SSTC mosfet reverse diode blocking- any useful?
Dr. Dark Current, Sun Apr 04 2010, 11:42AMAFAIK MOSFETs conduct in "both ways" so the junction should be able to drain the reverse recovery charge before the switching occurs right? Or is there a possibility that the reverse diode would conduct right before the switching transition?
Anyone could share a case where adding the reverse diode blocking circuit actually saved his MOSFETs from exploding?
I'm asking because I just had a transistor blowout for no apparent reason, it may also have been a funky PLL chip, I'll get a new one and put it into a grounded metal box, but don't want to pop the FETs again if that wasn't the problem
Re: SSTC mosfet reverse diode blocking- any useful?
Dzejwor, Sun Apr 04 2010, 10:29PM
I using circuit like this in all my SSTC's and I have no problem with MOSFET's MBR2545 is a 45V Shotky diode and 15ETX06 is a fast recovery diode (as fast as possible) Nominal current of shotky should be same or higher as MOSFET.
Dzejwor, Sun Apr 04 2010, 10:29PM
I using circuit like this in all my SSTC's and I have no problem with MOSFET's MBR2545 is a 45V Shotky diode and 15ETX06 is a fast recovery diode (as fast as possible) Nominal current of shotky should be same or higher as MOSFET.
Re: SSTC mosfet reverse diode blocking- any useful?
Steve Conner, Mon Apr 05 2010, 10:52AM
Shorting a diode only drains its reverse recovery charge very slowly, compared to what happens when you apply the reverse voltage.
I guess the charge drains eventually, but I've seen it persist for tens of microseconds in brick IGBTs and still produce a nasty recovery spike.
Steve Conner, Mon Apr 05 2010, 10:52AM
Shorting a diode only drains its reverse recovery charge very slowly, compared to what happens when you apply the reverse voltage.
I guess the charge drains eventually, but I've seen it persist for tens of microseconds in brick IGBTs and still produce a nasty recovery spike.
Re: SSTC mosfet reverse diode blocking- any useful?
Dr. Dark Current, Mon Apr 05 2010, 12:27PM
So do you think it makes sense to block the diode? Because I also get increased dissipation this way.
Dr. Dark Current, Mon Apr 05 2010, 12:27PM
So do you think it makes sense to block the diode? Because I also get increased dissipation this way.
Re: SSTC mosfet reverse diode blocking- any useful?
GeordieBoy, Thu Apr 08 2010, 12:29PM
As Steve said the slow and painful reverse recovery of the body-diode does not take place if it is "shorted" by the MOSFET channel. Instead it just hides for a while until reverse voltage is applied and then it rears its ugly head!
From what I remember, recombination is a slow process, so it takes applied reverse voltage to sweep out the charge carriers quickly. This is when you see the reverse recovery current spike.
Some IGBTs also exhibit this behaviour when operated in ZCS circuits. Even though the current falls smoothly to zero before turn-off, you still see a small current bump when voltage is applied to an IGBT that should be blocking.
For load currents with leading power factor, or any other switching sequence where the load current is commutated from a conducting diode to the opposing switch, it is mandatory to block the slow body-drain diode of MOSFETs and provide an alternative free-wheel path for the load current.
If this is not done a large current similar to shoot-through will pass down each bridge-leg at the instants where switches are turned on. This current can easily be high enough to destroy the MOSFETs and reliability suffers. The current spikes also create horrendous switching noise on the DC bus's stray inductance and on current sense signals, as well as being an EMI nightmare!
-Richie,
GeordieBoy, Thu Apr 08 2010, 12:29PM
As Steve said the slow and painful reverse recovery of the body-diode does not take place if it is "shorted" by the MOSFET channel. Instead it just hides for a while until reverse voltage is applied and then it rears its ugly head!
From what I remember, recombination is a slow process, so it takes applied reverse voltage to sweep out the charge carriers quickly. This is when you see the reverse recovery current spike.
Some IGBTs also exhibit this behaviour when operated in ZCS circuits. Even though the current falls smoothly to zero before turn-off, you still see a small current bump when voltage is applied to an IGBT that should be blocking.
For load currents with leading power factor, or any other switching sequence where the load current is commutated from a conducting diode to the opposing switch, it is mandatory to block the slow body-drain diode of MOSFETs and provide an alternative free-wheel path for the load current.
If this is not done a large current similar to shoot-through will pass down each bridge-leg at the instants where switches are turned on. This current can easily be high enough to destroy the MOSFETs and reliability suffers. The current spikes also create horrendous switching noise on the DC bus's stray inductance and on current sense signals, as well as being an EMI nightmare!
-Richie,
Re: SSTC mosfet reverse diode blocking- any useful?
Dr. Dark Current, Thu Apr 08 2010, 08:34PM
The circuit runs nicely and fails without warning again...
I had a heating element in series with mains input which limits the current to ~10A, and then a half-bridge of FETs with just two 1.5uF DC block caps, no filtering. There also was a simple overcurrent shutdown with a small TO-92 SCR (dunno how fast but I guess under 1us). I believe the coil was running slightly above resonance, so switch before current zero cross.
I think the only thing which could be killing my transistors is indeed the reverse recovery spike, gonna install those external diodes . . .
Edit: hell, maybe I know why my MOSFETs were dying... When tuning the coil I might have accidentally touched the pins on the frequency trimpot with a screwdriver, which sent RF feedback into the timing resistor and messed up the chip. I know this because I touched the trimpot pins again and the coil made funny sounds but luckily didn't die now.
Dr. Dark Current, Thu Apr 08 2010, 08:34PM
The circuit runs nicely and fails without warning again...
I had a heating element in series with mains input which limits the current to ~10A, and then a half-bridge of FETs with just two 1.5uF DC block caps, no filtering. There also was a simple overcurrent shutdown with a small TO-92 SCR (dunno how fast but I guess under 1us). I believe the coil was running slightly above resonance, so switch before current zero cross.
I think the only thing which could be killing my transistors is indeed the reverse recovery spike, gonna install those external diodes . . .
Edit: hell, maybe I know why my MOSFETs were dying... When tuning the coil I might have accidentally touched the pins on the frequency trimpot with a screwdriver, which sent RF feedback into the timing resistor and messed up the chip. I know this because I touched the trimpot pins again and the coil made funny sounds but luckily didn't die now.
Print this page