Bypassing PSU caps

[pro_crip]

I understand the reasons behind bypassing said caps, what I don't understand is how to implement it. Do I pick a certain % of the bypassed cap or do I experiment with a range of values? The bypass cap gets wired in parallel with the bypassed cap, but excuse the brain fart, does this mean I wire the bypass across the terminals of the PSU cap? If you saw my medicine cabinet you'd understand the brain fart. I really want to try some of them thar kommie surplus teflons. Thanks a bunch.

Rich

[Grainger49]

It seems to be Black Magic. 

Public Warning:  I have read scientific reports that say that the wrong cap can muck up the whole deal. 

On the other hand, which is where I sit, I have always been happy with the bypasses in the power supply that I have made.  The old wet thumb estimate was to bypass a 100uF with a better 10uF bypassed with an even better 1uF... well, you get the picture.  The rule was 10%.  I have bypassed 220uF with 2.2uF Obbligato caps, then with a 0.1uF KK Teflon.  There is no mathematical progression, neither linear nor the golden ratio.  In any application, audio path or power supply, bypass is a parallel path, yes + to + and - to - of the existing capacitor. 

P.S. Excellent East Tennessean there with "Them thar..."

[pro_crip]

The jewish boy from queens thanks you as I munch on a corned beef on rye.

Rich

[Paul Joppa]

Technically speaking, the risk that I can see with the wrong value is that you might introduce a resonance with the equivalent series inductance of the electrolytic cap, which is not sufficiently damped by the equivalent series resistance. To resolve this you need to know the ESR and ESL of the cap being bypassed. But ESR will increase as the cap ages, I think I've seen an estimate of four times as much at the end of the cap's life. I don't know if the ESL also drifts.

I have occasionally thought about putting an RF choke in series with the electrolytic, with resistance and inductance large enough to swamp the cap's shortcomings and so allow a precise calculation for the bypass cap. But in the end, I've always just gon looking for a cap rated to perform well to above 100kHz and used that, unbypassed.