What is the correct size of a power supply bypass capacitor?
The idea is for a large capacitor bank feeding the main amp. The amp is an F5.
I heard that the bypass caps should be 1/10th the size of the main bank.
So, if there is 94,000 uF per rail, then the bypass caps (ie. those actually in the amplifier side of the PS) should be about 10,000 uF.
Is this correct?
Thanks!
The idea is for a large capacitor bank feeding the main amp. The amp is an F5.
I heard that the bypass caps should be 1/10th the size of the main bank.
So, if there is 94,000 uF per rail, then the bypass caps (ie. those actually in the amplifier side of the PS) should be about 10,000 uF.
Is this correct?
Thanks!
I don't agree that putting capacitors in the wrong place improve performance.
Decide what you want a capacitor to do. Then select a capacitor that can do that job effectively. Finally select a location to allow the correct capacitor to do it's job well.
That is very different from adding bypasses onto the main smoothing of an amplifier. But before I understood I made exactly that mistake. I have learned.
Decide what you want a capacitor to do. Then select a capacitor that can do that job effectively. Finally select a location to allow the correct capacitor to do it's job well.
That is very different from adding bypasses onto the main smoothing of an amplifier. But before I understood I made exactly that mistake. I have learned.
I find that as I go up to 300 uf there are differences, above that not so much. It may be coincidence that that is equal to 8 ohms at 60 cycles, maybe not. I actually use film capacitors for that. But if you use an electrolytic then consider a small bypass of 1 nf with a good RF capacitor and another of around 20 uf film, but not mylar!
I strongly disagree with the "common knowlege" that bypass caps are necessary. The most important thing to consider, as Andrew T alluded to, is the application. In my experience, using bypass caps incorrectly can be worse than using none at all. One popular justification is the supposed inferior high frequency abilities of electroytic caps, not always true. At times they may be beneficial in supressing undesirable high frequency oscillation, but if not implimented properly, might even make it worse. If the circuit is well designed, physical layout done correctly and proper components are utilized, they will generally not be necessary, although there are always exceptions. There is also the unintended consequences of the inherent resonant frequencies of different caps creating problems. So the long-winded short answer is, know what is necessary to get the job done through design analysis, measurement and even experimentation. Guessing, or "standard answers" won't get you there.
Mike
Mike
The HF transient current drawn by an amplifier is fed from the HF decoupling (ceramic or MKT).
The MF transient current drawn by an amplifier is fed from the MF decoupling (electrolytic).
The steady state or LF current drawn by an amplifier is fed from the main smoothing bank.
If the main smoothing bank is located remotely from the amplifier it may be necessary for sound quality reasons to add in some local smoothing at the amplifier. Consider the cable linking the remote smoothing to the local smoothing as a lossy inductor with some resistance.
Your PSU then becomes rCLr"C"L"c (yes, a three stage RCRCRC supply), where r = resistance feeding the first capacitor bank (transformer+wiring), C = first capacitor bank, Lr" = umbilical inductance and resistance, C" = local capacitor bank, L" = PCB trace inductance, c = decoupling.
The MF transient current drawn by an amplifier is fed from the MF decoupling (electrolytic).
The steady state or LF current drawn by an amplifier is fed from the main smoothing bank.
If the main smoothing bank is located remotely from the amplifier it may be necessary for sound quality reasons to add in some local smoothing at the amplifier. Consider the cable linking the remote smoothing to the local smoothing as a lossy inductor with some resistance.
Your PSU then becomes rCLr"C"L"c (yes, a three stage RCRCRC supply), where r = resistance feeding the first capacitor bank (transformer+wiring), C = first capacitor bank, Lr" = umbilical inductance and resistance, C" = local capacitor bank, L" = PCB trace inductance, c = decoupling.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.