20mF 2mH 20mF per channel... should these be bypassed also?

look up Nelson Pass site for transformer and capacitor advice.
I suggest for a ccs loaded SE classA monoblocks about 6 times output power for each channel and about 10mF to 20mF per amp of peak current. The follower side of the supply will have current rangeing from 0A at max -ve power to 2 times idle at max +ve power.
here is what i got so far

going with a 25vrms 10A transformer per channel.

to a 25A bridge

then to 68mF 50v cap -> 2mH choke -> 68mF cap per channel.

fused at 8A.

question is will the ripple current be too high for these caps?

Nippon U32D 105C 0022L 63

is what they say.

should I bypass these caps?

oh and this is only a 30v supply not a split +/- supply.

Caps are too small

If I read your post correctly you are building a 15 Watt SS amplifier.

30/2 = 15V peak times 1/SQRT2 = 10.65 VRMS max

At 8 Ohm load this is an RMS output current of 1.4 Amps

Now the caps are recharged every 8.3mS (bridge) so the voltage drop across the input cap could be as high as (A DC current aproximation)

Seconds times Amps = Volts times Farads

(.0083 times 1.4 amps) divided by (68 e -6 Farads) would equal a ripple of 170 Volts. Obviously the caps will be drained long before this happens.

6800 uF is more like it; resulting in a 1.7 Volt 120 Hz ripple for a DC load of 1.4 Amps
I think Hermanv misread your milli and micro. How do we teach everyone to use the correct letters?
Your plans sound substantially good. 2.5Apk into 6ohm (minimum) = 27mF/A & xtrm=16*va. But you're right to query the ripple current ability of a single PSU cap. I would use multiple // caps here, but then I'm different!
Try adding plastic film //with PSU caps & listen to your sound quality. Same again for on board decoupling.
What are you using as DC block on output to speaker?

Hi Andrew;

If millifards were meant you are correct. 68,000 uF is a fine choice. Proabably 33,000 per each side of the choke is plenty.

I also agree strongly with your series cap question.


Consider if you can, using a center tapped transformer and a split supply. The problems in sound quality caused by a series coupling capacitor to the speakers are legion. Some will even spring for film capacitors of hundreds of microfarads just to avoid this. With a split supply the problem moves over to the power supply capacitors, but you have to pay for those anyway. (The PS filter capacitors actually end up in series with the speaker if you follow the speaker current all the way around the loop)

The parallel films are a good idea but to do very much good (IMHO) they need to approach about 1% of the total capacitor value they parallel which would be a lot of uF to pay for. It depends a lot on the quality of the electrolytics you use.

so if i wanted to not use the 68mF caps but instead use smaller caps but paralleled.... what do you suggest??

yeah... the cap coupling to the speaker.... well....

it needs to be like a 50v 8,000uf cap... of course paralleled with film caps.

dunno what brand though.... should I just use what i have in the PS? since technically the PS is also in the "signal path" ?

maybe a nice 50v BG cap?...

Thanks for all your guys help..

p.s. is my 25vac 10A tranny per channel good enough? and finally what DCR should I limit my inductor in the CLC supply?
Electrolytic caps

There is an internet store, Michael Percy that specializes in high performance audiophile parts. Sonic Craft and Parts Connextion are others.

Michael sells among others Nichicon capacitors that are reputed to sound better than most electrolytics for speaker series capacitors. Even with a DC polarity it is generally believed that non-polar electrolytics sound better. He also carries large values of film capacitors but they are not cheap.

Michael beside carring good parts will offer some advice if asked, I've always found him sensible and knowledgeable.

I have no financial connection with Michael, I like him and hope he thrives.
why not a stabilized psu with lm338? Solid as a rock and very simple.
Look at carlosfms' thread:

this chip can handle pulses up to 12A.

For audio use i wouldn't use rectifier bridges, recovery time of those diodes is too slow. You also have to bridge every diode with a capacitor (typically 22-47nF) and alternatively a small resistor with this capacitor in series...

using a regulated psu requires much less elko capacity than in unregulated psus.

use one transformer for each channel, two independent secundaries per trafo...
CLC is nice and simple and with a large second C will perform very well.

do not go LC unless you have done a lot of research on how to design it correctly.

Pass has used & gives designs for regulated, go and read again!!!!

I like the idea of the cap multiplier mostly because it was brought back by the late ( & great ) JLH. ESP has some ideas on modifying it and there is a web site dedicated to JLH.

However all my designs & mods are based on RC (very small R) so I'm just talking through a hole in my hat.
Power supplies

CLC supplies will be quiter and most probably sound better than CRC, LC may be better yet because there is no low impedance path for RFI coming through the diodes to get into the rest of the circuits.

Part of the problem is getting an L thats truly capable of handling the peak currents and a second problem is carefull grounding techniques are required to take maximum advantage of the CLC desgn (Or any other good design I suppose)

Once you start to think about a capacitor multiplier it becomes a very small step to change to a fully regulated supply.

What is this worth? Only the end user can tell. Like most things about quality sound it can be tiny details that make the difference between adequate or average and stellar. Most of us spend hours or many dollars (or both) trying to get that last bit of performance from our systems.

Contemplating a regulated supply does warrant consideration if the rest of the reproduction chain is good enough for the difference to become noticable. Once you've spent the money and wasted heat of a capacitor multiplier a regulator is only a single, small, low cost IC away.

Many fine amplifiers do not use regulated supplies but others do, its not a slam dunk that the simpler supply is better.

To a large extent this decision depends on your skill, experience and access to test equipment, CLC has an advantage in that it offers very few traps into which one can fall.
here is the thing

its for an output stage using mosfets in a source follower config.

so there is some inherent PSRR just in the output stage.

ripple down to nothing is not the idea.... just a good "sounding" ps is.

Also cost is a factor.

if i go with a LC PS then my power tranny is a 30v 10A unit.
if i go with a CLC PS then my power tranny is a 25v 10A unit.

one power tranny for the LC supply. 2 for the CLC.

with the LC looking at least a 10A 10mH choke. with loads of C.

for the CLC dunno yet.


my FINAL decision:

Per channel:

Hammond 167S25 25v 10A
27A IXYS bridge rectifier
2x 10,000uf Nichicon MUSE caps 35v
2.5mH aircore inductor 1 ohm DCR
2x 10,000uf Nichicon MUSE caps 35v
1x 10uf film cap (unknown as of this time)
1x 0.33uF RTX cap

how is that??

good for 3-4A per channel at 25-26v DC

Inductor size

Design looks good. 50V caps do give better margin but add size and cost.

That inductor: 2.5mH (milihenry right?) has a reactance of 1.9 Ohms at 120 Hz (full wave bridge).

If you cant afford a bigger choke you will get little benefit over a CRC design. The DC resistance of the choke does almost as much filtering as the inductance. I agree that a common drain (source follower) will have good PSRR so you still end up with a working design.

You're just getting almost no benefit from the choke.
25Vac transformer on no load will increase to regulation % above nominal, on max mains voltage this increases again (in UK +6%).
then times sqrt 2 to convert to DC and the answer comes to 39.7Vdc.
You must use 40Vdc or higher. DO NOT use 35Vdc caps before the choke. You might get away with 35v after if you can guarantee that a load will pull the final caps BELOW their safe rating.
As an opinion I think you will get more benefit by changing the ratio of caps before the choke to after the choke. I suggest before =1 and after about 2 to 3 times. i.e. 4,700uF before and 10,000uf to 15,000uF after.
This is a classA amp? then your cap values are a bit low. The choke helps because it stores some energy but at your current values I think the energy store is too small to be really effective, can anyone confirm? PSUD should give some answers.
Finally if cost is important your decision to eliminate the LC is correct. The choke would probably have cost more than the combined cost of all the other components . Even CLC will be dearer than CRC and RC is cheapest of all (the R is just the transformer secondary + wire to rectifier + wire to caps).