P3A Power supply
 

Roughly how many channels in the P3A amp (or any amp, really) may share a single transformer? I'm looking to build 8-channels of amplification based on the P3A and, ideally, use a single transformer (1x1kVA + 20,000uF per rail). Each channel will only be 40W continuous/8R (+/- 28.3v rails, 40v CT xformer).

Any thoughts?

Thanks in advance.

40VCT may be a little short for 40W/8R, is it OK to be 1dB low?

I would use 22-0 + 22-0, and use a one bridge for the + rail, and another for the - rail. One complete supply per channel.

This will help avoid high current ground problems.

The ATI multichannel amplifiers are built this way.

For 8R loads and 60hz the minimum filter capacitance is 3,300µF per channel. Use 6,800µF for 4R. For 50hz the minimum would be 4,000µF and 8,000µf respectively. Each rail for each channel should have a 22µF ~47µF bypass cap. If you add 0.1µF film caps too, be careful as it may oscillate (try adding 2R2~4R7 resistors in series with the 0.1µF).

The transformer should be 2VA for each W out at the lowest expected impedance. For 40W8R and 60W/4R that would be 120VA per channel, or 1KVA for eight channels.

Use an inrush current limiter.

You don't need anything like 1kVA per channel. That's just a waste of money, weight and space. I wouldn't bother with anything over about 200VA per channel, myself and even that is plenty.

If I was building this system, I would go with four 40VCT (20-0-20) 250VA transformers (one per two channels) and one bridge and cap bank per channel. You can go with bigger transformers if you use bad quality ones (bad regulation). Also, IMHO 20,000uF per rail is overkill - go with 10,000 per rail and save yourself some money. For what it's worth I can't ABX my 4,700uF per rail P3A from my 10,000uF per channel version and I would be happy to give a case of beer (Amstel or Windhoek) to the first person who can.

You can go with more channels per transformer, but you will find that over about 300VA price doesn't scale linearly with VA (at least not in South Africa).

Oh and as djk says an inrush current limiter is a must, especially if you are using toroidals.

I think you misread the previous post.

The real question here is this :

Though seperate bridges and capacitances are laudable for
interaction of the various amplifiers it precludes the use of
a single bridge and very large capacitor bank that under
dynamic conditions will give lower ripple than seperate
supplies.

Each amplifier obviously must be separately fused.

With just two channels playing for example, they would
benefit from a way over specced power supply.

I think it really depends on the intended use of the eight
channels, unless there is a fundamental disadvantage
of a single supply I am unaware of.

:) sreten.

There might be some slightly (negligeable - measurable but not audible) crosstalk issues between channels. But that depends on the P3A's PSRR and I can't say if it's going to affect SQ...

I might have misinterpreted. A single supply for all eight channels would be fine. The only reason I recommended four seperate transformers is that, around here, four 300VA toroidals are cheaper than one 1.2kVA toroidal.
As you said it depends on what they are being used for. If they are playing the same program then it won't help, but if they are playing different programs then one single large capacitor bank will be better. However you would need pretty fat wiring inside the PSU, but that's not really a problem.
Yes.
By Rod Elliott's claims and my measurements the P3A's PSRR is very good. It's not on the level of some opamps but is good for a discrete design. I don't think that one supply for eight channels is going to have a significant effect on SQ either way assuming proper design and construction techniques.

Thanks everyone for your replies.
First of all, I was asking about using one supply to be shared by all 8 channels :-D - so hell yeah I'm not using 1kVA per channel! I'm a student and do not have that kind of money!
So basically I was asking if interaction between channels sharing a supply would be so bad that the idea isn't feasible. It's difficult to quantify 'so bad', but I'm not trying to eek the last final percent in performance. I just want a solid, good-sounding amplifier.
BTW the 8 channels are for driving a pair of 3-way speakers (with the bass channel doubled up with 2 drivers per side... so 3.5 way? hmmmm dunno).

Thanks

One supply for eight channels certainly won't be 'so bad'. In fact I would guess that in the great majority of listening environments it won't be bad at all.

If possible I'd suggest you to separate the amps for the bass and use a higher voltage rail for them, since the woofers take most of the power (and with 40W might run out of dynamics when playing loud)

Cheers

Andrea

Good point. If I were building this system with Rod's projects I would go with the P3A at low rail voltage for the satellites (I really like this amp, as I am sure most can guess) and a different amp for the bass channel (subwoofer). A P3A with 35V rails would do the job nicely, but you could also take a look at P68 (300W subwoofer amp) or P101 (Mosfet amp) if you need that extra power into dynamics.

For music you don't need so much subwoofer power, but for Home Cinema a powerful sub is really nice (until the neighbours call the cops during Saving Private Ryan, that is).

Wingfeather, have you been to Rod's forums? Follow the forum link on his page. There are a lot of guys their who are very knowledgable about his projects. Rod's forums are a good complement to diyAudio.