Fault protection in PV half bridge converters

[fourtytwo42]

I am still investigating a serious failure I had a while ago where a complete half bridge (including the pcb) was destroyed by an unknown fault. This half bridge was protected by a circuit that detects sudden changes in the capacitive divider (other half of bridge) centre tap voltage (that I had seen used elsewhere) the reason this was chosen is high frequency high current current transformers are practically unobtainable and also difficult to make.

Upon close examination however I have discovered this method is unreliable and incapable of detecting some faults altogether, for example the attached simulation at 130uS throws an ohmic short on the low mosfet (a typical failure mode) the red trace shows the high mosfet drain current rising from its nominal 20A to over 160A enough to destroy the device in the 20uS on time shown here (25Khz converter) yet the change in ripple voltage at the capacitor centre tap (in this case 2 x 10uF) is insufficiently detectable from the normal variations to be usable to protect the mosfet. The same situation occurs should the high mosfet be the one to fail first. It should be noted that to save the mosfet it needs to be cutoff typically <5uS after the current becomes excessive.

As I have already said a current transformer is not a practicable solution but do not be fooled into thinking the centre tap voltage variation can be used to protect half bridge mosfets from failure.

I am presently looking into some other solutions..........

PS corrected frequency
Changed pdf to png so it displays an image here.