Testing Capacitors

[crackle]

I was recently extolling the virtues of the brown polyester/polystyrene capacitors with radial leads. I must say that the ones I have previously used which were of the "Suntan" make have been no problem and seem reliable.
However I ordered a different make recently and although they arrived from China in only 8 days, I had some concern over the physical size of the 4.7uF 630v capacitors.

I decided to do some testing just to confirm if they stood up to the rating of 630v DC.

I put together a simple high voltage tester using an old mains transformer from a radio and 4 IN4007 diodes as a bridge rectifier.



Here are the results of leakage tests on some 630v Metallized Polypropylene Capacitors from China. (NOT THE "SUNTAN" MAKE ONES)

The tests were carried out using a transformer with a 600v AC secondary, so I started off by winding the variac down to give 500v AC on the output of the transformer. Giving a 700v DC test voltage after rectification and smoothing. Any 630v DC capacitor should have sufficient working tolerance to be able to withstand 700v DC.

The 1uF capacitor passed OK, neon glowed steady, then started flashing with a steady decrease in the flash rate to the point at which it stopped and stayed off.
The 2uF cap also passed OK, same results.
The 4.7uF cap did the same but then after a little while the occasional flash indicating leakage.

I did the same tests at full output of 610v AC corresponding to 854v DC
1uF passed
2uF questionable, see test with AVO
4.7uF the neon would not stop flashing, continued to flash at about every 1.5 seconds FAILED TEST

I connected my AVO in as an ammeter and did some more tests on the 2uF and 4.7uF
I set the variac to give full output of 610v, these are the results;
1uF 0uA or less than the needle width.
2uF just < 1uA
4.7uF 2.5uA FAILED TEST

Then I did more tests on the 4.7uf capacitors at an input voltage of 460 AC being the closest I could get to 630v DC test voltage.
The 4.7uF cap was conducting 1.5uA FAILED TEST
During these tests on the 4.7uF capacitors I heard the occasional faint click as if there was a breakdown in the dielectric of the capacitor under test.

Further tests were done at 400v DC, the leakage on the 4.7uF capacitors was 0.5uA
On the 2uF capacitor the leakage was < 0.5uA barely measurable on the AVO
At 250v DC both the 2uF and 4.7uF capacitors showed no leakage.
My tests appear to indicate that the 4.7uF 630v capacitors may be only be suitable for 250v DC working.


I did some other tests on some yellow axial .1 and .22 uF 630v caps and absolutely no leakage current at the full voltage of 854 volts.
I also tested a some Suntan make, capacitors 1uF and 4.7uF and these again passed with absolutely no leakage at full volts.

The 4.7uF "Suntan" make capacitors are about 50% physically bigger than the faulty or wrongly marked make.

Mike

[MrBungle]

Interesting and nicely documented findings but it doesn't surprise me. You'll find those cheap Chinese brown blob poly caps inside a lot of cheaper LED bulbs. They tend not to survive very long in that role which is a common failure point for the entire device. Either that or the cheap under-rated resistors tend to explode.

Incidentally they are probably wrongly marked on purpose. They may last a few days in a typical installation which is long enough to absolve anyone in the supply chain of any blame and that is all they are rated for. You're probably right with the 250v assertion.

With respect to Suntan, I regularly use their capacitors (from Rapid) and they are always good quality and in spec, up there with Vishay/AVX/Kemet which was a surprise for the price! Their X2 caps are good replacements for the RIFA flamethrowers.

[David G4EBT]

Very informative Mike - what a neatly designed tester and well documented tests and results.

I guess to some extent, it's a good thing that many of the caps that we have to replace are in situations where the actual WV is well below the stated maximum on the caps, but that's not always so. Maybe it goes some way to explain why the likes of Farnell/CPC/RS and the like haven't been put out of business by the Chinese. Sometimes it can be false economy.

Given the recent safety concerns, with 700V DC at your disposal, I hope you wore the necessary personal protective equipment!

Can't be too careful!

https://www.bing.com/images/search?v...x=1&ajaxhist=0

[Craig Sawyers]

Amost any half-way decent X2 has to be better than the exploding RIFA X2's. Most recent one was a few days ago when my HP 4275A component bridge let loose with piles of smoke and cracking noises. Had to displace the mains transformer to find the culprit - RIFA X2.

[Techman]

Yes, very good. The only thing I would say is that the test should have been done the other way round, starting at 250 volts and working upwards. These tests could be inconclusive as the capacitors may have been partially damaged starting with the high voltage, making the lower tests meaningless to an extent. Having said that, I'm sure that on this occasion the tests you've made have proved what we all know about these rubbish components!

Someone bought an RF amplifier round for me to repair the other day. He supplied me with the new transistors which he'd bought from the 'usual' source. I told him that they'd be 'fake' and a waste of time, and I was right! I don't know what they were, but they certainly weren't anything like what the label on them said. The only thing you could say about them was that they 'looked' like possible originals and they were 'NPN' like the originals, but that's as far as it went, and collector and emitter were also reversed, total rubbish!

I wish people would just stop buying this Chinese junk! Just about all of it is 'fake'. However, I have to admit to an exception. I was recently looking for a now 'hard to find' RF output transistor and someone who 'blogs' on the net had mentioned some that had actually worked, as long as you didn't mind the 4 to 5 weeks wait on the delivery. I bought one, and although it has to be some sort of 'fake' (it clearly has its type number on a stuck on label, which I was really tempted to try to pick off to see what was underneath, but resisted), It worked extremely well, so well in fact, that I've sent off and ordered another! So to be fair, there perhaps are some exceptions, but probably not many.

[Techman]

Just as another note to the above tests that you'd made - You could work up from the lowest voltage to the highest and plot the results. Then work down, and again plot the results and compare them to see if there's any difference, pointing to any possible 'self healing', or not, of the dielectric taking place once it's been punctured at the higher/highest voltage.

[crackle]

Originally Posted by Techman

Yes, very good. The only thing I would say is that the test should have been done the other way round, starting at 250 volts and working upwards.

Yes maybe, if I had really wanted to do a full analysis of the capacitors, but it started out as a good / not good, test.
If they had been good, like the Suntan caps I later tested, then the 700 volts initial test would have been no problem for them.

Mike

[Horris]

A very good and interesting test Mike.
As I said in my PM to you I have used Suntan capacitors for the best part of 10 years now and have never had any problems with them.
They are available from Rapid Electronics in Colchester.They have just introduced a range of axial poly caps.
I have no connection with Rapid other than a satisfied customer

Dave

[Techman]

Originally Posted by crackle

it started out as a good / not good, test.

I know what you're saying, Mike, but having put together such a nice little testing circuit, it seems a shame not to do a bit more of a test with it - not that I'm asking you to take up valuable time doing this, having found they're bad compared to the quality ones, by just doing the over voltage test.

[Diabolical Artificer]

A few question's about your test jig Mike. Why did you use 3M as your bleed resistors same goes for using
4M series resistor? Wouldn't this knock the current to the CUT down too much? Second question, why the 0.1 cap across the neon? Think I know the answer but not sure.

Andy.

[crackle]

Hi Andy
I put the circuit together in consultation Michael Watterson when we were discussing the capacitors. My understanding of the 4M resistors in series renders the test probes safe, to humans and from accidental shorts and reduces any chance of excessive damage to the capacitor under test and oneself. (I tested this and all that happens is the neon lights.) Four 1M resistors were used in series because of the high voltage. However BEWARE that the capacitor under test, when charged is NOT SAFE it will hold a potentially LETHAL CHARGE, it will need to be left in circuit until discharged.
Similar with the bleed resistors, and in this case needed to be large in value to reduce the load because the smoothing caps were very small.

The 0.1uF capacitor will charge up and hence the neon "fires". In practice when the circuit is first turned on with a capacitor under test the neon appears to be on. As the test capacitor nears full charge the neon starts to flash rapidly but at a declining rate till it finally stays off, one or 2 flashes may be seen, but with the failing capacitor the neon never stopped flashing, or it stopped for a while then started to flash slowly again.
The whole cycle including discharge takes quite a while. I shorted the cap under test to finally discharge it once it had dropped to around 200v

It was a matter of keep your wits alert as I did not want to pick up a capacitor charged with 800 volts.

The test jig was not a pretty sight, it was put together in about 15 minutes, it is dismantled again now once the point was proved.
The point of this thread was to warn others of the capacitors as I had mentioned buying them from China, I dont think it would be appropriate to mention them by name, unless the mods say it is OK.

Mike

[Diabolical Artificer]

Thanks Mike.

Andy.