IR21844 MOSFET gate Driver

[darkfeffy]

Hello all,
Practically speaking, how many n-channel MOSFETS can one (1) IR21844 Mosfet gate driver drive at once? four? five? (in parallel, of course)
Thx
Ed

 

[crutschow]

Not enough info.

It depends upon the switching speed you need and the gate capacitance of the MOSFETs. The IR21844 has a specified amount of gate drive current. The more capacitance it has to drive the slower it becomes.

So you need to specify the speed required and the size of the MOSFETs you want to drive.

 

[darkfeffy]

irfp360 n-channel mosfets (23A max at 400V). I want to know if I can drive 5 mosfets (in parallel) at around 100kHz with one single driver.
Can this go?
Thanks
Ed

 

[smanches]

It should work fine to drive 5 of those MOSFETS. Each FET is going to need about 21mA at 18v, and the driver can supply 1.4A. If you're going to drive at a lower voltage though, you'll have to increase the current by the same amount.

I = Q / Dt = 210nC / ( 1 / 100000) = 21mA @ 18v. Someone check my math, it's still morning...

 

[Speakerguy]

No math check needed, it should be fine. 100khz is pretty low. And five @ ~23nF total combined Ciss should be OK. Maybe put in some gate drive resistors just to keep peak current down. They might not switch quite as fast as they might otherwise but we should still be talking ~100ns or so. Unless I'm wrong

 

[smanches]

Now that I've had coffee, I see I was wrong. Needed to use the turn-on time, not the frequency in the calc. I knew something seemed wrong.

I = Q / Dt = 210nC / 79ns = 2.66A@10V (<- EDIT: IR did switching speed tests at 10V, not 18V)

That seems more like a proper drive current, although at max switching speed. This is not the switching frequency, just how fast it turns on. As it is, you could not drive a single FET at full speed with those drivers.

100Khz is 10us period. There is a total of 146ns for both rise and fall times at that current. That's 1.46% of the total period, so you have a little room to work with, but not a whole lot.

Anyone have an idea about at what percentage you're switching too slowly for the given frequency? Faster is almost always better in power circuits, but it would be good to know what the limits might be.

 

[Speakerguy]

Anyone have an idea about at what percentage you're switching too slowly for the given frequency?

I don't know if there's any real guidelines, other than "don't burn up too much power in the output stage". Since the switching time is the only time the output stage is dissipating power, 'faster is always better' and the only real metric is how much heat you're dissipating into the heatsink. If you are running at 24V a certain switching time might be OK, running at 340VDC might toast your MOSFET's with the exact same drive setup since switching losses are now more than 10x due to voltage.

My only other concern might be power dissipation in the driver itself. Those little PDIP chips can't handle even 2W. I'd have to do some maths to check and see if it's ok.

 

[crutschow]

The switching speed affects the power dissipated during the switching time (it's roughly ½ V x I x Tr x f). If the transistor is switching high current/voltage then the switching losses for a slow switch time could be a significant portion of the total transistor dissipation (the transistor "ON" resistance generating the rest of the dissipation). For minimum dissipation you would want the gate rise time to be faster then the transistor rise time.

So we need to know the voltage and current that each transistor will be switching, as well as the desired efficiency, to determine how fast it needs to switch.

 

[darkfeffy]

Thx alot guys;
Very helpful remarks. I finally reduced the switching speed to around 69kHz. Seems to work just fine with 5 mosfets in parallel at no load. When I will feel more comfortable, I will drive these switches right upto 100kHz as my inductors are not that rich in inductance.
Furthermore, by running the converter (actually a buck and later on a boost) on full load, I do not expect the PWM controller (a UC3823n) to loose control
Cool forum...