The problem is due to breakdown of the dielectric between the capacitor plates. The plates are self-healing, meaning plate damage is limited, but this means there's a gradual loss in the surface area of the energy storage capacitor plates over time.
Ultimately this results in a loss in capacitance. As the capacitor ages then the breakdown gets worse and may be exacerbated by the fact that, for a given delivered energy, when the defibrillator is in calibration, the actual voltage across the storage capacitor may increase slighty (Q=CxV).
This further encourages breakdown. As the rate of dielectric breakdown increases this popping or crackling sound can be heard as energy being dissipated in the capacitor, resulting in increased electrical leakage from the fully charged capacitor.
I think age, temperature, cycling and electrical stresses, i.e. rate of charge, P.D. across the capacitor and number of full-charge cycles, contribute to this problem.
In practice I've had to change capacitors that fail the energy discharge test (360J to 1J in less than 20s) because the capacitor has gone out of tolerance (measurement has confirmed this and changing the capacitor has solved the problem).
The sounds of discharge inside the capacitor at 300J-360J can get that bad that operators think it's the mains socket is arcing, so they report the machine faulty despite it being functionally OK and operating within tolerances.
We have about 40 LP9 defibillators left in our hospital now and about 20 faulty energy storage capacitors on a shelf yet to be disposed of responsibly. Our LP9s are about 14 years old on average though so we've had reliable service out of them.
