[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Testing of Geek Group Caps

Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>

Hi All,

I "finally" got around to testing the Geek Group 150 and 100 nFcaps they 
sent too me for dissipation (many thanks to the Geek group and sorry for 
taking so long to do this ;-)).  The test is basically running a known 
current through them at 350kHz for 10 minutes and seeing how hot they 
get.  The maximum safe current rating can be estimated from there.  The 
caps all measure about 0.5% high in value and had greater than 20x10^9 ohms 
of DC resistance at 2000 volts.

The test set up is a 350kHz 900 watt power supply that can drive 4 amps 
through the cap in question.


The test cap is put in a tube that helps keep air flow off it to maintain a 
stable ambient temperature.


A thermal couple probe and Pearson current monitor record the temperature 
and current of the cap.


Load resistors and an MMC dump the power.  The MMC is set to about 2nF to 
counteract the ~120uH of inductance in the wire wound resistors.


The scope and power supply are shown here.  The scope monitors the current:


I found a big foam box also helped to keep the temperature stable during 
the test:


First I tested the 100nF caps at 4 amps, 346kHz, for 10 minutes.  For the 
first cap, the temperature rise was 1.07C.  The second cap gave 1.24C.  If 
we set a limit of a 10C rise, then we can predict that the caps would 
handle 12.22 and 11.36 amps RMS at 350kHz.  This follows well with the 
specified 12.1 amps in the data sheets.


I then tested a new sample of the 150nF cap and got a 0.690C temperature 
rise which suggests a current ability of 15.23 amps which somewhat exceeds 
the specs.

This can be contrasted with my measurements at:


Where an similar older cap heated 2.6C at 3.5 amps!

So it would appear that Cornell Dubilier "figured something out" since the 
internal cap heating per a given current is now dramatically less!!  They 
now seem to be able to take ~3 times the current of older versions!!!

I took one of the 100nF caps and ran the voltage up to breakdown:


The first snap was at about 4400 volts and the second at 6500 volts.  At 
6500 volts after the breakdown, the cap resistance dropped to 2 x 10^9 ohms 
at 6500 volts.  So breaking down at 3X the rating did a "little too much" 
internal damage ;-))  But at 2000 volts the resistance was again above the 
20 x 10^9 ohm limit of the Bertan 225.

I unrolled the cap and found that the two pops occurred in the ribs just 
like they are supposed to.  Not much real damage at all.  I did note that 
the internal dielectric layer is very thick at 0.0005 inches:


But the rest of the cap is just as expected.


The deep internal layers showed no signs of thermal damage.

So it looks like the Cornell Dubilier caps just keep getting better!!  Not 
sure how they got the dissipation down...