My observation is that they fail usually after about 3000 hours with regularity.
In many of the previous comments in the forum on this recurrent topic there are both a lot of complaints and testimonials regarding the WZ-68. When they work I love the things but have had at least 8 failures.
I want to mention here the fact that I also use them almost constantly during the day and evenings. I swap out amps regularly in the case of the ST-120's but still they get an average of 4-6 hours each every day if the time is averaged out.
I have replaced the diodes with a more robust version (IN-5408,3amp 1000V) as well as the sag reisistors I even drilled big holes in the caps on a couple for better cooling but none of this has solved the fried thermistors. It always seems that after roughly 3000 hours at operating temp the thermstors disintegrate.
It makes some sense as current inrush limiters operate much like a resistor and heat is a by product of resistance.
I am not sure what more can be done to mod them to be more robust n their current configration.
Replacing the thermistor with a higher value but in the same physical size (so it still fits in there) may not be much help ether if overheating is actually the issue. I now think it is the main culprit in many WZ-68 failures.
Whether direct overheating of the thermistor itself or the combined heat of the whole cap.
It could also be too close to the power tranny which also produces considerable radiant heat adding to the problem.
When I look at the WZ-68 circuitry it seems to have good specs but the compact format and the copper shield still retain a lot of heat. Even the ones I drilled out the cap on are still too hot to touch.
I'm thinking of making one on a perf board that can sit out in the open to test this theory.
Of course it would under some kind of cover to avoid accidentally touching it and perhaps death by electrocution.
Meanwhile I have finally bit the bullet and ordered 2 brand new Mullard GZ-37's to keep at least two of my systems operating.
I know I live in a place where the electrical service is very poor (Costa Rica) but I have latch breakers on all of them and a variac to control voltage on the one in use but the thermistor failures keep occurring regularly. With the doubled protection I think a bad grid is not the problem. More likely the problem is the thermistor deteriorates from heat over time.
Bob mentions his have been good for years without a failure but he also mentions he does not log a lot of continuous long hours either.
I have a system playing music or home theater at least 6-8 hours a day which is a quite a lot but it may explain why these things fail on me regularly.
Meanwhile I will go back to a traditional quality tube rectifier (GZ-37 Mullard) while I run the perf board model I mentioned to test its longevity. If in time it proves the thermistor proves to be reliable that would confirm my overheating theory.