Ebable function of IXDD414 gate drivers / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Ebable function of IXDD414 gate drivers

Move Thread

LAN_403

Marko

Thu Jun 26 2008, 03:12PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi all,

I'm looking into building a MOSFET H bridge inverter with short circuit and overcurrent protection, using a current sense transformer and logic to quickly shut the drive off until power is cycled. I plan using 500V mosfets in range 14 to 50A.

But, one thing I noticed possible is latch up of power devices if dU/dt gets too high due to fast gate drive in short circuit condition, where drain currents are extremely high;

I've read it's actually better to slow down the turn-off in this condition.

IXYS

is explaining some of this for their IXDD414 drivers, which I'm going to use, so I wonder how exactly should I design my circuit.

Still, some of their figures don't make sense to me: they say I should dimension the resistor to have time constant of 100us with *miller* capacitance ??, which would be in megaohm range for mosfets like IRFP450 and not really sensical.

They use 1600ohms for their large mosfet, and I'd expect something around that value.

Also, since I will be using a GDT, it's impedance wold probably dictate the turn-off.
I'm not sure if the whole thing would be possible with GDT actually.

It would also not be possible to use this pull down mosfet, but I don't really acre about just leaving the resistor connected and driving the IXDD414 into high impedance state if overcurrent occurs.

So is this possible, and worth of effort, or should I just use the old fashioned instant-off protection style?

I've done that before with nothing more than CT and small SCR and protected IRFP450 halfbridge from dead shorts on the output.

So, any suggestions?

GeordieBoy

Thu Jun 26 2008, 03:40PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

The rate at which the device current rises at turn-on in a switching circuit is normally limited by leakage inductance of a transformer, buck choke, or motor or whatever the H-bridge is actually driving. It is quite hard to protect an H-bridge from a bolted short right across the bridge legs because there is very little inductance to limit the rate of rise of current.

Essentially by the time your CT has sensed the current is over the limit and has passed the message on to the drive circuitry to cut off the gate drive, and the GDT has finally discharged the MOSFET gates, the drain current is probably already sky-high.

If your design must withstand this later example, I would look at adding artifical inductance to slow down di/dt with a bolted short. Also look at only enhancing the switching devices as far as you need to support the expected load current under normal operation. This achieves two things. Firstly you don't have to remove as much gate-charge if you want to turn them off in a hurry. And secondly this will limit prospective short-circuit current, and then use de-saturation detection on the power side to detect the fault condition and quickly remove gate drive.

-Richie,

Marko

Sun Jun 29 2008, 10:59PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

GeordieBoy wrote ...

The rate at which the device current rises at turn-on in a switching circuit is normally limited by leakage inductance of a transformer, buck choke, or motor or whatever the H-bridge is actually driving. It is quite hard to protect an H-bridge from a bolted short right across the bridge legs because there is very little inductance to limit the rate of rise of current.

Essentially by the time your CT has sensed the current is over the limit and has passed the message on to the drive circuitry to cut off the gate drive, and the GDT has finally discharged the MOSFET gates, the drain current is probably already sky-high.

If your design must withstand this later example, I would look at adding artifical inductance to slow down di/dt with a bolted short. Also look at only enhancing the switching devices as far as you need to support the expected load current under normal operation. This achieves two things. Firstly you don't have to remove as much gate-charge if you want to turn them off in a hurry. And secondly this will limit prospective short-circuit current, and then use de-saturation detection on the power side to detect the fault condition and quickly remove gate drive.

-Richie,

OK, I see.

In first case, I thought it would be the desaturation of the devices that would limit the current, but the overcurrent condition is really what is important.

So, I should really just turn off as fast as possible in any case, and ignore this enable function and slow turn-off mode?

Marko

Avalanche

Wed Jul 02 2008, 08:59PM

Registered Member #103 Joined: Thu Feb 09 2006, 08:16PM
Location: Derby, UK
Posts: 845

From what I've seen, the slow turn-off thing is generally only used when Vds(sat) overcurrent protection is used, this type of overcurrent protection is very fast to repond because a comparitor is used to keep an eye on the saturation voltage of the device. Once it trips, all the devices in the bridge should be subjected to the slow turn-off, via fast optos, or whatever.

In your case, I wouldn't bother with slow turn off, because with the added delay, by the time you broadcast the slow shutdown signal, the current will already be high enough to probably damage the device. As long as you have a sensible gate drive (ie not stupidly fast) and snubbers across drain and source, you should get away with switching off the devices normally.

Look into Vsat overcurrent detection if you feel like it, it's usually only found in modular drivers but you can knock something together with discrete parts if you want. It is fun to throw a spanner across two phases of a 10kVa inverter and see nothing but an LED come on

Steve Conner

Wed Jul 02 2008, 10:07PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Sounds like fun untill the boss catches you doing it

I looked into desaturation protection back when everyone was playing with IGBT bricks. The problem was that we drove them at way higher frequencies than they were meant for, and the desat circuit couldn't tell the difference between the device voltage drop, and the voltage induced by di/dt in the package stray inductance. I was measuring about 100V worth of L*di/dt when I put enough current for a Tesla coil at 60kHz through the Powerex bricks, and the device volt drop is maybe 3V.

In the intended application, motor drives and inverters, the big bricks work great with desat protection. They are actually specified to withstand a bolted short with full gate drive for a couple of microseconds, long enough for the desat circuit to kick in and shut everything down. And by desaturating, they clamp the current and stop it from ramping any higher. So the devices and protection circuit work together to make it pretty bombproof, as Avalanche found out with his spanner. "Vsat" rises to essentially the whole DC bus voltage when the spanner is applied, so maybe desaturation detectors would have worked for me after all. But in the end, I decided to just limit H-bridge peak output current with a CT, and not let anyone with a spanner near the thing.

The Miller capacitance is bigger than you think. It's probably best worked out by looking at the size of the Miller plateau on the datasheet gate charge vs. gate voltage chart.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob