IGBT & MOSFET faults

Usually there is some circuitry that looks at the power switch and decides if it is on when it is commanded on, and if it is on when it is supposed to be off, if wrong it assumes a fault and can then fire a bypass switch and usually an opto-coupler back to the control to identify the faulty switch so the software can alter the control to compensate, obviously if too many switches fail - you have to stop...

Orson Cart said:

Usually there is some circuitry that looks at the power switch and decides if it is on when it is commanded on, and if it is on when it is supposed to be off, if wrong it assumes a fault and can then fire a bypass switch and usually an opto-coupler back to the control to identify the faulty switch so the software can alter the control to compensate,

Click to expand...

could you please provide more detail about this subject.

obviously if too many switches fail - you have to stop...

Click to expand...

yes, you are right. the converter with faulty switch can be used in fault ride-through mode as long as the number of failed switches does not cause DC-link voltage large swing or the demanded level comes larger than 2^N. N is number of healthy cells.

In practice, how we can recognize faulty mosfet/igbt? I mean while the circuit is working, not testing the component.

By monitoring every semiconductor device for current and voltage, and comparing what is happening to what should be happening.

This concept is not unusual in aerospace or military grade hardware, or life support, where very comprehensive self checking is pretty routine.
But the cost and complexity is just not practical for cost sensitive commercial grade or consumer grade equipment.

Its going to be much more economic to just carry a complete backup spare, and just plug in unit B, when unit A lets the smoke out.

Isn't this trend little awkward?
at least a voltage sensor for every switch is needed, too way costly.
I am looking for a much more economic way! I have many switches in the topology.

Thank you

kappa_am said:

Isn't this trend little awkward?

Click to expand...

Yes, very awkward and very expensive, but that is how it is done.

How many times have you heard that the count down for a missile has been aborted because of a detected anomaly in one of several multiple redundant backup units ?