DPRG List  

[DPRG] Power MOSFET drive currents

Subject: [DPRG] Power MOSFET drive currents
From: Chuck McManis cmcmanis at mcmanis.com
Date: Sun Feb 9 14:00:02 CST 2003

A common mistake Randy, you just have to go one level deeper and you're there.

Now do the calculation for 40KHz, 80Khz, 160Khz, and 320Khz. These are the 
even harmonics for a square wave. You can take it right up to infinity if 
you want. The higher you go, the "squarer" the wave form gets.

The goal is to get the FET on _right now_ and to that end you want as close 
to a square wave as possible going into the gate. There are a number of 
parasitic effects that go on (the miller capacitance for one) but the goal 
is to get the FET fully enhanced in the shortest possible time to mitigate 
switching losses.

So if you have a FET like the IRF1405 (a popular one with the OSMC crowd 
and others) the FET has a turn on time of 18 nS and a rise time of 140 nS 
so any time you get better than 140nS to full enhancement your in good shape.

To get there however you can compute the current in a different way. The 
amount of charge that is needed on the gate to fully enhance the FET is 
180nC (nano-columbs) and you'd like to have them there in 18 nS (so that 
the gate can switch on as fast as possible). Recall that current is simply 
columbs per second so 180nC/18nS is 10 nC/nS and if you multiply both sides 
by a billion you get 10C/sec or 10 Amps. Not that you could get 10 amps 
into the gate without serious magic, if you get 2 Amps into the gate (which 
is what I shot for with my design) you get a 2nC/ns or 90ns to get 180 nC 
into the gate for a switch on time of 90nS. If you look at the scope 
pictures on my web page you can see that with the IR211O you can get there. 
It gets worse however the more FETs you drive (more columbs to push in the 
same amount of time) and this is the limiting factor on 64 fet per leg 
monster speed controllers :-) (I talked to a kid who was trying to build one!)

The challenge is that while you can think of a gate as a capacitor for a 
large number of things, this isn't one that lends itself well to that.


At 12:22 PM 2/9/2003 -0600, Randy M. Dumse wrote:
>Hey Guys,
>I wonder if anyone could clarify something for me. I figure a
>typical power fet might have a gate capacity of a few thousand
>For reference I lookd at IRF1405 and found input capacitance to
>be typ. 5480pF
>So if a 20KHz drive signal were applied, a current through the
>gate would see a 1.45Kohms reactance.
>For reference I use the calculator at:
>So with a 10V gate drive, the current through the gate would be
>6.89 mA.
>Am I thinking about this right? If so, why are the MOSFET
>drivers like the TC4244 rated for multiple amps (yet have high
>output resistance)? Can't see that much current capacity is
>needed or desireable, unless this is to handle peak currents at
>the gate is drained of a couple hundred nanocoulombs in a few
>nanoseconds. Am I missing something?
>DPRGlist mailing list
>DPRGlist at dprg.org

More information about the DPRG mailing list