Power supply and capacitance....

[Lisandro_P]

OK, i moved on to building the power supply of my amp. My choice is a dual capacitance multiplier (for the positive and negative rails), feeding the two amp boards. Now, i've settled for 20000uF as the input capacitor. My concern is, and specially after checking prices: has anyone tried paralleling a lot of smaller value capacitors (say, 10x2200uF or 20x1000uF)? It's a bit cheaper, ESR it's lower (for whatever's worth) and i imagine long term relaibility is better with smaller caps... am i right?

[AudioFreak]

all things being equal, the smaller caps will do a better job (lower esr, better high frequency response, higher ripple current) also think about using an equal number of film bypass caps (polypropylene or polyester) of say 2.2uF one across each of the electrolytics as this will improve the high end of the supply and the amp just make sure the amp will be stable with a low power supply impedance at high output frequencies ... what sort of amp are you planning on building? ... also if you haven't sourced the transformer/s yet think about using large inductors on the primary and secondary to increase the charge pulse width (duration) and more large inductors to form a pi filter for the unregulated output. This will of course make the trasformer work harder so you may need a higher rated one.

[Edited by AudioFreak on 11-05-2001 at 03:52 AM]

[Lisandro_P]

The amp i'm building is a 2x20w JLH 1996... all class-A, so low ripple is a must. The capacitance multiplier seems to work great with these kind of amp (drawing almost constant current). I thought about using multiple smaller caps after checking the prices on them, just wasn't unsure if there was a point... dunno, i just don't like those big can caps, somehow they feel like a problem waiting to happen.

Anyway, the idea is perhaps 10x2200uf or 5x4700uF per rail as the input cap of the circuit. These are all bypassed of course. I thought about the pi-filter, but given the ripple filtering of the cap. multiplier, adding bulky high current inductors seems like too much... anyway, how does they behave (pi filters that is)?

[AudioFreak]

I would still suggest that you use 20 x 1000uF but if the price is an issue and you can get 10 x 2200uF for less than that will be almost as good ........ sure the second option will give you a little extra capacitance but if you get quality caps then 20 x 1000uF will serve you better in the long run. dont get 5 x 4700 unless you really have no choice cause most of the specs cant compare. btw imho ripple current is still an important figure even in such a design.

The pi filter essentially works like a low pass filter and the higher the frequency the greater the attenuation ie. the higher the frequency of the line noise the less of it that actually gets thru to the amp.

Also even if you dont use a filter, I would still suggest that you use a large inductor on the primary and another large inductor on the secondary of the transformer as this will significantly improve the duration of the charge pulse for the caps ........ imho this is a worth while investment ...... but be aware that if you decide to do this, the transformer and rectifier will have to work significantly harder for a large percentage of the time so you will need to factor this in when considering current ratings for the transformer and rectifier.

[Edited by AudioFreak on 11-06-2001 at 03:20 AM]

[grataku]

Lisandro,
I just built the Hiraga and I used a bunch of 2200 and 3300 50V panasonic caps FC series from digikey bypassed with a couple of 10u MKP. I went with a regulator as opposed to a capX.
I mounted the caps on a piece of perforated board and used Cu strips to connect the caps. I had to make two of these boards one for each side of the regulator.

[Lisandro_P]

Originally posted by AudioFreak
I would still suggest that you use 20 x 1000uF but if the price is an issue and you can get 10 x 2200uF for less than that will be almost as good ........ sure the second option will give you a little extra capacitance but if you get quality caps then 20 x 1000uF will serve you better in the long run. dont get 5 x 4700 unless you really have no choice cause most of the specs cant compare. btw imho ripple current is still an important figure even in such a design.
(...)

I'll see if i get to it, but given the simulations i've run (with ripple below 100uV) it seems like a bit too much. How much inductance are we talking about?

[Lisandro_P]

Originally posted by grataku
Lisandro,
I just built the Hiraga and I used a bunch of 2200 and 3300 50V panasonic caps FC series from digikey bypassed with a couple of 10u MKP. I went with a regulator as opposed to a capX.
I mounted the caps on a piece of perforated board and used Cu strips to connect the caps. I had to make two of these boards one for each side of the regulator.

Have you had any problems with the tracks heating up because of the current? I once built a power supply that was wired that way and had to put thick wiring in parallel with the copper tracks.

[AudioFreak]

Which inductors are you referring to? the one for the pi filter or the ones at the transformer?

The inductor for the pi filter will be about 2mH air-cored series with the caps ..... half the caps on each side of the inductor ..... this usually gives between 3dB and 30dB noise improvement.

[grataku]

Lisandro,
12.5 mm X 1.2 mm qualifies as being a Cu bar rather than Cu strip.

grataku

[Lisandro_P]

Originally posted by AudioFreak
The inductor for the pi filter will be about 2mH air-cored series with the caps ..... half the caps on each side of the inductor ..... this usually gives between 3dB and 30dB noise improvement.

2mH at 5 amps.... *bulky* I'll see how the capx performs and if it's needed, i'll get them.