Interesting low voltage capacitor leakage tester

[Argus25]

Originally Posted by Diabolical Artificer

Hadn't heard about the waveform test, will have to look it up.

Andy,

You might not find this as I have not published the article on this yet, perhaps somebody might have.

I found this effect in some vintage bumble bee capacitors. They behaved as though there was a capacitance, about x10 the value of the original capacitor, with a high value resistance in series in the order of meg ohms , placed in parallel with the original value.

So, in a timing circuit, like a valve oscillator, where the impedance was high, the oscillator frequency runs grossly low, but in a differentiator circuit with circuit impedances less than say 50k, there is no apparent effect.

The only plausible explanation I found was that the paper insulation, which contains salt contaminants, had absorbed water over the years to create a weak electrolyte between the foils.

In any case the effect is easily seen looking at a discharge waveform into a high impedance load. Two exponential decay profiles are clearly visible, an initial fast one, then a slower one due to the large value. It is like a very severe case of dielectric absorption.

The capacitors can also have parallel resistance. So what in effect was one component when new, ends up with age behaving as a network of 4 components, and the effect of this depends on the circuit the capacitor is used in.

When I first observed this effect I realized it would be possible to make a unique type of capacitor tester that tested the quality of the capacitor's dielectric by comparing the exponential charge or discharge profile with a known good capacitor.

[ms660]

9 volts can't jump like wot 500 volts can.

That's my theory.

Lawrence.

[Argus25]

Another thing to watch out for;

This is particularly the case for X2 capacitors that have a metalized paper film.

You can check them on a leakage tester, perfect, no leakage. And check them at their rated voltage, all ok. Seems wonderful, until they are checked on a capacitance meter and you find they have lost half or more of their capacitance.

I found this out after using X2 capacitors in series with the 3W motors of 115V synchronous clocks, so as to drop the voltage and run them on 230V (these motors also had altered gears so the rotational speed was correct on 50 vs 60Hz).

After setting up the uF value correctly, initially, so that the voltage across the motor winding was 115V, the voltage dropped with time over about 6 months.

Testing showed the X2 capacitors had lost capacity.

So I disassembled the X2 capacitors and unwound the foils and found that the paper and its metallic film had been eaten away over time (but no shorts). I have never seen this with poly caps. So I replaced the X2 caps with 1500V rated poly caps as a trial for the same application, the experiment now 3 years old, with no degradation of the poly caps. This is why I fit 1500V rated poly caps in place of X2 caps in my own gear and never use X2's, but obviously some are better than others.

I also found that the X2 caps in my air conditioners had lost capacity too.

[G8HQP Dave]

Yes, X caps losing capacitance is a common phenomenon. I have had to replace them in a doorbell and oven controller - both now left at my previous house. I expect to have to replace one in the receiver for my heating programmer eventually. There must be lots of items thrown away (or having expensive repairs by swapping a whole PCB) when all they need is a new cap.

[ukcol]

I have had to replace them in a doorbell and........

Same here. I had a door bell that sounded like it had a sore throat, checking the class X capacitor that was used in the power circuit as a capacitive dropper showed that its capacitance was less then half its original value. A replacement fixed the bell.

I don't know anything about the details of class X capacitor construction but I suspect this loss of capacitance is associated with the "self healing" function of its design. Perhaps when a voltage spike appears across the capacitor it monetarily breaks down and the fault current blows a hole in the plates near the breakdown. I can imagine that over time this process would gradually reduce the area of the plates and hence the value of capacitance.

This is all speculation along the lines of a "thought experiment" and could well be complete b ks.

[Argus25]

I don't know anything about the details of class X capacitor construction but I suspect this loss of capacitance is associated with the "self healing" function of its design. Perhaps when a voltage spike appears across the capacitor it monetarily breaks down and the fault current blows a hole in the plates near the breakdown.

Of the ones I opened up, they didn't actually have plates or foils. It had a very thin metallic deposit on paper. I think you are right, voltage spikes must slowly blow it away and they call that "self healing"

Maybe large islands of metalization get isolated to account for the loss in uF value. The ones that are in a 4 year old air con unit I have are down to about 1/2 capacity now. They are only being used as mains RF filters so as they fail the unit still runs. But in a series dropper application it not good. In a computer switchmode PSU I recently posted about for the 5155 computer, all the power is transferred via two X2 looking capacitors to the primary of the main transformer, but they are still ok, so I think they must be a poly cap and not a metalized paper cap, or by now one would think they would have lost uF value.

These capacitors though make a good point about capacitor testing, probably there are Four tests at a minimum that a capacitor must be subjected to, to determine if it is ok, or not:

1) leakage test
2) uF value test
3) ESR
4) Rated voltage withstand test


Just one test alone isn't really good enough. So obviously the ideal capacitance tester built into one box would do these four tests. At the moment though we have separate meters for these most of the time.

[Slothie]

:
1) leakage test
2) uF value test
3) ESR
4) Rated voltage withstand test
:

These are the basic failure modes of capacitors; Mr Carlson's tester only really tests (1), and to be fair the "magic eye" tests of the Jackson and Heathkit do too, they need the high voltage to get a detectable current. Failure (4) would not be detected by Mr Carlson's tester of course and maybe the other 2 might if internal arcing happened at a low enough voltage.

I suppose what I'm getting towards is that Mr Carlson's interest, and thereby that of most of his viewers, is in repairing old valve and transistor radios and amplifiers, and because these kind of devices are not stressing capacitors like a transmitter, RF amplifier or television might, 99% of failed capacitors in these kind of devices are going to be electrically leaky due to electrolyte corrosion/drying or moisture ingress; his design, being able to detect very small currents will detect these very common failures better than the other types despite the low voltage. Capacitors that appear fine at low voltages but fail at higher ones are going to be rare, and probably evidenced in other ways, e.g. discoloration, loss of capacity or holes in the side

Perhaps if his youtube viewers build his tester (I'm probably going to) then they wont buy up all the "proper" testers and they'll be available for those who know how to make better use of them!

[Radio Wrangler]

Originally Posted by ukcol

I don't know anything about the details of class X capacitor construction but I suspect this loss of capacitance is associated with the "self healing" function of its design. Perhaps when a voltage spike appears across the capacitor it monetarily breaks down and the fault current blows a hole in the plates near the breakdown. I can imagine that over time this process would gradually reduce the area of the plates and hence the value of capacitance.

This is exactly what class X capacitors are designed to do.

The peak voltage of transients on the mains distribution network can be very high. In some places transmission-line effects can multiply surges.

An anti-RFI capacitor needs to have enough capacitance to make the equipment it's in pass RF emission tests, but it also has to fit in a reasonable space. This limits the voltage it can stand.

Together these mean that a capacitor straight across the mains will get flashed-over from time to time. Class X capacitors are rated to survive this without going short-circuit. The paper burns and self-extinguishes but remains insulating and the metallisation melts back without shorting across the blasted dielectric. So these capacitors reduce in area with time.

You could view the reduction in capacitance as a sort of dosimeter for mains transients!

It also means that their effectiveness in the job of suppressing RF emissions will fall with time and equipment which would pass EMC tests when new will get noisier.

Many years ago I was at a presentation by Rifa on how their capacitors worked. They were aware of that last bit but recognised it as being inevitable and recognised that enough margin had to be built into the RFI measures to handle the expected product lifetime under average mains quality conditions.

David

[ukcol]

Originally Posted by Argus25

......probably there are Four tests at a minimum that a capacitor must be subjected to, to determine if it is ok, or not:

1) leakage test
2) uF value test
3) ESR
4) Rated voltage withstand test


Just one test alone isn't really good enough. So obviously the ideal capacitance tester built into one box would do these four tests. At the moment though we have separate meters for these most of the time.

There is your next design project then Hugo.

Originally Posted by Radio Wrangler

This is exactly what class X capacitors are designed to do.

Thanks David.

[Argus25]

Originally Posted by Radio Wrangler

You could view the reduction in capacitance as a sort of dosimeter for mains transients!

That is a great remark.

It just goes to show how nasty the transients on the mains can be, large voltage spikes originating from a very low impedance source. Combinations of inductors and large gas discharge arrestors might be a good way to snub some of them off, but these all occupy more space. The only caps I'm prepared to accept across the mains that are not X2's are 1500V rated and always fused. Probably 2kV would be better.

In early vintage radios I have seen 1000V waxies on the mains inputs, and surprisingly a lot of these did better than the modern X2's, but they were bigger.

[Refugee]

I have in the distant past seen a 1000V waxy pulled in half like a Christmas cracker in a dual standard TV.
The reported fault was that it needed a safety check because it had gone bang with smoke and then carried on working.

[VintageTest]

Paul of "Mr Carlson's Lab" on YouTube does appear to know an awful lot about virtually all corners of electronics, but I too am sceptical about detecting leakage at very low voltages. However, the little box appears to do what it says on the tin, and his video appears to back this up. I'm not a Patreon member either (although there does seem to be some nice articles and projects on there) but being a Scot, I'm allergic to paying for - well - virtually everything - but my stump up in this case. Oh, I agree totally with Argus25 on the don't-reform-old-electrolytics too. I tried this in days gone by, but like refilling lead-acid batteries with fresh acid, it only delays the inevitable!

[Anthony Thomas]

Has anyone built Paul Carlsons capacitor tester yet?