Paralleling Gate Drivers

[Mosaic]

Hello:
I need to drive 4, IRFP 3306s' NFETS in a 1Khz switching app, 20uSec pulse width. I have MCP1407 Mosfet Drivers.

I was wondering if I could parallel those drivers to drive the 4 Paralleled Fets?

They are independent parts and may not sync properly. Would adding Schottky diodes in line with the gate drive outputs protect against this imbalance and still benefit the turn on time of the FET bank? That would prob reduce the turn off time though. Perhaps a 1k to ground at the gates will keep that reasonable.


As the Total gate charge is perhaps 90 nC per Fet...and the peak current at 12V is about 4Amps from the MCP1407....then with 4FETS I should realise a 90nSec charge time for the combined gates?

If so I can just use 1 MCP1407 driver as that is fast enough.

 

[jpanhalt]

Deleted -- tried to edit and it duplicated

 

[jpanhalt]

I have paralled 5 mosfets with a single driver. Use a small resistor on the gate of each. Keep the resistor close to each gate. I used a "dead bug" or "Manhattan" type of construction for it. Will post picture, when I can find it.

Here's an article from Fairchild on resistor selection: https://www.electro-tech-online.com/custompdfs/2013/04/Fairchild_AB-9.pdf

Here's the picture:



John

 

[Mosaic]

I have some 1 ohm, 2.3 ohm and 5 ohm resistors flame (retardent) on hand (u use 1.5ohms?)....is that good enough for the gate resistors? I also have some ferrite beads I could use, they're rated @ 500mA but i could parallel them if required, although the duty cycle is only around 1-2%.

I am mounting the Fets on an aluminium finned heatsink 120mm square with a 120mm fan forced air cooling on the fins. The drain/tab of the devices can be bolted to he heatsink and I can attach a 4 awg lug output to the heatsink. The switching current can pulse to several hundred amps.

 

[jpanhalt]

From the link already provided:

Use a 4.7Ω gate resistor for every power MOSFET used in a switching converter; in particular, use a separate 4.7Ω gate resistor for each paralleled MOSFET.

It is a good discussion and well worth reading.

IR (and other vendors) have lots of application notes on mosfets. When I designed the device I showed, I relied on IR's notes and used larger gate resistors than were probably necessary. I did not have a good oscilloscope at the time. I chose the Fairchild note, as it gives efficiency information related to gate resistor choice and tends to suggest lower values. The efficiency lost by using a slightly larger gate resistor (e.g., 3.3 or 4.7 Ω) is trivial. Also note, that the resistor will tend to balance turn-on time much like resistors in LED chains balance current.

BTW, if your aluminum heat sink is anodized, remember that the anodized coating is an electrical insulator. If you are depending on it for electrical contact you may want to remove the coating from a small area.

John

 

[Mosaic]

I read the note...it's very useful...I am looking at other notes as well..They also mention zener clamping the Vgs based on Vds transients due to inductive loads. My load isn't particularly inductive, perhaps 100nH.

What do u think of the ferrite bead vs the gate resistor?

 

[tcmtech]

Given your rather low switching frequency any resistance combination that keeps the theoretical peak currents under 4 amps would work just fine.

 

[misterT]

Given your rather low switching frequency any resistance combination that keeps the theoretical peak currents under 4 amps would work just fine.

I agree with your point, but switching frequency and switching speed are two different things (switching speed limits switching frequency). Even if your switching frequency is low you might want to switch as fast as you can for efficiency.

 

[Mosaic]

The application requires the fastest switching transient times. It's not about the base frequency in this case.

 

[jpanhalt]

I agree with your point, but switching frequency and switching speed are two different things (switching speed limits switching frequency). Even if your switching frequency is low you might want to switch as fast as you can for efficiency.

Agreed. One might equate unacceptable switching losses to the anticipation of high currents in this case! Since the discussion is of paralleled mosfets, I assumed that the TS (aka Thread Starter) was interested in high currents. The device I showed controls a 2HP DC motor at 12V.

There is also a need to "balance" the mosfets' turn-on times. I think Mosaic is focusing on ringing with his thought of using ferrite beads.

John

 

[tcmtech]

Well if it switching efficiency he is after then it may be best run a dedicated driver IC for each switching device.

To be honest I would be curious to see what the actual numbers say on the efficiency losses between theoretically perfect with a dedicated driver for each device Vs one drive split between all devices with a 4 amp gate drive current limit in place.

So what are you running that needs some 480+ amps switching capacity at a 20 uS PWM any way?

 

[Mosaic]

I am concerned about gate ringing and I'd love to know more about 'balancing' the fets without adding source resistance , but given the fast turn on, I don't think the balancing is that critical...