Author Topic: K72P-6 ultra low DA, DF capacitors  (Read 2814 times)

0 Members and 1 Guest are viewing this topic.

Offline shodan@micronTopic starter

  • Contributor
  • Posts: 18
  • Country: ru
  • Deleted account
K72P-6 ultra low DA, DF capacitors
« on: June 12, 2021, 01:29:46 am »
void
« Last Edit: July 14, 2021, 05:15:36 pm by shodan@micron »
Deleted account
 
The following users thanked this post: Echo88, Vgkid, doktor pyta, Kosmic

Online bsw_m

  • Regular Contributor
  • *
  • Posts: 241
  • Country: by
Re: K72P-6 ultra low DA, DF capacitors
« Reply #1 on: June 12, 2021, 12:28:52 pm »
Some time ago I measured the leakage current and the dielectric absorption coefficient of the K72P-6 22nF 200V capacitor.
The leakage was about 8fA at a test voltage of 50V.
Dielectric absorption was below 0.02%
 

Online bsw_m

  • Regular Contributor
  • *
  • Posts: 241
  • Country: by
Re: K72P-6 ultra low DA, DF capacitors
« Reply #2 on: June 12, 2021, 12:38:37 pm »
Now I decided to measure the leakage of a capacitor accidentally taken from the box at nominal voltage.
It turned out to be K72P-6 56nF 200V.
Since I am going to see a leakage <1E-12A, a 1TΩ KVM resistor was connected in series with the capacitor to avoid noise gain amplification. The capacitor body has been grounded to avoid leakage through the insulators. I will publish the results after the readings have been established. Please consider the time constant of this RC circuit. For shortening the experiment time, before measurement capacitor was charged to nominal voltage by shortening series resistor.
« Last Edit: June 12, 2021, 01:02:03 pm by bsw_m »
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 7934
  • Country: us
  • Retired, now restoring antique test equipment
Re: K72P-6 ultra low DA, DF capacitors
« Reply #3 on: June 12, 2021, 03:37:05 pm »
How would you measure the dielectric absorption (DA) on capacitors of this quality?  In audio circles, they are highly regarded.
 

Online bsw_m

  • Regular Contributor
  • *
  • Posts: 241
  • Country: by
Re: K72P-6 ultra low DA, DF capacitors
« Reply #4 on: June 12, 2021, 06:01:03 pm »
Dear friend BSW_M, i known, you have bag of K72P-6 magic, please show DF results with yours TESLA meter. It may be much-much more dramatic.  ;D
Took a random capacitor and measured it with my Tesla BM595.


I also decided to publish preliminary data on the measurement of the leakage current. As you can see, the readings are still not established, but the leakage current is about 1pA.
« Last Edit: June 12, 2021, 06:15:32 pm by bsw_m »
 
The following users thanked this post: shodan@micron

Online bsw_m

  • Regular Contributor
  • *
  • Posts: 241
  • Country: by
Re: K72P-6 ultra low DA, DF capacitors
« Reply #5 on: June 12, 2021, 09:31:25 pm »
I ended up measuring the leakage current of the 56nF 200V K72P-6 capacitor at nominal voltage (200V).
This test took approximately 9 hours.
« Last Edit: June 12, 2021, 10:16:30 pm by bsw_m »
 
The following users thanked this post: shodan@micron

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14067
  • Country: de
Re: K72P-6 ultra low DA, DF capacitors
« Reply #6 on: June 12, 2021, 10:44:36 pm »
I'm make TSP software right now for my K2450 to repeat ГОСТ 28885-90(GOST 28885-90) method

I'll try run simple TSP script link on K2450, it seems to be works!
Preparation:
* Short leads at 4 hours.
* Connect DUT to K2450 trough rear triaxials.
* Remove short.
* Run script.

Code:
* Short capacitors leads at 30 min (source 0V with 1A limit),
* Charge capacitor to 200V and hold 5m (source +200V with 100mA limit).
* Discharge capacitor to 0V and hold 5s (source 0V with 1A limit).
* Release leads at 3m (HighZ mode).
* Measure Voltage.
* Calculation.


So the russian standard for measuring DA is slightly different from the US one. In the US they charge very long, discharge for some 10-15 seconds and measure after some 15 minutes. So the procedures are similar the numbers are not directly comparable. Expect a slightly larger DA number from the US test.

To get more information out from the test, one can do a very short (e.g. 1 s)  discharge phase and than contineously measure the voltage rise. subtracting the voltage at the beginning is the same as doing a longer discharge.
This at least works with a good voltmeter that does not add much bias. In the old times one may had to disconnect the meter for the waiting time. If the capacitor is small it may help to repeat the test with reversed polarity.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14067
  • Country: de
Re: K72P-6 ultra low DA, DF capacitors
« Reply #7 on: June 13, 2021, 08:29:03 am »
The US test standard (MIL-PRF-19978)  calls for 1 hour charge, 10 seconds discharge and than looking for the voltage after 15 min (or the maximum if reached earlier in a more leaky capacitor).  The difference to the GOST is about a factor of 2-3 in the time scale and a longer charge time at the beginning.
For many capacitors DA tends to go up with the longer time scale and the longer charge time at the start also gives slight higher DA numbers, though not much.

The faster russian test procedure is slightly more convenient, as it does not need input impedance of the meter as high.


With very low absorption it is a good idea to connect the voltmeter only after discharging the capacitor. Otherwise one would include also dielectric absorbtion of the input capacitance of the voltmeter. So there should ideally be a switch (relais) at the voltmeter input. With a slightly different configuration one can use a single switch to short the meters input and do the charging / discharging  at the other side of the capacitor.
One can see the effect of the length of the discharge phase, by using a short discharge and record the data from the start. Except for very high DA caps the recovered voltage just add up. So instead of a longer discharge, one can use the voltage at that time as the new "zero".  So the information for a longer discharge time is already in the curve.
In this repect the US way with a long charge time is more universal. With a long charge time a single measured curve could cover different time scales in a single test.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf