Power supply bypass cap values.

[Paul Joppa]

There are no reliable "rules" about bypassing power supply caps. Sometimes they help, sometimes the hinder, and sometimes you can't tell the difference. The "1% rule" is really more like "something between 0.1% and 10%" and even then exceptions are not at all uncommon.

The only reliably useful answer is "try it and see!"

I could go into technical explanations of what can work and what can cause problems, but frankly I think there are so many effects we DON'T understand that it would be deceptive to talk of the few we DO. Just try it.  :^)

[adamct]

Cool. I'll try them in both positions and see how it goes.

[earwaxxer]

I have a 2.2uf Mundorf S/O 'bypassing' a 10uf P/O in my speaker crossover.

That is pretty significant in a crossover...

True.. I probably shouldnt call it a 'bypass' because I was combining the values on purpose to reach the 12-15uf or so that I was looking for. I perceived that there was a benefit beyond just lowering the crossover point for the ribbon tweeter. So, when I combined them I was hoping I could kill two birds with one stone (bypass benefit, and significantly lower the crossover frequency). That seems to be the case, lucky for me!

[adamct]

If we consider the RC time constant of the power supply:

We have 680 Ohms into 100uF, we have a time constant of 0.068S.

If we look at 8.2K into 22uF, we have a time constant of 0.1804S. 

A higher time constant will really shave down the noise, but at the expense of potentially dropping voltage or paying a lot for caps.  (I shoot for 0.1S under most circumstances)

-PB

Hmmmm. This is very helpful in understanding how you choose the values you use. Voltage affects volume, is that right? And why does increasing the time constant reduce noise? Why wouldn't it affect noise and the music signal equally? Other than cost of caps and reduced voltage, does increasing capacitance have negative effects on the sound? (I'm no longer trying to figure out what caps to install, just trying to understand how this stuff works.)

Best regards,
Adam

[corndog71]

I have a 2.2uf Mundorf S/O 'bypassing' a 10uf P/O in my speaker crossover.

That is pretty significant in a crossover...

True.. I probably shouldnt call it a 'bypass' because I was combining the values on purpose to reach the 12-15uf or so that I was looking for. I perceived that there was a benefit beyond just lowering the crossover point for the ribbon tweeter. So, when I combined them I was hoping I could kill two birds with one stone (bypass benefit, and significantly lower the crossover frequency). That seems to be the case, lucky for me!

Hey Eric,
When combining caps in a speaker crossover you should try to balance the caps equally.  For instance, if you're shooting for 12uF then you should use a pair of 6uf caps and if you still want a bypass then use a .1uf.  This is the technique used by a speaker designer I follow. 

Combining a 10uf with a 2uf will cause smearing of detail.

[earwaxxer]

(combine equal value caps in a crossover)  - That does seem to be the 'conventional wisdom', but I have found that is not the case. I have experimented with that as well. So much for the status quo!

[Paul Birkeland]

Hmmmm. This is very helpful in understanding how you choose the values you use. Voltage affects volume, is that right? And why does increasing the time constant reduce noise? Why wouldn't it affect noise and the music signal equally? Other than cost of caps and reduced voltage, does increasing capacitance have negative effects on the sound? (I'm no longer trying to figure out what caps to install, just trying to understand how this stuff works.)

I'm not quite sure what you mean by voltage influencing volume.  A capacitor of the same capacitance will be larger in physical volume as the voltage increases. 

To understand the time constant, consider the noise to be a 60Hz signal, and that the RC filter is a low pass filter.  As the time constant goes up, the effective level of the filtering also increases.  If you go down far enough, you may not attenuate the 60Hz signal at all!

[johnsonad]

Thanks PB and PJ for helping me to understand this. Madisound has ClarityCap SA's on sale. I picked up a pair of 18uF SA's for my S.E.X. with C4S. The time constant is still high at 0.148S. Now to see if I can shoe horn them in somehow.

[adamct]

Paul,

Maybe I misunderstood. When you wrote "A higher time constant will really shave down the noise, but at the expense of potentially dropping voltage or paying a lot for caps", I assumed the higher time constant was due to a larger capacitor, and that the downsides were reduced voltage and the higher cost of the larger caps. And then I was trying to figure out what the implications of lower voltage would be. My understanding is that lower voltage results in reduced volume, all else being equal. But maybe I misunderstood your statement, and a larger cap does not reduce voltage, you were just saying that if you keep the cap the same size and want to reduce the noise, then you have to reduce the voltage. Is that right?

And do I understand correctly that the noise is coming primarily from the power supply, which is why it is 60Hz?

And my final question is: how/why does the cap act as a low-pass filter?

Best regards,
Adam

[Paul Birkeland]

Ah, you would lose voltage by increasing the R value, not the C value.

Yes, the noise is primarily 60Hz or 120Hz from the power supply, depending on the design.

If you distill your model of a capacitor down to the resistive component, your RC filter simplifies to a voltage divider.  The actual impedance to across a capacitor depends greatly on the frequency of the AC voltage, and on the capacitor itself (construction, temperature, lifetime, etc.).  This value, therefore, has its own separate name (capacitive reactance). 

If we look at a good quality electrolytic cap, the ESR at 120Hz may be around 1 Ohm.  Having a 270 Ohm resistor in series with a ~1 Ohm resistor is a giant reduction in the AC signal, but better yet would be a 500 Ohm resistor in series with the same cap (at the expense of heat and voltage loss).

Some good reading can be found here, under Low Pass Filter. 

-PB

[adamct]

Thanks for the link, Paul. I'm reading a lot, but most of it has yet to sink in. I need to read it about 5 more times and keep experimenting with simple circuits. I appreciate your patience and generosity in taking the time to answer these questions.

Best regards,
Adam

[adamct]

The only caps I haven't replaced or bypassed in my S.E.X. are the following:

A --- 2x 100uf 450V
B --- 2x 220uf 250V
C --- 2x 22uf 450V (currently bypassed with .22uf film caps, but see below)

With regard to (A) and (B), I initially didn't plan on making any changes to these caps, since they aren't directly in the signal path and weren't recommended changes. But I recently replaced all 3 220uf 250V electrolytics in the power supply of my Speedball'd Crack with 200uf metalized polypropylene film caps. I can't explain it, but I feel like the change is distinctly audible, much bigger than I ever would have expected. Frankly, I was really sort of shocked, since I wasn't expecting much of an audible difference. Bass seems much stronger and resolution (if possible) seemed to get even better than it was before.  Best of all, the cost was surprisingly low.

So I am now revisiting my decision to leave the 100uf and 220uf electrolytics listed above for (A) and (B) in. I've been considering replacing the 100uf electrolytics with oil-filled 100uf motor run caps, and replacing the 220uf electrolytics with 200uf metalized polypropylene film caps. I would have to mount the caps externally and run fly leads to them. The voltage ratings would be equal to (or exceed) the voltage ratings on the stock electrolytics.

My questions are as follows:

1. Putting cost aside (again, the cost is fairly minimal), is there any reason not to do this? Is there a chance the fly leads will cause problems?

2. With regard to (A), is there any reason why oil-filled motor run caps are a bad idea in this position?

3. With regard to (B) is there any reason why films caps are a bad idea in this position?

4. Again with regard to (B), are 200uf film caps sufficient, or should I use 250uf caps instead? The price difference is negligible. Or do the caps really need to be exactly 220uf?

5. With regard to (C), rather than just bypassing them (as is currently the case), I plan on replacing them with 20uf 630V film caps. Is 20uf acceptable or is the 22uf crucial?

6. Again with regard to (C), should I couple the 20uf caps with 2.2uf caps wired in parallel to get closer to 22uf in total capacitance? (I already have all of these caps, so extra expense is not an issue.)

Best regards,
Adam

[Paul Joppa]

[see my reply in the Smack thread as well]

Electrolytics are usually +/- 20%, sometimes even more. So a film within that range is fine.

[adamct]

Thanks. Out of curiosity, with regard to the 220uf caps in the S.E.X., is there any theoretical benefit to using 250uf caps instead of 200uf? Is the S.E.X. shunt regulated?

Best regards,
Adam

P.S. For the convenience of anyone looking for Paul's reply in the Smack thread, see here.

[adamct]

...with regard to the 220uf caps in the S.E.X., is there any theoretical benefit to using 250uf caps instead of 200uf?

Anyone have any thoughts on the above? I can get 200uf, 250uf or 300uf caps for the exact same price, so I'm trying to figure out what I should go for.

Thanks,
Adam