Bypassing bias stage with caps

[EchoWars]

I've seen large values such as 100µf used to bypass the bias trimpot and diodes (basically joining the bases of the next stage, usually the drivers or pre-drivers), but I've also seen very small values used, such a .01µf.

Certainly it depends on the input impedence of the next stage as to what ought to be used, but the low value above seems real undersized, and 100µf seems to be real overkill. Voltage seems not to be a real factor, as there is unlikely to be must more than 3-4 volts across the cap.

What do you all do?

[markp]

I use 10uf accross the base-emitter and .1uf between the C and E.

[EchoWars]

OK...how about a visual?

This is about the most common bias arrangement I run into. Bypassing the DC bias arrangement would mean a cap from base of Q9 to base of Q10. In a scenerio like this, what would you use?

[markp]

Sorry, I use a bias transistor not just varistors or simple diodes.

[peranders]

Originally posted by EchoWars
...., but the low value above seems real undersized, and 100µf seems to be real overkill. Voltage seems not to be a real factor, as there is unlikely to be must more than 3-4 volts across the cap.

100 nF is an average (=normal) value but the value isn't critical and if you really want to determine the value I think you must have a distorsion analyzer. I would have settled for 100 nF to start with and 100 uF seems overkill.

To be true I have never investigated how much I need in this position so I gather everything from 1-100 nF will do probably.

[EchoWars]

Well, I cracked open a few service manuals to see what they did. Sansui uses a lot of 1µf electrolytics for this. The one Kenwood that bothered to bypass had a 0.47µf film cap. Several Marantz amps I have manuals for use 0.1µf caps, and a largish Sansui BA-F1 has a 3.3µf electrolytic in parallel with a 0.047µf film cap. Seems to run the gamut.

On the circuit above, I stuck in a 0.56µf 50V stacked film cap (since that's what I got laying around). Seems to have made little to no difference in distortion measurements, but also hasn't seemed to hurt anything. I'm inclined just to leave it there.

Maybe at higher frequencies it might actually do some good?

(BTW peranders, I always appreciate your replies...seems when no one else is interested, you always have something to say. Thanks.)