Fast power FET example?

[@rt]

Hi Guys,
I would like to drive a FET or power transistor of medium power (say 1 Amp at 12 Volts), that can operate up to 2MHz.
I have IRF540N here, and some darlington, but I’ve not got any of them to operate at that speed.

There is an example of a totem gate driver here for IRF540N:
http://joost.damad.be/2012/09/dimming-12v-led-strip-with-mosfet-and.html
The timing of rise and fall times it appears you could get a 2MHz signal, even if duty cycle is not perfect,
but everywhere else I read the IRF540 is only good up to 500kHz or so.

Would someone provide ay example of a circuit that can drive a 1A load at 2MHz?
The input PWM would be a micro or digital logic signal derived from a divided crystal frequency.
Cheers, Brek.

[tszaboo]

You just need higher current to drive the gate of the FET. 3-4A gate driving currents are normal for these switching speeds for normal TO-220 FETs. Or you decrease the capacitance of the FET. But unless it is really required, and for 12W it is usually not, lower switching speed should be used.

[@rt]

Hi, If I have a 3-4 Amp gate driver transistor then the problem is solved and I don’t need the gate driver,
and I can just use the transistors that were being used in the gate driver.

I have reproduced the circuit I linked to and driven it with a 3.57MHz crystal (7404 buffered),
then divided that by 2 with a decade counter. The output of the decade counter is correct,
and so is the output frequency of the totem gate driver. It’s the IRF540 that doesn’t switch.

[T3sl4co1l]

You don't want to use old transistors.  IRFxxx have been around practically since the beginning.  They have very high capacitance, especially drain-gate (which gives rise to Miller effect).

The 5V drive is also insufficient for a regular type part.  IRL540 should be used, which is logic level.  But it still won't be much faster, because the lower threshold means more charge has to be delivered over a smaller voltage range; and with the same series resistance available, even with a strong driver, you're limited in speed.  To get much greater speed from a given device, you have to goose the drive, with a higher voltage (say +12/-5V), or even with high voltage pulses (20 or 30V peak, decaying back to regular gate drive voltages, at a similar rate as the Rg*Cg time constant itself).

Modern devices have about 1/3 the total equivalent capacitance, rated gate resistance, and better Rds(on), so they are much better to design for.

As for the load: what are you doing, and why is 1MHz+ important?  You should analyze that ringing better.  There are excellent ways to deal with it, which aren't too hard to work with.  This deals with not only the ringing, but filters the high frequencies away from the load (not too important with LEDs, but important with loads like motors and Peltier devices), and most importantly, filters them away from radiating into space and disrupting communications from AM radio to air traffic control!

Tim

[tszaboo]

Hi, If I have a 3-4 Amp gate driver transistor then the problem is solved and I don’t need the gate driver,
and I can just use the transistors that were being used in the gate driver.

I have reproduced the circuit I linked to and driven it with a 3.57MHz crystal (7404 buffered),
then divided that by 2 with a decade counter. The output of the decade counter is correct,
and so is the output frequency of the totem gate driver. It’s the IRF540 that doesn’t switch.

That is exactly my point. To switch a FET fast, you need a lot of current, there are no workaround for that.
Just take a simulator. Put there a capacitor, and a signal generator, connect them together. Set the signal generator to output a trapezoid, with the required rise and fall times, frequency and gate drive voltage. Use one of the characteristic capacitance of the MOSFET as the load (like 500pF for the IRF540)
You will see very high currents coming out of your generator.
Either use a much smaller FET, or decrease the frequency.

[krivx]

You don't want to use old transistors.  IRFxxx have been around practically since the beginning.  They have very high capacitance, especially drain-gate (which gives rise to Miller effect).

I have been using IRFxxx for medium power stuff, including SMPS at 10s-100s of kHz, what would you recommend instead?

[mzzj]

I was going to tell you to take a look at IR's GaAs power mosfets but they seem to have disappeared

[T3sl4co1l]

You don't want to use old transistors.  IRFxxx have been around practically since the beginning.  They have very high capacitance, especially drain-gate (which gives rise to Miller effect).

I have been using IRFxxx for medium power stuff, including SMPS at 10s-100s of kHz, what would you recommend instead?

For those kinds of things, they're perfectly cromulent.  For maybe >300kHz and RF applications, they're on the slow side, and you should go shopping.

Rather than recommend particular examples, I'd suggest searching a distributor -- especially in the smaller SMT packages, you'll see some very nice, new parts.  I try to pare down search results by selecting the voltage and current minimums I need, and packages I expect, then selecting the lower third range of Rds(on) and Qg.  Between this, and packaging (cut tape / tube) and in stock filtering, I usually have 10-100 options to look through, and can check datasheets for specifics.

Tim

[Quarlo Klobrigney]

FQA11N90 http://www.mouser.com/ds/2/149/FQA11N90C_F109-109833.pdf
I know people have used these to 4 MHz. http://www.classeradio.com/rfvalues.htm
But I believe you will need a higher voltage in any case.

[honeycomb0]

I would strongly recommend to look for MOSFET driver IC's

This one just an example (its pretty old, but good):
http://www.irf.com/product-info/datasheets/data/ir2117.pdf

[ehughes]

Keep in mind that it is not voltage alone that turns on the MOSFET (Enhancement of the channel),   it is a total *charge*.      IF you can't supply enough charge in the time you are switching over, you will never switch (well).

I give a presentation on Class D Audio Amplifier once in awhile (see Attached).

See Slides 21 -26.

There are 3 primary contributors to loss in a MOSFET.     Conduction Loss are (Rdson) is the smallest at high frequency.

Some of the most important parameters when selecting a device

Qg - Total Gate Charge (Gate capacitance is wrapped up in this
Qrr - Reverse Recovery Charge
Cgoss- Drain to Gate  Capacitance and Drain to Source capacitance.


For fast switching you want to find a device with a low Qg and Qrr.       The Using Qg,  you can figure out how much current you need to switch at the rate you want to.   For your problem it will be on the order of amps.

There are several manufacturers of gate drivers (Micrel, IRF, etc).   that will work for your application.   The challenge is selecting a good MOSFET

[@rt]

I was going to look into this TI one: http://www.ti.com/lit/ds/symlink/tps28226.pdf
They claim frequency up to 2MHz.

[jdraughn]

Once I drove a TI CSD series 20V 5A mosfet with a PMD3001D and I drove the PMD3001D with a SN74LVC2G14.

The result is almost perfect waveform at 3MHz, and it only a fraction of cost compared to a proper gate driver IC.

Warning: these parts are tiny, and must be SMT mounted. Through hole devices just don't go to 2MHz without soft switching.

Did you parallel the two SN74LVC2G14 inputs and outputs?

[@rt]

I’m just hobbyist, it’s for general medium to high voltage “thing”
I want an optional micro controller board as an option for the 555 frequency generator.
This one has no totem gate driver, I did that on breadboard since.

[@rt]

Can I change it to (doesn’t have to be a FET).
I just assume that’s a fast way to switch a medium power.
Any Darlington pair I’ve tried behaves similar though, works at low freq, stops and/or breaks onto the MHz.