[Collins] 75A-4 Recap
Dr. Gerald N. Johnson
geraldj at ispwest.com
Tue May 23 16:20:59 EDT 2006
On Tue, 2006-05-23 at 01:35 +0000, Eugene Hertz wrote:
> On the subject of recapping and testing. I have a basic question.
> I have a cap tester that will show leakage in uA for caps tested
> up to 600v (sencore lc53). I have heard that ESR is what one
> should be looking for, somehow its a more accurate or telling measurement.
>
My test with power supply and VTVM shows full scale for 0.15 microamps
leakage. Is yours that sensitive? Just one microamp leakage from audio
driver to audio output grid with a 1 megohm grid resistor (uncommon, but
possible, more commonly 1/2 megohm) shifts the bias one volt. On a tube
with 30 volts grid bias that's not much but on a tube with only 3 volts
grid bias, that's unacceptable unless you have a bucket of tubes and
output transformers because that bias shift will move the operating
point of the tube to more plate current than needed. So the parts run
hot.
> My question is, can't leakage and ESR be equated? For example R=V/I.
> Couldn't ESR be somehow derived from the leakage? Could ESR=600v/uA ?
Leakage and ESR are different phenomena. Bridges sometimes try to equate
the two as a function of capacitor Q or dissipation factor. Dissipation
factor is 1 / Q. And in an AC bridge circuit with dissipation factor
that might show relative equivalence. The ideal bridge would have three
balance adjusments, one for series resistance, one for capacitance, and
one for leakage resistance. The bridges in common use (yesteryear) had
only two balance adjustments allowing the series resistance arm to be
the total Q adjustment. Even then there can be interaction between the
two making the bridge null hard to reach. I'm not sure one could
separate series and shunt resistance (and that shunt equivalent resistor
might need a range into gigohms which is hard to do with a high quality
precision wire wound rheostat) which may be why bridges tend to go for
equivalent series or parallel resistance only.
Trouble is sometimes DC leakage is voltage sensitive and the bridge
operating voltage is probably only a volt or two on the capacitor while
the DC leakage may get significant only above 150 volts (or some
threshold). DC leakage need not be linear either, it can have a
threshold and essentially be a breakdown phenomena, or arcing in the
dielectric.
On the other hand, ESR is the series resistance of the capacitor. A
capacitor can be leaky but have low series resistance. In some circuits
(like that grid coupling capacitor above) a bit of series resistance is
of no consequence, like what's a 1k series resistance feeding a half meg
load? A voltage drop from 500 parts to 499 parts. On the other hand in a
by passing circuit or a power supply filter circuit, a 1K series
resistance allows an AC current on the supply line to generate a volt of
ripple for each milliamp of ripple current. In the filter capacitor
situation a few ohms of series resistance is a sign of a failed
capacitor while the equivalent as a leakage resistance based on
capacitor Q or dissipation factor would not have a significant effect on
radio operation.
A capacitor truly has both a series resistance and a parallel or leakage
resistance. The third element of the capacitor is the capacitance. They
can be lumped together but as I've said, have different circuit effects.
For convenience some AC bridges lump them together measuring only the
equivalent of one or the other and some bridges aren't adequately
sensitive to leakage for my tastes for audio coupling capacitors and AGC
bypass and time constant capacitors.
A bit of leakage in screen bypass capacitors when in every capacitor in
the radio tends to make the total gain of the radio a bit low and the
performance a bit punk which gets worse with time as the added heat from
the extra current in the screen dropping resistor tends to make it drift
high in value while the added heat in the bypass capacitor makes it leak
more.
Leakage comes from defects in the dielectric media. Series resistance
comes from poor connections to the metal plates (or the electrolyte in
an electrolytic capacitor). They can't be equated.
>
> Anyone shed some light on this subject?
> thanks
> Eugene
>
>
--
73, Jerry, K0CQ, Technical Advisor to the CRA
All content copyright Dr. Gerald N. Johnson, electrical engineer
More information about the Collins
mailing list