View Single Post
Old 05-06-2010, 08:30 AM
Harvey's Avatar
Harvey Harvey is offline
Gold Member
Join Date: Jul 2009
Posts: 1,137
Mosfet Body Diode

Hi Chris,

The MOSFET Body Diode is actually a by-product of how those devices are manufactured. The internal structure makes a naturally occurring diode junction between the source (anode side) and drain (cathode side).

In addition to this, some devices also have an Avalanche rating where that diode becomes conductive above a certain voltage. For the IRFPG50, that voltage is 1000 volts. So above 1000V the MOSFET simply turns itself on hard and dumps as much as 800mJ of energy straight through in one fell swoop - or repeatedly, 19mJ. Now those values were obtained with very specific parameters, like a 50V VDD, 40mH Coil, a 25Ω Gate resistor and a maximum IAS current of 6.1A. But they do give an idea of the maximum power can be dumped through the device and still have it survive. The IDM (pulsed drain current) can be as high as 24A, and the continuous drain current can max out at 3.9A if the device reaches 100C. So even though it is robust, it still has it's weaknesses. Another critical area is the Gate to Source voltage. This is a CMOS device, and the Gate is susceptible to ESD. The limit on that is +/-20V relative to the source pin. So if you have a circuit that the source pin floats up to 24V and the gate is held down at ground, you just popped the FET. Conversely, if you drive the Gate to +24V and the source is clamped at ground, it will pop the FET. Also, these devices really need the temperature to be controlled. They can get real hot real fast if not properly dissipated via a heatsink or the like.

So the Body Diode can be viewed as a large 6.1A Zener Diode built inside. When the diode is forward biased (negative voltage on the drain relative to the source), there will be a 1.8V drop across it's junction. That is the minimum drop. So in that condition (like a ringing that goes negative) the diode will conduct and short the ringing to the source pin. At full pulse current of 24A, that means the device is dissipating 43.2W in that diode. The case allows for 190W dissipation if properly heat sunk. Any time that happens, you are wasting energy. (unless the MOSFET is your heater ) This is why I suggest floating the inductive ringing above that level to prevent the damping if you are using it to heat the inductor, or feed a bridge charger take off node. For example, if your Peak to Peak ringing voltage is 100V, then put your supply at 60V to keep the bottom peak above ground. The IRFPG50 can easily support ringing of 990V, so a 510V supply would work fine for that and all of the energy would be returned to the inductor. This way the MOSFET stays cool, the Body Diode does not have to conduct in either direction and all of the energy stays with the inductor to be dissipated or taken off for charging.

Although somewhat unconventional, the inductor can then be the primary of step down transformer. With a 500VDC supply, the MOSFET can pulse on in phase as the ringing begins to decay and keep the Primary ringing continuously near 990V +/- 100V perhaps. The secondary can then be used to AMP up that energy at a lower voltage and apply it to do work. The frequency is only limited by the components and RF losses due to their geometry. At 24A of pulsed DC current (just to charge the inductor back up to ringing) you could get as much as 16KW pulsed energy through the transformer. Of course at 24A, the power dissipated by the transistor would spike to 1152W and the device can only dissipate 190W. So this can only be achieved if the device is operated below the thermal impedance margins. Frequency durations longer than 0.1 seconds are reaching the limits for all duty cycles, while a 10% duty cycle with a frequency duration of 10s (100kHz) is at the bottom end where things are cool. So between 10% and 50% at 100kHz you run the full dissipation range with this device. So while the above sounds enticing, there are some very real limits that preclude even a 100kHz pulse at 24A while we may think it is within the maximum ratings. A more realistic approach would keep the current below 6.1A for safe measure. But even then, we are looking at 4KW at the transformer. IR tells us the power in FET, where it is 2 Ohms ON, to be 74.42W if the current were continuous DC. So it would get pretty warm even during normal operation. Devices with lower on resistance, or parallel MOSFET's can greatly reduce that dissipation loss.

We could probably help you solve the burn-up problem you are having if we had more information on the specifics of how they are being applied.


Reply With Quote