What exactly is the construction on those things anyway, mylar / metalized polyester ?
Metallized paper, actually. The choice of the dielectric is mainly related to self-healing properties (see for instance
http://www.minuszerodegrees.net/line_supression/evoxrifa_rfi_fact.pdf). Most X-class capacitors are either metallized paper or metallized polypropylene.
Also sort of curious why they fail so much. Normally epoxy sealed film caps are very reliable, but these specifically seem not to be. Maybe something about the epoxy isn't dimensionally stable over time, which causes the cracks and air exposure?
X2 are a particular type, in that they are subject to high AC currents (which causes the dielectric to swell, contract and vibrate) and are required to be flame-proof too. Cured epoxy has an ignition point of about 1000°F (i.e. more than 530°C) so it is a good choice for this application, but it needs to be resilient enough to withstand the deformations of the dielectric without crackling, and this for many years.
On a French forum I have found a post by someone claiming to have witnessed the problem with Rifa while he was working at Ciba-Geigy who (also according to him) was then the manufacturer of the epoxy resin used by Rifa. Mind you, this is just internet talk, but at least this claims to be a first hand account. You can find it (in French) there:
http://www.tsf-radio.org/forum/forum.php?show=282462#282462. Here is a Google translation of the interesting bits (I don't have the time to translate myself). I reproduce it with all due warning that this is an unconfirmed account by someone I don't personally know, so I cannot guarantee this is actually true:
"RIFA PME271 and RC metallized paper capacitors age badly even in storage. This comes from the design (casting in a mold and not in a dedicated box). Indeed, the resin used is an Epoxy Araldit resin (I worked for 35 years in the Ciba-Geigy factory producing this resin at the rate of 200 tons/day) very reactive which requires gelation for a few hours before the final polymerization at high temperature (80 to 120°C). However, this procedure is not compatible with the imperative of production, the molds not being able to be mobilized for such a long time. The unsuitable RIFA process produces hardening in a few minutes which results in high tensions in the material which produces cracks through which moisture can enter and lead to the combustion of the capacitor. I have already been confronted with this problem since the 70s because humidity modifies the properties of capacitors-increased losses- (I had returned lots of thousands of RIFA capacitors which were changed under warranty, but RIFA has never changed its manufacturing process. New RIFA capacitors. stored at room temperature (21÷25°C) show cracks and some even have shards of resin that have come off. In conclusion, it is necessary to avoid RIFA PME271 capacitors and use WIMA capacitors for example which are of conventional construction, i.e. a small case in which the capacitor winding with the output wires is introduced before the casting of the resin that can undergo the recommended procedure of preliminary gelation followed by final hardening at moderate temperature, which prevents the appearance of cracks."
Personally I have strong doubts about this version. I find it extremely difficult to believe that one of the major manufacturers of a safety-critical component has intentionally and systematically botched their whole production since 50+ years, and that their industrial customers (who are responsible of their own customers' safety) have never paid attention and keep buying those caps which are neither the only ones nor by far the cheapest X2 capacitors on the market. Also, I am not nearly as optimistic as the last sentence, as just last week I found two cracked Wima MP3 capacitors (the model mentioned above) in the mains filter of a V2000 VCR. Granted, the casing of the MP3 is not transparent so you cannot see whether it is developing cracks until the capacitor fails - but this does not mean they are not there.
It is my understanding that X2 capacitors are simply the best possible effort to achieve their goal (namely ensuring EMC, electrical safety and fire protection) for their expected lifetime, and just cannot be guaranteed to work for 50-odd years as is the case in the A77. When parts get old and especially when they are constantly subjected to stress, it is normal for them to need replacement.
![20230304_110910[1].jpg](/forums/data/attachments/2690/2690700-20d987f12816f113403580e8408d9de1.jpg?hash=INmH8SgW8R "20230304_110910[1].jpg")
