MOSFET gate resistors

Thread Starter

tracecom

Joined Apr 16, 2010
3,944
I have always been somewhat puzzled about the need for a gate resistor on a MOSFET, especially when not using PWM or other similar drive signals. Here is an interesting explanation which seems to indicate that no gate resistor is required in such circumstances. It also indicates that gate resistors can be higher in resistance than often used. In the case of a microcontroller output used to gate a MOSFET, it would seem that a gate resistor in the 300 ohm range could serve to protect the uC in the event of a MOSFET failure.

http://irf.custhelp.com/app/answers/detail/a_id/215
 

jpanhalt

Joined Jan 18, 2008
11,087
Here are two notes from Fairchild (only the one in the link is still on its site:
https://www.fairchildsemi.com/application-notes/AN/AN-9005.pdf

They are not inconsistent with each other, but tend to slant the argument toward lower resistances. The ultimate answer is to use what works. If inductive ringing is not a problem, then you can improve efficiency by using a smaller resistor.

John
 

Attachments

#12

Joined Nov 30, 2010
18,224
I believe you have it correct. The first consideration is, how fast?, the second consideration is about ringing, and the third is about protecting the driver.
 

wayneh

Joined Sep 9, 2010
17,496
Or you might say the first 3 considerations are, how fast, how fast, how fast. I mean, if the switching is infrequent and using a 1k resistor does not cause a problem with switching time, you're done. That covers a lot of applications - at least around here. It'd be nice to have a rule of thumb frequency, below which you can forget about needing to optimize the gate resistor. I would suggest that 1kHz is close? I haven't done the calculations but I'm guessing that even the old IRF540 could be switched quickly enough using a 1k resistor at 1kHz. At 10kHz, I'd certainly expect to start making calculations and reducing the gate resistance.
 

#12

Joined Nov 30, 2010
18,224
Just playing around with the IRF540 specs, firing 20 volts through a 4.7 ohm resistor into 870pf raises the gate by 10 volts in 2.83 nanoseconds...
BUT the datasheet also says the rise time of the transistor is 45 nanoseconds.
If your combined rise and fall time is 90 nanoseconds, that would be 10% of 1.11MHz.
That looks like 1.11MHz the high limit for an IRF540 which spends 10% of its time in transition.
Change the resistor to 1K and the rise time is 603 nanoseconds and 82.9 KHz is the frequency where the chip spends 10% of its time in transition.
If you don't want to hit it with 20 volts (max) that will slow things down again.

Just playing with the numbers.
 
Top