[R-390] C553

[Charles Steinmetz]

David wrote:

>That several hundred ohms series resistance might have limited the 
>event energy; optimum self-heal might require a low-impedance 
>source, same as tantalum electrolytics.  Do you know if the 
>manufacturers claimed that the energy stored in the cap is enough 
>for a good clear?

I get the impression that a typical MF cap clears with its own 
charge.  The clearing energies disclosed by manufacturers are 
substantially less than the energy typically stored in HV 
capacitors.  Also, note that the capacitor's internal resistance and 
inductance impede the flow of energy from the capacitor terminal to 
the fault, even if the power supply is connected to the capacitor 
terminal through a low impedance.

C553 operates with a much greater source impedance than my test 
circuit (2.2k + 12mH, although C511 complicates the calculus), so 
clearing of C553 wouldn't be any better than in my test circuit even 
if the series impedance does matter.

>It's interesting that your test protocol included power-off/on 
>cycles.  I seem to recall that at least one of the reliability 
>predictors factored this in, and it had more effect than I expected.

Yes, power cycles are stressful to caps, even if the dv/dt is 
significantly limited by the circuit.  This is especially true of 
metalized-film caps, which use vanishingly thin dielectric sheets and 
therefore have extremely high field gradients.  Every time you charge 
or discharge a cap, things move under these extreme forces both at a 
microscopic level and at a molecular level.

Best regards,

Charles