Cathode bypass capacitors: Electrolytic vs. film?

[bob p]

chuck, you mentioned how to build the shelf in other locations ... and all of the time I was thinking of going in the opposite direction -- that the shelf is the basis for the noise problem. if I'm going to build a high pass filter in the midrange area to roll-off bass response, then i don't want a shelf that will allow the hum to persist a few dB lower at the level defined by Rk; comparing a 1600 Hz knee to 50 Hz noise, they're 5 octaves apart. I'd rather have a constant -6db/octave slope across that 5 octave gap. I get more hum rejection that way.

i thought that was implied in my earlier post but the kudos go to octal for actually spelling it out.

[Chuck H]

chuck, you mentioned how to build the shelf in other locations ... and all of the time I was thinking of going in the opposite direction -- that the shelf is the basis for the noise problem. if I'm going to build a high pass filter in the midrange area to roll-off bass response, then i don't want a shelf that will allow the hum to persist a few dB lower at the level defined by Rk; i'd rather have a constant -6db/octave HP filter to roll off hum that's 5 octaves away. i get more dB of rejection that way.

i thought that was implied in my earlier post but octal actually spelled it out.

Maybe I explained myself badly (or interpreted your post badly?)... My suggestion was to use the shelf filter with a fully bypassed cathode. The shelf filter would be added to simulate the partially bypassed cathode. The end result should be that any hum from the cathode is now shunted via the bypass cap and the shelf filter simulates the original intended frequency response. I agree that a -6dB/octave would effectively roll off more hum than the shelf filter but there shouldn't be any significant hum (from the filament circuit anyway) with the cathode fully bypassed. Simply lowering the coupling cap value will not produce the same signal frequency curve as a partially bypassed cathode, but the shelf filter will and we've already eliminated the hum.

EDIT: Then again... There's no free lunch. Adding significant series resistance, as might be necessary when adding a shelf filter if you want to maintain the original circuit impedance, can increase hiss. There's are other possibilities that won't add so much series resistance, but DO change the circuit impedance, for better or worse. I just start with an ideal and try to make it work until I run out of chops. Then I come here for new ones

[trobbins]

Do you have a reference to the second plot, or a clarification on the voltage rating of the various cap types being compared? The label for the Y axis is a little ambiguous at first, as it looks like the division of Z by ESR. It looks like they are comparing low voltage caps of circa 10-16V rating.

Not many manufacturers show ESR curves, and if they do then not for low uF values in a particular model range. I was just looking at United Chemicon SXE datasheet, and it indicates a 100kHz ESR for 10uF 25V 25C of 5 ohm. ESR level for an alu electrolytic will fall significantly with a higher value cap of same voltage rating, or a higher voltage cap of same value, or a warmer cap.

[nickb]

Do you have a reference to the second plot, or a clarification on the voltage rating of the various cap types being compared? The label for the Y axis is a little ambiguous at first, as it looks like the division of Z by ESR. It looks like they are comparing low voltage caps of circa 10-16V rating.

Not many manufacturers show ESR curves, and if they do then not for low uF values in a particular model range. I was just looking at United Chemicon SXE datasheet, and it indicates a 100kHz ESR for 10uF 25V 25C of 5 ohm. ESR level for an alu electrolytic will fall significantly with a higher value cap of same voltage rating, or a higher voltage cap of same value, or a warmer cap.

It could be clearer...

The solid line is |z| and the dashed one is ESR as I understood it.

[Old Tele man]

It could be clearer...

The solid line is |z| and the dashed one is ESR as I understood it.

Yes, that's typically how the industry data books display them.