X2 Capacitors - loss of capacity

[jimmc101]

Earlier today I needed a 330nF X2 rated cap, a quick hunt through some scrap PSUs produced one.

Measuring it I found the actual capacity to be 89nF.
Intrigued I tried a few more used X2 caps I had lying around, the worst I found was marked 0.22uF +10% and measured 13nF!
Neither capacitor showed signs of damage and in each case the insulation resistance was greater than 1000Mohm at 500V.
Testing new caps showed they were all within tolerance.

It looks as if capacitors across the mains can suffer a large drop in capacity over time, presumably caused by repeated self healing after suffering repeated spikes on the mains.

I've not come across this before has any one else seen it?

Jim

[paulsherwin]

They do lose capacitance as they self heal, but I wouldn't expect the numbers you've been finding unless the caps have suffered something catastrophic like a lightning strike. I'll test a few from my junk box the next time I have access to a capacitance meter.

[FRANK.C]

Hi Jim
In the equipment I service this quite a common problem. The X2 capacitors are used as droppers for LT supply to the electronics. The relays on the board usually starts to chatter when the cap's cant supply enough current to hold them on.
We have to replace the complete board as the company wont allow us to do component level repairs.

Frank

[Lucien Nunes]

Yes, quite common. Only a few weeks back the fridge-freezer temperature warning flashed up. Turned out that the defrost watchdog timer (a motor-driven camswitch) had come to a halt in the defrost position because the self-healing motor ballast cap had fallen to about a quarter of its proper value. Replacing it with an orange drop from the toolbox spares pack got the timer running again; the old cap tested fine for insulation.

The cap is subjected to something less than mains voltage continuously, but with the tiny clock motor winding in series it is hard to see how it would be prone to transients vicious enough to puncture the insulation regularly. It would be daft if the switching spikes of the freezer's own compressor were sufficient to damage the cap progressively.

Lucien

[Richard]

Hi

I had to replace one on the PCB in our Whirlpool freezer a few weeks ago, it had dropped to about 50% of its value and then the relay would not pull in. It is used as a dropper feeding the PCB. Freezer is about 4 years old.

Richard

[Lucien Nunes]

I wonder whether all these caps that have lost value in ordinary use would have failed short-circuit had they not been self-healing types? If not then the self-healing capability has come with a sacrifice of ultimate durability of the dielectric.

Perhaps it's a bad idea to fit X or Y rated caps in vintage restoration work? They might be suitable for something with an expected life of a few years that will be on the scrapheap before the cap goes out of tolerance, but in a radio that will hopefully be kept for decades?

Lucien

[G8HQP Dave]

Don't forget that when in series with an inductor, such as a motor, the capacitor sees a greater voltage than the incoming mains, except for the very fastest spikes. At the frequency where the cap and motor are series resonant the voltage will be Q times the input. This issue probably won't occur when the cap is being used as a dropper for a circuit, as most circuits will look capacitive.

I can't believe that most of the plates have been spark-eroded away, so is it gas liberated from the dielectric which has increase the separation a little? Remember that breakdown will preferentially occur where the dielectric is thinnest - the parts which contribute most to the total capacitance.

[kenneth]

Does this (x2 loss of capacity) mean that it is risky to use X2 caps as replacements for voltage droppers?

[paulsherwin]

It's not risky to use them as droppers. If the capacity drops the voltage applied to the heaters will fall rather than rise. Also, voltage will only be applied to the caps when the radio is turned on, so the caps will take many many years to degrade. X2 caps in modern PSUs are often connected to the mains 24 hours a day, even when the appliance is supposedly turned off.

[Lucien Nunes]

A low voltage electronic circuit driven from a capacitor-fed rectifier with some kind of shunt stabilisation would look almost like a short circuit to the cap, so for incoming transients the capacitor might as well be connected across the mains. A heater chain could have enough ohmic resistance especially when hot to provide a useful defence for the capacitor.

For the synchronous motor in my freezer timer, the resistance of the coil is a significant chunk of the total impedance (IIRC a few k DC resistance) and it does not go anywhere near resonance. The cap is used as a dropper rather than a phase shifter so its voltage is less than mains under normal conditions. The situation would be different for a phase splitter cap on an induction motor.

Dave might have a point with internal gassing / delamination, I too was struggling to see how so much active area could have been lost. I have just flicked through a couple of datasheets for self healing capacitors. They don't mention any such failure mode but then they don't admit to capacitors routinely losing significant capacitance either!

Lucien

[murphyv310]

Hi.
From the beginning of "time" capacitors have been a pain in the b**t one way or another, still we cannot design and make a truly reliable capacitor. Perhaps one day!

[McMurdo]

The problem is the raft of cheap X capacitors from China and, suprisingly, Italy. The good ones from Wima, BC (Philips), Epcos (Siemens), AVX and Roe (Roderstein) etc seem to last ok.

[kenneth]

Originally Posted by paulsherwin

It's not risky to use them as droppers. If the capacity drops the voltage applied to the heaters will fall rather than rise

Thanks Paul. I'm relieved to hear that.

[kalee20]

One thing is now worrying me - if the X capacitors in a mains filter start losing their capacitance, then although the equipment will still function, the conducted RFI produced will probably rise and it could fail to comply with standards.

This being so - how does this fit into the world of EMI? It would be rare for anyone to check their 6-month old computer, and return it, because it'd still be working. But the thing could be getting noisier and noisier at RF. Multiply this for every computer, TV, etc...

[G8HQP Dave]

Some Chinese factories have a simple solution to slowly degrading X capacitors: just omit them after the prototype has been through the EMC lab, then they can't degrade!

[tvden]

Originally Posted by murphyv310

From the beginning of "time" capacitors have been a pain in the b**t one way or another, still we cannot design and make a truly reliable capacitor. Perhaps one day!

Yes I agree and we still dont have a simple 100% definitive tester either

[Valvepower]

Originally Posted by murphyv310

From the beginning of "time" capacitors have been a pain in the b**t one way or another, still we cannot design and make a truly reliable capacitor. Perhaps one day!

Hi All,

It’s slightly off topic but talking about capacitors being a pain in the b**t, I was recently quoted a delivery date of sometime in the middle of 2011 for 5000pcs of a 0.22uF X2 capacitor from Evox, as there appears to be a world wide shortage of film to make capacitors at the moment.

Also Evox and Arco were recently bought by Kemet so they have a bit of a monopoly now.

Terry

[QQVO6/40]

Ahhhhhh the consumer appliance manufacturers Holy Grail! Planned obsolescence! Another conspiracy theory. One way to make a quid, make it fail and you have to go and get another one. Hihi. Oh well I just had to throw that one in.

[PJL]

Only use an X2 capacitor where it doesn't matter if you didn't fit one in the first place. I am with the Chinese on this one.

[G8HQP Dave]

But it does matter. Far too many CFLs, computer PSUs, DVD players etc. spray interference around because the suppression components have been omitted from the PCB during normal production. A fading capacitor is better for EMC than a missing capacitor!