Author Topic: Using Models to Predict Capacitor Leakage Current  (Read 23160 times)

0 Members and 1 Guest are viewing this topic.

Offline joeqsmithTopic starter

  • Super Contributor
  • ***
  • Posts: 11630
  • Country: us
Using Models to Predict Capacitor Leakage Current
« on: May 24, 2018, 04:41:31 pm »
Too bad the other thread was locked.   Moderator's, if you feel this is an out of bounds thread, please feel free to close it. 

I had posted how I was a bit confused about how a test like this would be valid and assumed how such a design would assume the leakage current was linear.   Reading through the comments, the following was from the designer: 

Quote
Leakage is resistance, it's just that simple. We don't need high Voltage to find that issue. This is a very common misconception, due to the lack in understanding of this failure point. The newest tester posted on Patreon is incredibly sensitive to any resistance (Leakage) in components. The trick to understanding this clearly is: When you increase the Voltage across a fixed resistance, that fixed resistance draws more current. This added current draw is what allows the shadow deflection angle to change in an eye tube. (The eye tube is just a current meter) So really, the high voltage is just there to make the instrument more sensitive to the leakage resistance, since that resistance is very high in value.
[/i]

So it does appear they have a simplistic view of what the model would be.    I had linked a few papers the have some information on leakage currents and mentioned a case where some ceramic parts a friend of mine was having problems with were far from linear.   Personally, I would like to continue the discussion to see if others have seen cases where the leakage currents are not linear. 

This paper talks about the effects of ESD on ceramics and their leakage current.   I may try and look of some of the old capacitor catalogs and see if I can find some better information on this subject (assuming the thread stays active).

https://cecas.clemson.edu/cvel/Reports/CVEL-14-065.pdf

« Last Edit: May 25, 2018, 11:12:34 pm by joeqsmith »
 
The following users thanked this post: fpliuzzi, Performa01, egonotto, GEOelectronics

Offline GEOelectronics

  • Regular Contributor
  • *
  • Posts: 145
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #1 on: May 24, 2018, 05:09:08 pm »
I agree with you. If a capacitor is to be used at a certain Voltage, it can only be thoroughly tested at that Voltage. Resistors too.

I do that on a daily basis, my instruments happen  have that ability, up to 1000 V DC.

For capacitors search "Forming Voltage" "Working Voltage"
 Here's a description of a LC-103 high end cap-inductor tester:

Analyzes capacitors for all four failures
out-of-circuit:
– Value from 1 pF to 20F
– Equivalent series resistance
– Leakage with up to 1,000 volts applied
– Dielectric absorption


BEWARE, this is definitely not 1000V at 10 Amps. It must be current limited by design.

Even the old Heathkit cap testers Magic Eye tube tested electrolytics at their working HV, otherwise they won't "form" properly.

Geo>K0FF
« Last Edit: May 24, 2018, 10:34:03 pm by GEOelectronics »
 

Offline PA4TIM

  • Super Contributor
  • ***
  • Posts: 1161
  • Country: nl
  • instruments are like rabbits, they multiply fast
    • PA4TIMs shelter for orphan measurement stuff
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #2 on: May 24, 2018, 10:10:46 pm »
Here an application note from Keithley about cap-leakage testing : http://www.scientific-devices.com.au/pdfs/WeTransfer-NZvJB6Cw/Sensitive%20&%20Semiconductor/Capacitor%20Leakage%20Measurement%20using%20Keithley%20Model%206517A.pdf
They use a sourcing electrometer, I have one from HP, the HP-4329A, that can measure down to 50 fA and 10-1000V

I added 2 more documents
www.pa4tim.nl my collection measurement gear and experiments Also lots of info about network analyse
www.schneiderelectronicsrepair.nl  repair of test and calibration equipment
https://www.youtube.com/user/pa4tim my youtube channel
 


Offline PA4TIM

  • Super Contributor
  • ***
  • Posts: 1161
  • Country: nl
  • instruments are like rabbits, they multiply fast
    • PA4TIMs shelter for orphan measurement stuff
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #4 on: May 25, 2018, 08:37:41 am »
I was pretty sure this document was linked here but I must be wrong. Very interesting, it is about measuring partial discharge in dielectrics. 
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19940028538.pdf
They do some modelling too.

It would be a good thing to have a model of a leaking cap and some leakage simulating device/fixture. This must be possible with the help os an opamp (adjustable current source triggered by the test voltage of the leakage tester), comparator (breakdown trigger) and/or maybe a sort of sparckgap/breakdown "device" maybe the avalanche behavior of semiconductor.

If I understand well this is a form of leakage/breakdown due to imperfections in the material. One of the methods is a hiss test. That made me remember a cap in a Valhalla voltage/current calibrator. I found that by listening with a "stethoscope"  (some flexible festo vaccuum tubing and a Y connector) I heard it ticking. Only above a certain voltage. After I knew  where to look I did the leakage test again on my just finished 2nd leakage tester (I tested it before at my old one, that uses a max 750 uA current up to 365V, the new one can source upto 10mA at almost 600V)  I then noticed smal meter reactions synchronous with the sound when charged to the max circuit voltage.  The voltage on this cap was not fixed when in use. Only in current mode the max voltage was over it (if I remember well)

This could be an interesting way to test leakage using an ultrasone detector.

www.pa4tim.nl my collection measurement gear and experiments Also lots of info about network analyse
www.schneiderelectronicsrepair.nl  repair of test and calibration equipment
https://www.youtube.com/user/pa4tim my youtube channel
 

Offline joeqsmithTopic starter

  • Super Contributor
  • ***
  • Posts: 11630
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #5 on: May 25, 2018, 11:20:41 am »
I was pretty sure this document was linked here but I must be wrong. Very interesting, it is about measuring partial discharge in dielectrics. 

As I am reading this, I get to figure 2 and thought, no way I would hook into that.   Then figure 5 shows the clamp.  I wonder if any test equipment was damaged while putting this paper together.  lol

Page 16, #5   Yep, turn the thing off and ground it before you work on it!  Words to live by!  lol 
#6 is even better.  I just get this feeling that the student learned a few lessons.  Imagine, the TC connected and you not thinking about it and pick up the meter...   One mistake like that, I too would make a note in my paper about it. 

Page 23, I am surprised 5 minutes was enough.  Then again, considering the test, maybe they just didn't care and it is a fairly small chamber. 

Once they get into the results, things get a bit cloudy.  If you are going to take the time to write about not connecting the TC to the high voltage,  maybe they could have added more details to the results. 
Page 29, conclusion.. Wait, what..  You just showed cases where the PD went down with temp, not up.   

Page 30, OK, I give him a pass.  At least he comes out and states it.   Imagine trying to model this.   

Good read.  Thanks for posting this paper. 

Online PA0PBZ

  • Super Contributor
  • ***
  • Posts: 5121
  • Country: nl
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #6 on: May 25, 2018, 11:22:57 am »
Even IF the leakage current was linear (I don't think it is), wouldn't you need the rated voltage to judge if the leakage is too big?
Keyboard error: Press F1 to continue.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #7 on: May 25, 2018, 11:26:54 am »
Even ignoring damaged parts which will suffer all kinds of weird breakdown effects, leakage tests need to be done at the capacitor's rated voltage or higher; low voltage tests will only reveal gross anomalies.

The same applies to semiconductors and especially high voltage ones.  A curve tracer is especially useful for detecting faulty transistors which operate at low voltages with mysterious behaviors that become readily apparent at high voltages.
 

Offline GerryBags

  • Frequent Contributor
  • **
  • Posts: 334
  • Country: gb
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #8 on: May 25, 2018, 11:44:52 am »
Even ignoring damaged parts which will suffer all kinds of weird breakdown effects, leakage tests need to be done at the capacitor's rated voltage or higher; low voltage tests will only reveal gross anomalies.

The same applies to semiconductors and especially high voltage ones.  A curve tracer is especially useful for detecting faulty transistors which operate at low voltages with mysterious behaviors that become readily apparent at high voltages.

Hi David, wouldn't a curve tracer be useful for testing for cap leakage? you'd have control over how slowly the voltage was brought up and would get a clear indication of where the leakage current begins and how much there is, no?
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #9 on: May 25, 2018, 12:15:47 pm »
1) How do you want to know if the leakage current measured at a low voltage is correct or not? On what basis do you base yourself? All manufacturers give the leakage current at the rated voltage in their technical specification.

2) How can you know if the capacitor will support its working voltage if you do not test it with a voltage higher than its working voltage?

What is the purpose of a low voltage test that does not guarantee the correct operation of the capacitor?

3) And for electrolytic capacitors, how do you know if the capacitor needs to be reformed or not?
An electrolytic must be reformed if its breakdown voltage is below its rated voltage. Without applying a sufficient voltage (with current limitation), it is impossible to know it...

Results: when it is powered on, the surge current may be too high, the capacitor heats up and can even explode ....

Doing tests that are useless is for those who have nothing to do with their lives, not for those who work.
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #10 on: May 25, 2018, 12:40:07 pm »
When we talk about HV tester, we often quote the Heathkit IT-11 and IT-28 ... It's a mistake because these devices
1) do not measure the leakage current, they only give a very rough estimate per magic eye
2) they are very dangerous because there is no limitation of current on the range 600V .... One can do much better with the safety level, (see the project of PA4TIM)

An HV tester can also have other applications like testing high voltage zener diodes, gas tubes for regulating high voltage, reforming electrolytics, measuring the breakdown voltage of diodes and transistors ......
« Last Edit: May 25, 2018, 12:49:04 pm by oldway »
 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7694
  • Country: de
  • A qualified hobbyist ;)
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #11 on: May 25, 2018, 01:42:55 pm »
I had posted how I was a bit confused about how a test like this would be valid and assumed how such a design would assume the leakage current was linear.   Reading through the comments, the following was from the designer: 

Quote
Leakage is resistance, it's just that simple. We don't need high Voltage to find that issue. This is a very common misconception, due to the lack in understanding of this failure point. The newest tester posted on Patreon is incredibly sensitive to any resistance (Leakage) in components. The trick to understanding this clearly is: When you increase the Voltage across a fixed resistance, that fixed resistance draws more current. This added current draw is what allows the shadow deflection angle to change in an eye tube. (The eye tube is just a current meter) So really, the high voltage is just there to make the instrument more sensitive to the leakage resistance, since that resistance is very high in value.

One could also argue that an amplifier stage would be able to deliver enough current for the magic eye in case of a sensitive leakage current detection. Tubes have a quite high input impedance. Maybe vendors chose the less sensitive approach to save a tube or two.
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #12 on: May 25, 2018, 01:56:42 pm »
Quote
Maybe vendors chose the less sensitive approach to save a tube or two.
That's right, but not for this reason.

According to my opinion, the true function of the Heathkit IT-11 and IT-28 testers was not to measure the leakage current per se, but the breakdown voltage.

As the leakage current increases considerably when you reach the breakdown voltage, you should not have a device too sensitive ... when the magic eye closes, there is really a high leakage current and you can be sure that the capacitor is really defective if the closing of the magic eye occurs at a voltage below the nominal voltage.
 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7694
  • Country: de
  • A qualified hobbyist ;)
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #13 on: May 25, 2018, 02:00:57 pm »
Even ignoring damaged parts which will suffer all kinds of weird breakdown effects, leakage tests need to be done at the capacitor's rated voltage or higher; low voltage tests will only reveal gross anomalies.

Take a HV electrolytic with a degraded oxide layer for example. If the oxide layer is badly degraded a LV check works. But if it's mildly degraded the LV check might give you an ok while a breakdown would occur at voltage lower than the cap is rated for. This is something which happens with NOS. A LV leakage check is not 100% reliable, it will deliver a few false positives. So we should educate users of the limitations of LV leakage testers.
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #14 on: May 25, 2018, 02:19:24 pm »
Quote
This is something which happens with NOS

This type of problem does not only occur with NOS electrolytic capacitors but also with electrolytic capacitors which have been used for a long time at a voltage well below their nominal voltage. The breakdown voltage decreases, but remains above the operating voltage.

For other capacitors, I have explained in the other thread why they may have a normal leakage current but a reduced breakdown voltage....

I post it again !!!!
« Last Edit: May 25, 2018, 05:15:40 pm by oldway »
 

Offline GEOelectronics

  • Regular Contributor
  • *
  • Posts: 145
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #15 on: May 25, 2018, 02:21:28 pm »
Oldway said: An HV tester can also have other applications like testing high voltage zener diodes, "

My home made Zener tester uses a little over 1000V to test ALL Zeners, even he 5 and 10V ones, but the most use is for higher Voltage Zeners, up to 300V. Over the years of testing literally thousands of 100,200 and 300V Zeners both with leads and surface mount, rarely are any 2 the exact same V, so they must all get hand selected.

You are correct that HV must be used to perform certain tests, and I add that capacitors vary by type radically. One type I test is vacuum capacitors. There is a special tester for this, it uses to 50kV high frequency AC. If the vacuum is intact, no sparks inside. If there is air inside there are sparks. CRTs tested same way.

Leakage is not always "simply resistance" that can be measured with low Voltage.

"Meggers" have been around since the early days for this same reason.

Capacitors that use liquid or paste dielectric are another can of worms.

Geo

edit-30 May 2018 Capacitor illustrations added for clarity.
« Last Edit: May 30, 2018, 04:09:24 pm by GEOelectronics »
 

Offline GEOelectronics

  • Regular Contributor
  • *
  • Posts: 145
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #16 on: May 25, 2018, 02:23:30 pm »
Oldway said "I post it again !!!!"

Please do or link the locked posts, I am new here and missed them

Geo
 

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #17 on: May 25, 2018, 02:24:34 pm »
I post it again, here we are:

Here are some explanations about the need to do the leakage current testing at the rated voltage.

1) For electrolytics, there is a fundamental problem since all electrolytics have a leakage current, this is normal.
The problem is to know if this current is normal or not.
For this, it is first necessary to measure this current, what the "invention" proposed does not and then compare the result of this measurement with the technical specifications of the manufacturer.

These technical specifications for measuring the leakage current are made at rated voltage ...
http://www.rubycon.co.jp/en/catalog/e_pdfs/aluminum/e_twl.pdf

It is therefore necessary to test the electrolytic capacitors at rated voltage.

2) For other types of capacitors, it is essential to check if the breakdown voltage has not decreased.
A capacitor may have a low leakage current at low voltage while having a greatly reduced breakdown voltage.
This is due to a physical characteristic of the insulators: the dielectric strength.
see here to understand what it is:
https://en.wikipedia.org/wiki/Dielectric_strength

You will find that the dielectric strength of polystyrene, for example, is 19.7 and that of air is only 3.

What happens when a capacitor has been punctured by a surge?

The insulation is pierced and some of the insulation has been replaced by air or gas from the spraying of the insulation.

As a result, at this point, the polystyrene is replaced by air and the dielectric strength is then only 3 instead of 19.7.

The breakdown voltage is then reduced by 6.56 times and this is not noticed during the leakage current test at low voltage because the leakage current of air and polystirene are substantially the same.

Thus, testing a condensator at low voltage is dangerous because it let you think your capacitor is good when it is'nt.

When you power on such a capacitor at rated voltage without current limitation, it will blow up and may injure you.
« Last Edit: May 25, 2018, 02:32:26 pm by oldway »
 
The following users thanked this post: GEOelectronics

Offline GEOelectronics

  • Regular Contributor
  • *
  • Posts: 145
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #18 on: May 25, 2018, 02:40:02 pm »
Oldway said:When you power on such a capacitor at rated voltage without current limitation, it will blow up and may injure you."

Amen. A short time ago I accidentally overVoltaged a  LV surface mount high value cap, by no more than 6V and it exploded with shrapnel and smoke and a LOUD bang. Learned that lesson the hard way. Made me an instant convert to safety glasses.

Geo

 

Offline PA4TIM

  • Super Contributor
  • ***
  • Posts: 1161
  • Country: nl
  • instruments are like rabbits, they multiply fast
    • PA4TIMs shelter for orphan measurement stuff
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #19 on: May 25, 2018, 02:58:14 pm »
Maybe change the titel of the topic to "several ways to test capacitor leakage current" or "everything you wanted to know about leakage but were afraid to ask"  :)
I have a curve tracer (tek 576), they are able to test leakage but I never used it for caps.

If I can find some time I shall try to do some linearity tests. I think first try a new cap and go in steps of 5 or 10V up to the WV.  Ideal would be a programmable  voltage source and a logging meter but I do not have those (that is with enough uA resolution) so I have to do it by hand.
To do it according the datasheet I should apply the voltage, wait the stated minutes and then note the current, discharge and do the same but now to a higher voltage.

Then test leakage by charging it, waiting a few minutes and measure the discharge current. and also step up the voltage.
Maybe also between steps measure C and D

If this gives the expected results and both methods agree we can expand it to other types. So we know if the test set up is correct

And then to bad caps. The problem with the latter is that this is  unpredictable because it can be defect is many ways. Is it severe leaking or is it breaking down, is it short ? The problem is to find some usable caps that broke down because that is a very rare failure (in my line of work). And they vary from dead short to breaking down just under the WV. 
www.pa4tim.nl my collection measurement gear and experiments Also lots of info about network analyse
www.schneiderelectronicsrepair.nl  repair of test and calibration equipment
https://www.youtube.com/user/pa4tim my youtube channel
 
The following users thanked this post: GEOelectronics

Offline GEOelectronics

  • Regular Contributor
  • *
  • Posts: 145
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #20 on: May 25, 2018, 03:07:08 pm »
"The problem is to find some usable caps that broke down because that is a very rare failure (in my line of work). And they vary from dead short to breaking down just under the WV.  "

Seems like you just described the life cycle of a wet slug tantalum bypass. And it doesn't matter if they were powered up, never powered up, or new in a box, only age.

Geo>K0FF
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #21 on: May 25, 2018, 04:39:36 pm »
Even ignoring damaged parts which will suffer all kinds of weird breakdown effects, leakage tests need to be done at the capacitor's rated voltage or higher; low voltage tests will only reveal gross anomalies.

The same applies to semiconductors and especially high voltage ones.  A curve tracer is especially useful for detecting faulty transistors which operate at low voltages with mysterious behaviors that become readily apparent at high voltages.

Hi David, wouldn't a curve tracer be useful for testing for cap leakage? you'd have control over how slowly the voltage was brought up and would get a clear indication of where the leakage current begins and how much there is, no?

A good curve tracer can be used that way.  I only have a Tektronix 7CT1N but it has a "leakage test" mode which applies a continuously adjustable constant DC up to 300 volts to the device under test instead of a voltage sweep and measures down to 10 nanoamps/division.  The only reason I would use it is because it is the highest DC voltage tester that I happen to have but 300 volts is not high enough for what I consider a "high voltage" part. 

With capacitors I do not think I could ramp the voltage up slowly enough to see a clean breakpoint because the charging current covers the leakage current.  What might work is to control a constant but low charging current (higher currents for higher values of capacitance unless you like to watch grass grow) and display the voltage ramp.  That might make for an interesting project for someone but I think a sourcemeter or a high voltage power supply and current meter is good enough.

As far as a useful high voltage capacitor and diode tester, I have considered designing something dedicated to measuring leakage over a wide voltage range however when doing repairs, this is of questionable use because if you have to pull a part to test it, it is almost always better to just replace it.  A tester which measures capacitance, leakage, ESR, and dissipation at different DC voltages and frequencies would be more interesting but is a lot more complicated than a standard LCR meter so I am not convinced there is any market for it except at the very high end.  I only just recently had an inspiration about how to protect a standard LCR meter front end from high voltages; most of them are pretty fragile or at least all of the ones I know of are.
 
The following users thanked this post: GerryBags

Offline oldway

  • Super Contributor
  • ***
  • !
  • Posts: 2172
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #22 on: May 25, 2018, 05:13:44 pm »
Quote
The problem is to find some usable caps that broke down because that is a very rare failure (in my line of work).

I'm repairing audio amplifiers and receivers manufactured in the years 70's and I agree with PA4TIM.

- a repairman usually does not keep the defective parts to avoid mixing them with the good ones.

 - a lot of defective capacitors can be identified by a simple visual inspection or a measure of capacity, no need of a leakage/breakdown tester.

 - contrary to what many people think, I have to replace more defective transistors than faulty capacitors.
« Last Edit: May 25, 2018, 05:16:55 pm by oldway »
 
The following users thanked this post: GEOelectronics

Offline joeqsmithTopic starter

  • Super Contributor
  • ***
  • Posts: 11630
  • Country: us
Re: Using low voltage levels to test HV capacitors leakage current
« Reply #23 on: May 25, 2018, 05:28:07 pm »
If I can find some time I shall try to do some linearity tests. I think first try a new cap and go in steps of 5 or 10V up to the WV.  Ideal would be a programmable  voltage source and a logging meter but I do not have those (that is with enough uA resolution) so I have to do it by hand.
To do it according the datasheet I should apply the voltage, wait the stated minutes and then note the current, discharge and do the same but now to a higher voltage.

Keeping in mind that the manufacture's procedures may be different for each device. 

I was thinking to try and make some different defective parts.  Along the same lines, I am thinking to setup some sort of automated test to measure various leakages to the Joe Smith standard.   Every part would be tested the same way.  Then attempt to damage them and see how I effect the linearity.    If they were perfectly linear we could easily make a model and predict how they would behave over the entire operating voltage. 

Offline Cliff Matthews

  • Supporter
  • ****
  • Posts: 1910
  • Country: ca
    • General Repair and Support
Re: Prediction of Capcitor Leakage Current Using Modelling
« Reply #24 on: May 25, 2018, 08:12:54 pm »
Precise title Joe.  :-+  Ok, I'm in for some reading/learning..
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf