Bad X capacitors?!

[LongRunner]

Yes, they can fail too. See: http://www.vintage-radio.net/forum/show ... hp?t=59610

To my knowledge, besides self-healing events, internal corona discharge (which progressively eats away at the metallisation) is what eventually does them in. Of course, the only consequence of their failure in EMI filters is an increase in differential-mode interference, making their failure even more covert than typical age-related failures (dried-out electrolytics, etc.). It could be that you have a PSU (maybe several) with an X capacitor that has been practically open circuit for years, without knowing it.

Not having a capacitance meter of my own (yet) I can't personally attest to that experience, but it's something to take notice of. Could we knock points off units with low-quality X caps? There are endurance tests for X and Y caps, but it's not as though the makers of crappy 'lytics don't give endurance spec's.

[Behemot]

This happenes for all film/foil caps. Misused foil caps even from Panny or Rubycon will fail. Besides I have no idea what manufacturer is good and what is bad, there are thousands of them. Many times you are not even able to identify.

As for safety caps, they are designed so they only loose their ability to act as capacitors, that means loose the capacity because of metal film evaporization. They should never short.

For measuring these caps you need very low capacity meter, ESR Micro is only able to measure the bigger ones from 0,1 uF.

[LongRunner]

To straighten out the terminology: The capacitors in question are metallised film types. Film and foil caps are different – those have thicker aluminium foils, which provide lower ESR, but they are bulkier for the same capacitance/voltage and cannot self-heal. (As I understand the terms, "foil" conventionally refers to a thin sheet of a metal while "film" conventionally refers to a thin non-metallic sheet, most often plastic.)

Anyway, I know that it's an inevitability, more-or-less (although some manufacturers are claimed to be worse than others). For what it's worth, for dropper applications, there is a "high stability" variant with series instead of mono construction. The trade-off is that they are bulkier for the same capacitance and voltage rating.

If surges are the main problem, using X1 capacitors (designed for equipment for fixed installation) instead of X2 might help. (Strangely, Kemet seems to be under the impression that you can't use X1 caps in plug-in devices. ) If it's corona discharge doing most of the damage, failure would be quicker at higher mains voltages (a given make and model of cap might be stable for years on 120V, but not for very long on 230V).

For the time being, I would suggest the "innocent until proven guilty" treatment, as far as scoring deduction goes.

[LongRunner]

I found this article, which states that in recent times, manufacturers have reduced the thickness of the sealing compound and/or the casing (the article doesn't make it clear which); presumably in a desperate attempt to make capacitors just that bit more compact. Given the increase in moisture susceptibility that resulted, this turned out to be futile.

Until they develop a new dielectric with superior volumetric efficiency and HF performance comparable or superior to polypropylene, I think they'll have to stop there.