[Prev][Next][Index][Thread]
Re: Fw: Fair Radio Sales HV capacitor
"Now if you REALLY want to non-destructively identify your capacitor's
dielectric, you can do this by measuring it's dielectric loss at RF
frequencies. This will tell you whether the dielectric is Mylar (loss
factor of 0.005 - 0.016) or Polypropylene (<0.0002). Determining this
with any precision requires having access to calibrated laboratory
equipment, using a specifically-configured RF impedance bridge (called a
Schering bridge). The measurement technique is described in "Dielectic
Materials and Applications" by Arthur Von Hippel. However, I also
suspect that you can use the much simpler "finger test" - measuring case
temperature rise after relatively brief running times - to determine
virtually the same thing!
Now it is interesting that you didn't see any performance difference
between your rolled PP/oil and this cap! Is it possible that there were
other significant system losses that masked the loss from the commercial
cap? For example, could your home-brew cap have had substantial internal
corona losses? The latter may be why the homebrew unit subsequently
failed."
Bert:
A loss factor as high as 0.016 would still imply a Q of at least 60. I
suspect the losses in the gap are great enough that the loss in the
capacitor doesn't make a great deal of difference. I would think the
type of construction - extended foil or inserted tab - would be at least
as important. Of course, the heating test is what really counts. By
comparing the heating of the capacitor and that of the gap you should
get enough info to figure out what is happening. I think I'll buy a
couple and mess around with them.
Can report on an ancient experiment involving Sprague "Vitamin Q" HV
filter capacitors. Have some 0.05 ufd, 16 kV jobbies which are
relatively small, so didn't expect much success. Their loss is so high
that I can't measure it (at RF) with the bridges I have, and attempts to
measure the parallel resonant impedance with a suitable inductor failed
to reveal said resonance! (At a frequency of around 100 kHz.)
Ed