BPA Regulated Power Supply

[h_andree]

Hello all,

I'm interested in building a BPA configuration monoblocks
with 2 x 4 lm3875/lm3886 per chassis. They are going
to drive 4 ohm speakers.

I would like to keep the voltage low because of thermal
considerations. Probably 28-29V DC.
Therefore I was considering the following
components:

Per chassis:
625 VA 2 x 25V secondairy toroid transformer
MUR 1520 Diodes (8)
About 100,000 uF of filtering caps

Now the questions:
I think one lm338 per rail is probably not enough
(For 4 chips). So should I try the 15A regulation
as suggested on the lm338 web-site or
should I go with 8 power supplies each with a 317/337
combination?

Has anybody compared regulated vs unregulated
supplies. I saw some posts on caps after the regulation
but am still not sure on the quality of the sound.

Any suggestions?

Thanks

Harry

[jleaman]

I have a few of the regulated boards ill post a picture of them. Id presume that regulated is better than none regulated psu for any thing..

[macboy]

You may also want to consider a "capacitor multiplier" instead of regulation. You still get the important benefits of regulation (less ripple and noise) but the voltage will sag a bit under heavy load. A good place to start is Elliott Sound's project 15.

[tlmadsen]

I am not sure that I think a regulated suppply for IC based power amplifiers is "the answer to all problems" but if you really want to go that way, why not use the LM3886 as regulator ??. Given a steady voltage ref.(like a LED) it should work fine.

BR

Thomas

[jackinnj]

each regulator is going to dissipate quite a bit of heat -- consider using a discrete regulator with a TL431 as the error amplifier/voltage reference -- this can be configured for both the positive and negative rails. lm317/lm337's can be paralleled, or you can use an external pass transistor with these units --

0.1F is overkill -- the Overture chips have excellent PSRR --

[h_andree]

I know that the overture series has a good PSRR.
However, I believe that if you decide to go with
regulation you have to be within a regulation
limits.

With bridged parallel at +/-28V i can get maybe
10 amps out of the amp continuous. That
means that if I want to power supply not to drop more
then 1V at 10 amps within 1 cycle of 1/120s I need
about 100,000uF for the total amp.

Is this incorrect? You can argue that you let the ripple be larger.
But then why regulate?

My plan is to research regulation for a single
chip first. Then if it sounds better then unregulated
(like was posted) I would like a plan to scale it to
a more powerful amp.

I've build some gain clones and they sound nice for vocals
and Jazz but lack slam with large speakers, specially
for symphonic music.

I believe that Jeff Rowland uses regulation. IMHO that's
one of the best sounding amps I've heard.

Harry

[Nuuk]

FWIW, I've tried regulated PSUs (LM338) and the CarlosFM snubbered PSU and if anything the latter sounds better. In addition, GC's with the snubbered PSU run cooler and are easier to implement.

[h_andree]

Hello all,

I'm experimenting with the BPA design. I'm running 2 chips in
parallel right now. I only got 0.3mV difference between the
2 chips. I get a great power out of them. They run very
cool. I did match the resistors to 0.01%. I'm using Holco
H8 from Mouser and have them matched even further.

So right now I'm using a 2134 buffer and 2 x lm3875 in
parallel. Adding the buffer improved the sound.
I feel there's less glare.

I'm using an unregulated supply with 2 x 68,000uF caps.
I will update when the regulators come in.

Harry

[h_andree]

Snuberizing the previously mentioned supply
did not make any noticable change in sound.

Why is that?

Harry

[carlosfm]

Originally posted by h_andree
Snuberizing the previously mentioned supply
did not make any noticable change in sound.

Why is that?

Harry

Those caps are very big.
I wouldn't use 68,000uf caps.
What values did you use for the snubbers?