Switching frequency large gate charges

I have been using a circuit like the attached pic for low powered h-bridges.

My problem is speeding up the switching times for large gate charges or multiple mosfets.

If i use too many Mosfets in Parallel the circuit starts behaving like an on/off switch unless i lower the PWM frequency, which, to me means the switch is turning on but not discharging before the next pulse so it acts like its always on, nothing gets hot.

I know i can use gate drivers for this but i have lots of discrete components.

So my questions are:

When the gate is switched off is the gate charge all discharged through the 10k resistor to ground?

Can i reduce the gate pulldown resistance to speed this up?

You are not turning the devices off actively. Only the 10k resistor is available to discharge the gates.

Go buy a low-side MOSFET driver chip like a MIC4422, you won't get a cheaper simpler solution to heavy switching. With multiple MOSFETs you need a resistor per gate, perhaps smaller values will do like 22 ohms. With a single MOSFET no resistor is needed (it prevents differential oscillation mode).

Gates are basically large value capacitors (1nF to 30nF or so) and need large currents (100mA to 1A is typical) to switch snappily, and this applies equally to switching off as on.

Ok, Thanks for the tip on the driver i will use those in the future for sure, it would take a fair amount of either time or money to get one to where i am unfortunately. Will do some experimenting with changing values with what i have for now..

Failing that use a push-pull driver to the gates, ie class-B - that's just an NPN and PNP.

Thanks again, a push pull gate driver would do the trick. I changed the pulldown resistor to 150 Ohms and the gate resistors to 25 Ohms. This is really toeing the line however. The pulldown resistor is dissipating about 0.5 watts and gets pretty hot ( using two 0.5watt resistors in parallel). The voltage drop is acceptable, the gates still get about 9.2 volts. I can go to about 500hz pwm only with this setup.

Should i make the push pull driver in an emmitter follower configuration? It seems to me with some brief reading that this will create a "dead space" during switch time and reduce shoot through. I have been wrong before though.. a lot.

I made a test driver from some irf1404's (5 in parallel). This gives a pretty good gate charge (800nc) to charge/discharge. Next I have to convince the wife that I need an oscilloscope to see what's going on. I will post the final schematic when I have done some more testing.

If you use a push pull driver, make sure you use matched pair transistors designed for this. A MOSFETdriver,as mentioned, may be the better solution. Maybe: http://www.farnell.com/datasheets/1863842.pdf

Thanks Larry, I agree! I MOSFET driver is a better solution.. I have the access to lots of parts so I am going to build one. As it stands the motor controller is useable with just the pull down resistor and it's cheap. A two dollar MOSFET driver would add 33 percent to the cost of the project lol! Good idea about the matched transistors.

For anyone interested this is the circuit i ended up using to drive the mosfets. Tested up to 20khz using 5 irf1404’s in parallel. I used these transistors for the push pull part of the gate drive.


Attached is the gate drive schematic and a pic of the test circuit. The driver should be ok up to about 100 amps 24 volts providing the wiring can handle it. Can maybe handle small scooters and the such. I don’t know, haven’t tested it much. Not bad for a 6 dollar driver though.

If anyone notices problems or improvements i should make to the schematic fire away. I don’t want to give anyone bad info. I am just learning.

No decoupling capacitors on the gate driver??? You need good stiff ceramic decoupling (0.1uF + 10uF say) - thats where the gate charge comes from in the short term (remember you need perhaps an amp in 100ns or so, thats 10^7 amps/second which ain't going happen without proper decoupling right on the driver 12V rail.... If you can get access to a 'scope do - that'll tell you so much about what's actually going on.

Also its more important to drive the gate low fast than high(*), so using the PNP pre-driver means it won't switch off as quickly as its switches on, only the 1k resistor is available to pull the class-B driver stage low. I would use a single logic-level n-channel MOSFET predriver, with 1k ohm pull-up.

(*) to reduce dead time and because the threshold voltage is much closer to 0V than 12V - you want fast transition through the plateau to threshold range. With a class-B driver you can only pull down to about 1V anyway, making the switch-off-speed slower than with a gate driver that uses MOSFETs internally.

As I said MIC4422 will do the job better and cheaper.

decoupling caps are there… sorry ,not shown in schematic. you can see them in the picture. i am using a 0.1uf ceramic and a 10uf electrolytic on the 12v rail.

I wanted the gate tied low to be off if there is signal loss. Hence the pnp pre driver, I’ll change to npn and invert the input. Mark is right, it does slow down the turn off speed. I will fiddle with it a bit more. I’m using ltSpice to run sims now, really helps to visualize whats happening with the circuit.

I hear you about the mic4422 Mark… i do… i’m in too deep now though lol. Only cost about 50-60 cents so far for the gate driver…
Thanks again for your help