Aleph 2 project

[Thago]

Hi, I'm very busy these days collecting parts for my Aleph 2 project. I will be using the KK-PCB's, combined with Kristijan's PSU, which is essentially a CLC design using 8 times 47000uF per channel, and 2 times a 2.2mH choke. I have some questions about the parts to be used for this PSU:
- what would be best for the choke: air core or transformer core? If I would use air core, would 1.4mm copper wire be sufficient (0.46 Ohm), or would 2mm be preferrable (0.30 Ohm)? FYI, a transformer core with 2mm wire woul be 0.10 Ohm. Air core 2mm and transformer core 2mm would cost more or less the same.
- is it necessary to use that much capacitance? Or would 8 times 33000uF or even 8 times 22000uF be sufficient?
- is the quality of the capacitors used in the PSU a very critical factor?
- would 8 times 47000uF be better than 4 times 100000uF? I understand that large capacitors have difficulties with high frequencies; does the Aleph 2 PSU have to deal with very high frequencies?

It's a lot of questions, I know...

Thanks in advance for your replies!

[MEGA-amp]

- find the biggest air core your budget will allow..between 14-12 gauge should be fine.. bigger is better here

-the commercial aleph 2 used 4x25000uF per monoblock. 8x33000uF or 8x22000KuF would be more than adequate. Be sure to include some form of soft start.

-quality? RIFAs, BlackGate are good from what I read here($$). I'm sure somebody will tell you it makes a difference between exotic and cheap. I buy the biggest, cheapest I can afford..works for most.

-8x47000uF is most likely your best bet; lower ESR than 4x100000uF.

Take a look here

[analog_guy]

In general, electrolytic capacitors have poor ESR and ESL. especially at high frequencies. It is better to parallel multiple capacitors, rather than use a single capacitor of equivalant value. Another useful trick is to parallel several film capacitors (0.1 - 1.0 uf) across the electrolytics. The ESL of the film capacitors is considerably lower.

What you want to do is present a uniform power supply impedance (usually as low as possible) over the frequency range that the amplifier will draw current from the power supply.

One of the major PC CPU makers had a similar problem with power supply resonance, and it nearly killed their yield. The resonance in this case was at approx 500 MHz.