Re: Measuring MMCs

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: Mddeming-at-aol-dot-com 


Hi Steve,
     . Any time you measure a capacitor in a circuit, you will get a false 
reading. A single cap with a resistor across it is such an RC circuit. Most 
meters are designed to measure only a single cap with one end disconnected 
from the circuit. Many do this by measuring the reactance of the cap and 
back-calculating C from Xc. Meters that use an HF test signal will br more 
accurate than a LF or DC meter, because as freq. increases, Xc decreases, 
and the bleeder R tends to "look" infinite by comparison.
     An MMC is a large RC network. Trying to get an accurate measurement of 
the network capacitance by looking only across the two outer terminals of 
the network will always be in error. In a TC, while charging the caps, the 
50/60 Hz transformer "sees" both the R and the Xc. When discharging, 
however, the frequency is ~2000 times higher, Xc is 1/2000th of what is was 
during charging, and the Rs "look" infinite to the tank oscillations. Use 
of an HF Sig. Gen. and calibrated scope would give much closer results.

Matt D.


In a message dated 8/8/04 2:06:15 PM Eastern Daylight Time, 
tesla-at-pupman-dot-com writes:
Original poster: "S&JY" <youngsters-at-konnections-dot-net>

To all MMC users,

I always wondered why calculated and measured MMC capacitance differs more
than expected.  So I did some experiments that proved that bleeder resistors
are the culprit.

Using a capacitance meter good for 1% (5,000 count), I measured a typical
tank cap value with various bleeder resistance values across it, and got
these results (R in megohms, C in nanofarads):

    R       C
infinite   29.3 (the actual value)
40.2       32.1
30.1       33.1
20.3       35.4
10.1       45.28

So you can see, bleeders can cause serious errors when measuring MMCs.  No
doubt, the results will vary depending on the type of capacitance meter.

So how does one get rid of the effects of the bleeder resistors when
measuring MMCs?  It appears that the error is caused by DC current flowing
through the bleeders.  So the cure is to put another good (low leakage)
capacitor in series to block DC.  Then calculate the value of the MMC.

As an example, My MMC is 15 paralleled strings of 12 22 nF caps, which
should be 27.5 nF.  My bleeder resistance totals 24.3 meg.  Direct
measurement indicated 30.8 nF, which is wrong (15% high!) because of the
bleeders.

I put a 331 nF cap in series with the MMC, and that combination measured
24.6 nF.  Calculating the MMC capacitance gives 26.7 nF, which is about 3%
low but within the 5% capacitor tolerance and believeable.

(I tried using a 45 mF PFC cap in series with the MMC to avoid the need to
calculate results.  But it didn't work with my meter because the meter
current is a tenth of a microamp and it would take almost forever for the
two caps in series to reach equilibrium).

Bottom line - don't trust capacitance measurements of MMCs that have
bleeders, unless you add a cap in series and calculate the MMC value.

Have others noticed this effect, and how did you solve it?

--Steve Y.