I designed a circuit around LTC4366 as per the picture below. However numerous MOSFETs have been damaged and become useless.
Mosfets used : STH12N120K5-2; IXFT50N85XVH; STW13NK100Z - All are rated to above 800V and 12A
The Load is operated with 380VDC and can handle 450VDC MAX (the threshold in the circuit is set to 420VDC). The power supply can deliver 800VDC if it fails (and this is what I want to protect the load from)
1) If the Gate Cap (Cg) is 3.3nF the Vout never goes up. The circuit behaves as if it was in the protected mode.
2) If the Gate Cap (Cg) is 10pF the circuit works and turns off the MOSFET when the input voltage reaches approx. 420V. However if I set the power supply to 550VDC, the LTC4366 resets the failure every 9 seconds and after a few cycles the MOSFET simply dies and shows a short between Drain and Source
In the first case with 3.3nF gate cap, the inrush current is very small and it takes roughly 62 msec to charge the output cap . The timer timesout much before this and shuts the gate off. the part will still retry after 9 sec but fail to start as timer is set for almost 10.5 msec. You might want to take a look at scope shot of the timer to confirm if it reaches the timeout threshold during startup .
In the second situation , since gate cap is small , the inrush is larger . Hence the part turns on. Note however due to resistive load , the FET carries load + inrush current . Since it retries with several 100V Vds the temperature rises and eventually the MOSFET fails.
I will recommend using the LTC4366-1 instead of LTC4366-2 so that the part remains latched-off after such a fault . With such high Vds and current latching-off would be better instead of auto-retry functionality.
I bought the #1 version and am using a 100pF as the C gate and it seems to work but I am still confused. The charge pump manages to keep the gate voltage 12V above the source until the Vin is smaller than around 110Vdc. When the Vin reaches its nominal value at 380Vd, the Vg is sitting at 280Vdc, which gives Vgs = -100V. However the MOSFET is still on.
I am going to try using a zener between gate and source to keep Vgs below 30Vdc (which is the maximum allowed).
Can you shed some light please?
LTC4366 has an internal zener clamp to ensure 12 V between Gate -Source to protect the mosfet gate oxide. You measured input , gate and output w.r.t GND or Vss?
When the input was 380 V, and gate was at 280V , did you measure the output (MOSFETsource voltage)?
-100V Vgs does not sound right . If the gate is at 280V (abs. voltage wrt GND) and if output is still 380V then I expect there is drain-source short.
My apologies for the misunderstanding.
I was measuring from GND to SOURCE (380V) and from GND to GATE (280V) and didn't realize the GND is floating as the Cg is in between them. When I measured the GATE in respect to SOURCE I could see the 12V there.