In article , John Phillips
wrote:
[snip]

I agree that in the case of a coupling capacitor the ideal design drops
a negligible *signal* voltage across it. The article and its test are
of dubious relevance there.

However, consider the case of a single-pole RC filter. In the pass-band
substantially the full signal voltage appears across the capacitor.

Yes. Three cases had occurred to me after continuing to think about this:

1) Something like a Baxandall arrangement when the tone was set well away
from 'flat' sic. Here quite noticable ac levels could appear across some
of the caps without being cancelled in effect by that on another cap.
Although in most control amps in my experience this would be or the order
of a volt or less - far less than the 70v RMs used in the page we were
discussing.

2) 'Miller' shunt caps between the base and collector (or equivalents for
other types of device) as used in some designs across the output devices.
Here the voltage swings seen by the caps could be almost 'rail to rail'.
Hence they could easily be 70v RMS in some cases. However the cap values
would normally be small, and within a loop, and shunted by the device
(whose own capacitance may be far more nonlinear!) I doubt anyone would be
using an electrolytic for this! I've never noticed such a design giving
noticable distortion from this, but it is something I would avoid applying
anyway, so have not investigated it.

3) The C of the series RC of a 'Zobel'. As (2) this could see rail-to-rail
ac. However this should be gripped fairly tightly by the output impedance
of the amp, so although it may ask the power amp to deliver a nonlinear
current through the Zobel, it should have little effect on the output as
seen by the loudspeakers.

There are other examples like the C of an input LP RC roll-off filter to
prevent an amp being driven into slewing. However in most cases like this
I'd expect the level of the ac to be well below 70v RMS.


So I suspect there can be relevance in measuring capacitor non-linearity
with significant AC signal level.

Yes. That was why I was asking in an earlier posting for some examples to
see if there were any. I haven't really encountered any that seem
comparable and relevant, but for all I know they may exist, e.g. in some
valve designs. The above all occurred to me as 'in principle' areas where
an effect might show, but in each case I've never seen/heard anything
significant as a result in practice.

I am not sure if in a valve amplifier this may happen inter-stage with
the sort of signal level the article uses for the test, though. As an
input RC filter the signal voltage across the capacitor will be somewhat
smaller than that used.

Yes. In a twin-rail amp using SS devices I tend to expect caps to only
appear in a few places like an input dc block, or the block on the feedback
to reduce the dc gain. In these places the ac levels seen in normal use
should be generally relatively tiny.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html