Quote:
Originally Posted by MPaulHolmes
It seems to me that you could do a near equivalent of the desat detection in software. If the overcurrent was tripped, mark the time. If current doesn't get under control in a certain amount of time, open the main contactor, because a tripped overcurrent circuit without reducing current would imply a shorted igbt I suppose. It would be slower than the desat detection I think, but that may not matter.
|
That's not quite the same as desaturation detection.
Desaturation detection samples the voltage drop across the device when it's supposed to be fully on. If it rises above the desired level, you turn the device off.
One effect this has is to protect against an output dead short, where the device's on resistance is the largest in the circuit. But it also protects against unstable operating regimes e.g. oscillations at exactly that operating point.
A desat detection circuit can be as simple as a diode, resistor and capacitor. The resistor is attached to the gate drive, the diode cathode to the collector/drain, and the capacitor to ground. When the gate drive is high, the resistor charges the capacitor up. When the capacitor reaches the voltage of the collector (plus a diode drop), it stops charging and has a sample of the device voltage drop.
The RC constant isn't critical. It should be long enough to compensate for the brief period that the gate drive is high but the device hasn't turned on. And the voltage shouldn't sag too much while the gate driver is off, but it shouldn't so long that the detection is slow.
The circuit can be improved with additional parts: a matched diode in series with the sample output to compensate for the diode voltage drop; diode+resistor pair in series with the gate drive so that the capacitor doesn't discharge quickly when the gate drive is low. But it's best to start with the simple three component version and see if there are problems rather than reinvent a complex sample-and-hold.