I have 3 active devices in series, SCRs in this case but I'm guessing that's not important. Each is rated for 2.4 kV. The total voltage I will actually apply is 6.6 kV.
Leakage current is guaranteed to be 80 mA maximum, but the datasheet does not give a typical value. At 2.2 kV per device, that suggests that each device's off resistance can be as low as 25 kohms. 6.6 kV * 0.08 A ~= 500 W, which for the application is a lot. Putting 2.5 kohm resistors in parallel to force equal sharing is out of the question because the 5 kW it would burn is way too high.
I figure the device that eats up the most voltage also has the lowest leakage. Therefore, if any one device eats up enough voltage that it would be out of specification at full voltage, I ought to be able to put a shunt resistor in parallel with that device so that it is in specification.
Note that, in a string of three devices, only one or at most two of them will get a shunt resistor. That is what makes this network a bit unorthodox. It's also completely ad hoc. Every example I've seen of using resistors to achieve equal voltage sharing used identical smaller resistors conducting at least 10x more current than device leakage. I can't do that because it would eat up my entire power budget.
My question is this:
If I do this, is the resulting shunt resistor likely to be valid for several years? Will most SCRs keep the same leakage current provided they are not abused?
Edit: What this question is really about is the stability of an SCR's leakage current. If you measure it in the factory, does it stay relatively stable over the life of the device? Or is it prone to go up and down?
