[R-390] Megger and Capacitor Measures ... Good vs. Bad ?

[Alan Victor]

On Sat, Oct 25, 2014 at 5:15 AM, Charles Steinmetz <csteinmetz at yandex.com>
wrote:

> Alan wrote:
>
> After looking at a few known good caps, it became
>> apparent that a baseline for good versus bad would be useful. As it turns
>> out, this work was already done very nicely back in 1955 by a couple of
>> folks at the Diamond Ordnance Fuze Labs in Washington, DC. They plot the
>> MEGOHM x MICROFARD product for a variety of caps from the time, namely
>> mica, ceramic, paper, glass, and some poly plastics. At room temp, all of
>> these caps fall in the range of 4-6.2 (meg x uf ) product! Hence, a 47 uF
>> looks like about 100k ohm while a .01 uF should look like 400 -500 meg
>> ohms
>>
>
> Some data points (everything below assumes that the capacitor is being
> measured at or below its DC voltage rating):
>
> At 200v, an apparent leakage resistance of 100k represents a leakage
> current of 2mA.  What kind of dielectric did the authors test at 47 uF?  An
> ordinary aluminum electrolytic of that value (not even a low-leakage type),
> which should be the leakiest capacitor of that value you can find, is
> specified at around 250uA maximum leakage at room temperature (>35 megohm x
> uF).  Any plastic film capacitor is specified for leakage much, much lower
> than that (for example, WIMA FKP3 metallized PP caps are specified at
> 500,000 megohms minimum, while Series 225 Orange Drops are specified at
> 25,000 megohm x uF product minimum -- 5,000 times better than the spec you
> quote).  Typical ceramics are specified at ~10,000 megohms.
>
> 5 megohms x uF sounds awfully low to me, even for 1955.  It's also *very*
> suspicious that caps made for tuned RF circuits (glass, mica, ceramic),
> which need very low leakage to deliver high Q, didn't score much higher
> than caps intended just to block or bypass DC (paper, plastic).  I haven't
> read the study (do you have a link?), but I'm suspicious that there may
> have been systemic measurement errors.
>
> In any case, it is my belief that a capacitor checker or megger is
> entirely unnecessary for working on tube radios.  Indeed, I'd go so far as
> to say it's usually counterproductive because it is very often used way too
> early in the troubleshooting process and focuses the tech on individual
> parts when (s)he should be keeping an open mind and looking at the circuit
> as a whole.
>
> Note that most of the suspect caps in tube radios are (i) the main filter
> capacitors (electrolytic in all but the oldest BAs) and (ii) paper bypass
> caps on power supply lines and tube cathodes.  Bad bypass caps can almost
> always be found easily with a VTVM, since there are invariably decoupling
> resistors between the raw power bus and the bypass caps -- leaky caps will
> cause the local B+ at each bad cap to be low.  (If the radio no longer
> works (blows fuses or smokes), then any leaky B+ bypass caps can easily be
> found with an ohmmeter when the radio is unplugged -- they will generally
> read less than 1k to ground.)
>
>
> Best regards,
>
> Charles
>
>
>
>
> ______________________________________________________________
> R-390 mailing list
> Home: http://mailman.qth.net/mailman/listinfo/r-390
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:R-390 at mailman.qth.net
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>

Hi Craig, Charles and all.

Thanks for the inputs and information. It was not my intention to bring a
dead
horse back to life. In any case, I agree Charles on the troubleshooting
technique
you raise, while Craig raises a good point as well. However, if I find an
issue
with a circuit and circuit theory points to the likelyhood its this cap, I
would like to test it after removal with some level of confidence. If the
measure
says it's OK, as Craig highlighted, put a new cap in its place anyway.
However, I know
I have NOT found the problem.

Charles, on the numbers, I believe I mis-interpreted the plot scale. The
y-axis in this
paper reads LOG MEGOHMS-MICROFARD PRODUCT. As all the caps investigated in
this work had leakage
R values in the 10^10 to 10^15 ohm range, my error. So the plot figure 9 y
axis values are probably
10^4 to 10^6 (MEG-OHM x MF) in range. They reported mica and ceramic at
10^4, paper at 10^5 and the plastic
polystyrene etc... in the 10^5 to 10^6 range. No, they did not measure
large electrolytics.
There C values for measurement were 1000pF through .033uF. Applied V at
200volts. So the
47 uF I mentioned earlier should be ~ 85 M-ohms. Need to revisit what the
megger is reporting.
Incidently, a diode should be placed in series off the megger to the C
under test. If
you can't crank the generator with any constant rate, the C discharge back
through the
megger makes it hard to get a good reading. Charles, I am not in love with
the megger, but
seems like a reasonable tool for the job. If a cap is really bad, a simple
VOM can pick it
out. Its this borderline cases and the units that measure A-OK on an RX
bridge that are
potentially questionable.

The paper and plots were not found on line, but through a local libray
search.
It was published in '55 in the IRE TRansactions.

Thanks!