Amp sim output load capacitance, how much?

[cbdb]

Whats a good capacitance to put accros the load resistor (8R) when simming an amp? I am not using a Zobel network in the sim yet, but am still trying to get stability. I know this has been discussed before but my searches didnt find anything. I have been using 10nf and now when i change it to 100nf I have an oscillator. If 10nf is too small I need to go back to the drawing board! Thanks in advance.

[Mr Evil]

2uF is often seen, because it is supposed to be similar to an electrostatic speaker. D. Self suggests that 100nF is a more difficult load, so it's probably a good value to test with if you want absolute stability. An output inductor helps a lot with this.

[cbdb]

Did you mean 2 nf?

[Mr Evil]

No, definitely 2uF. I wouldn't worry if your amp isn't stable into that, because it's not a normal load. Even 100nF is very high, but I think it's worthwhile making sure an amp is as well behaved as possible into difficult loads.

[jcx]

just Google for wire/cable capacitance pF/ft - 10-50 pF/ft seems to cover all but the most exotic low inductance cable designs

the 2 uF direct load is an absurd standard - leakage inductance and series resistance of the ESL step up transformer would prevent the amplifier from seeing a pure C load of that magnitude

[kaos]

Piezo tweeters are typically around .2 to .3 uF, so someone hooking one of those up to the amp is kind of nasty, especially if there’s no R load to help dampen things out. Electrostats are sometimes modeled as 8 ohms paralleled with 2 uF (don’t ask me why, as someone already commented, real electrostats probably look quite a bit different). My current amp prototype shows the greatest over-shoot and ring around 20 nF, less than that the amp drives well by itself, and over that the output inductor progressively helps more and more.

[Bob Cordell]

Originally posted by cbdb
Whats a good capacitance to put accros the load resistor (8R) when simming an amp? I am not using a Zobel network in the sim yet, but am still trying to get stability. I know this has been discussed before but my searches didnt find anything. I have been using 10nf and now when i change it to 100nf I have an oscillator. If 10nf is too small I need to go back to the drawing board! Thanks in advance.

It's good to test the amp with a wide variety of capacitance loads, but don't expect it to be stable under all conditions without the use of a series L-R network. Depending on the amplifier details, you can go as low as 1 uH in parallel with 2 ohms or so. When probing for stability of the amplifier itself, probe on the amplifier side of the L-R network.

Most people will agree that a series inductance as low as 1 uH is sonically transparent. Others think that values up to perhaps 5 uH will not degrade the sound. In any case, be generous in constructing the coil and make it air core.

The Zobel network (series R-C shunting to ground), is usually needed to stabilize the amplifier when it sees an inductive load or no load at all.

If you want to go as high as a 2 uF load test, it is reasonable to put 1 or 2 ohms in series with it.

Cheers,
Bob

[wg_ski]

2 uF is -j4 ohms at 20k. If you really wanted to make sure your amp is stable into and can drive a fully reactive load, that would be the test. If you were going to be driving electrostats or banks of piezos it might be good to check. Probably a 10kHz test (-j8 ohms) is probably sufficient for even extreme loads, however.

[cbdb]

Thanks for the answers. One more question, would I be correct in assuming most speaker/cable combinations are an inductive load? (Not including electrostatics, or pizos). So do people test stability with capacitive loads because they are a more difficult test? (I guess if you have a long high capacitance cable that isnt connected on the speaker end you dont want the amp to go unstable. So much to worry about! Oh well, I guess thats what makes this stuff so interesting.)

[cbdb]

Just tried a series R with the load cap, makes a huge differance. Adding .1R lets me go from .01uf to 2 uf. Is this a resonable thing to do ? My phase margin goes fro 70 degrees to 30 degrees (still borderline stable? but the interesting thing, my gain margin increased from 6db to 12db!