High Frequency PWM - Mosfet Gate waveform

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
I am looking for some advice on driving the gate of my mosfet.

I am currently using a voltage divider with a capacitor to help hold the dc level and sinking this to ground with an optocoupler to effect PWM of the gate for variable power delivery.
(the v divider carries 5mA and the gate charge of my particular fet is 210nC)

With my PWM frequency set to 2khz, i can see the square waveform and notice the smoothed rise (showing the charging of the gate capacitance). However the shape is still a square wave aside from the visible charging of the gate and slight knee indicating the miller capacitance.

This switches clean throughout the duty cycle and allows the fet to dip in and out of conduction without spending too much time in the linear region. The result is a nice cool mosfet.

Now the issue... when i crank the PWM frequency up to 20khz (to eliminate audible buzzing) i get alot of dissipation in the mosfet and device failure after a minute or so.

When i observe the PWMed gate waveform i can see what looks like shark fin with the miller capacitance knee and a relatively vertical drop to zero V - far from the square wave seen at 2Khz.

The gate needs to be driven more adequately to avoid the linear region.
I could try increase the current available to the gate to enable faster charging by getting higher rated resistors and altering the values of the v divider.
I know mosfet driver IC's are also an option.

Can anybody provide some guidance?

Thanks!
 

GopherT

Joined Nov 23, 2012
8,009
I am looking for some advice on driving the gate of my mosfet.

I am currently using a voltage divider with a capacitor to help hold the dc level and sinking this to ground with an optocoupler to effect PWM of the gate for variable power delivery.
(the v divider carries 5mA and the gate charge of my particular fet is 210nC)

With my PWM frequency set to 2khz, i can see the square waveform and notice the smoothed rise (showing the charging of the gate capacitance). However the shape is still a square wave aside from the visible charging of the gate and slight knee indicating the miller capacitance.

This switches clean throughout the duty cycle and allows the fet to dip in and out of conduction without spending too much time in the linear region. The result is a nice cool mosfet.

Now the issue... when i crank the PWM frequency up to 20khz (to eliminate audible buzzing) i get alot of dissipation in the mosfet and device failure after a minute or so.

When i observe the PWMed gate waveform i can see what looks like shark fin with the miller capacitance knee and a relatively vertical drop to zero V - far from the square wave seen at 2Khz.

The gate needs to be driven more adequately to avoid the linear region.
I could try increase the current available to the gate to enable faster charging by getting higher rated resistors and altering the values of the v divider.
I know mosfet driver IC's are also an option.

Can anybody provide some guidance?

Thanks!
Could you please post a schematic of your driver circuitry. Normally, 5mA is way too low. Also, you don't need any big wattage resistors to drive the gate. Current is flowing a short duration to just charge the small gat capacitance so a 1/4 watt or 1/8 watt (or even smaller) resistor is fine.
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269

Heres my temporary layout.
The resistor values have changed slightly to allow for 6V at the divider required to drive the gate.
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
I changed the Voltage divider but i am reaching the operating limits of my optocoupler (sinking 25mA).

I also managed to fry another mosfet when i switched on the transformer.
I guess the inrush is doing it?
 

bertus

Joined Apr 5, 2008
22,270
Hello,

The bleeder resistor (5k6) is not really needed as R1 and R2 will already be enough.
R3 will not be needed as R2 (2k2) is in parallel.

Bertus
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
Hello,

The bleeder resistor (5k6) is not really needed as R1 and R2 will already be enough.
R3 will not be needed as R2 (2k2) is in parallel.

Bertus
I can see that the bleeder isnt needed now, thanks for pointing that out.
Likewise with R3!
 

crutschow

Joined Mar 14, 2008
34,285
Cheers, ill look into it!
Would you suggest that over a dedicated driver IC?
I dont have any PNP BJT's so i will need to place an order either way.
Thanks !
If you need to order parts, then a dedicated driver IC would generally be preferable since it likely has better performance and fewer parts.
P.S. Also order a few general-purpose transistors such as 2N3904 and 2N3906 while you're at it, for future projects.
 

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
i dont suppose anyone can suggest an ammendment to the above schematic to allow for two mosfets to pass AC?
That would allow me to drop the bridge?
Take a look at something like this:
http://www.mouser.com/ds/2/149/FOD8320-108270.pdf
Much faster than the 4N26 and it has under voltage lock out which I think you need with your set up due to the time it will take for the voltage to come up. Or make sure the PWM isn't running when you power up the driver.
The push-pull output seems analogous to the totem pole driver above, Am i correct in thinking this?
Also, the under voltage lockout - how does one set the threshold voltage for this?
 
Last edited:

ronv

Joined Nov 12, 2008
3,770
Yes, the same as push pull.
The under voltage is usually fixed in the driver ic. But, different drivers will have different under voltage settings.
IGBTs often like 15 volts on the gate.
 

ronv

Joined Nov 12, 2008
3,770
Pretty sure. The 0 to 30 volts is under absolute maximum spec. which means it won't burn up. But to work it has to be over 13.5.
But your mosfet will probably work fine with 15 volts on the gate. Do you have it's datasheet.
I would just build a little 15 volt power supply with a 15 volt zener since you have 30+ volts to work with.
 
Last edited:

Thread Starter

urb-nurd

Joined Jul 9, 2014
269
Pretty sure. The 0 to 30 volts is under absolute maximum spec. which means it won't burn up. But to work it has to be over 13.5.
But your mosfet will probably work fine with 15 volts on the gate. Do you have it's datasheet.
I would just build a little 15 volt power supply with a 15 volt zener since you have 30+ volts to work with.
it can tolerate 20V between the base and emitter.
i may just go ahead with a zener and drive it at a higher voltage - as i have found a similar opto but the uvlo threshold is 8.3.
That will do me!
Thanks for the assistance
 
Top