driving mosfets

Graham Stabler

I was wondering if for moderate switching frequencies (say 4khz) a mosfet driver should be used between the prop and the mosfet, might be worth it anyway due to low gate drive voltage but I'm not sure. I'm switching 0.7A through the fet, as yet not selected.

Graham

Bongo

Yes, a driver is better.

1) You can use a standard (easy to get and cheaper) device rather than a low gate voltage device
2) Gate Input capacitance requires some driving. For example, you will probably find that the device gets hot if you drive the gate via a 10k resistor, and runs well with a 200 ohm gate resistor. So the driver should be able to drive 200 ohms at 10 volts for common mosfets. A small pnp transistor is good.

bongo

Leon

I've got away without a driver with a small stepper motor and a small DC motor with MOSFETs driven from a 3.3 V MCU. It was a low gate voltage device. However, if you read the data sheets for those devices closely, you will find that the low gate voltage is only valid for a very low current (like 250 uA), and a driver should be used.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 6/19/2008 12:23:05 PM GMT

Graham Stabler

I've used this mosfet a lot for model aircraft stuff:

www.irf.com/product-info/datasheets/data/irlml2502.pdf

Tiny on resistance even with low gate drive.

But it's not convenient for general use due to the tiny package, I'll dig for something similar for through hole.

Graham

Post Edited (Graham Stabler) : 6/19/2008 3:12:08 PM GMT

Leon

Those look useful, I'll see if I can get some. Farnell and RS stock them.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 6/19/2008 2:25:24 PM GMT

Beau Schwabe

Graham Stabler,
·
At one time I came across an interesting MOSFET in a TO-92 package capable of driving 5 Amp loads never breaking a sweat.· I'll have to dig around, but I think it was an A16 or A76, or something.· The MOSFET had an extremely low Rds"ON" which meant that it would take quite a bit to develop enough voltage across the S/D junction to amount to any heat dissipation.· As I recall, these were logic level MOSFET's that were virtually indestructible.
·

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

parsko

Is it this one?

This link also seems to be helpful!

Post Edited (parsko) : 6/19/2008 4:56:48 PM GMT

Beau Schwabe

parsko,

No, 40 mOhms is still too high... I think I remember this thing being in the single digit milli-Ohms.· I'll look this evening.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

parsko

Beau,

Did you see the other link? It looked like there were some that were in the single digit milli-ohms...

-Luke

Philldapill

Beau, that's impressive! I still have a few HUF76145P3 that I bought in bulk($50 for 800). They're something similiar except 75A, low gate voltage, and 4.5mOhms ON-Rds.

Graham, I never thought gate drivers were all that until I tried one. I went from about 9uS rise time to something like 500nS rise time. They are definately worth it when your freq. goes up since they will spend more time in the partial-conducting state if you don't use one.

I got a few of these drivers from TI as samples. They are 9A peak and work very well. You can get them in a PDIP package for prototyping.
http://focus.ti.com/docs/prod/folders/print/ucc27322.html

Beau Schwabe

parsko,
My head hurt turning it clockwise like that... lol
I did not see anything in a TO-92 case, although a TO-220 isn't bad.
Point is, is that the smaller the transistor, generally the smaller the gate capacitance, and if you can get one with plenty of drive above what you actually will need, then that is a bonus.· Another important aspect that is often overlooked is the Rds "ON"... This value has gotten lower and lower over the years with improvements in silicon technology, resulting in·smaller processes.· A transistor with a 180nm gate technology, can actually handle·the same amount of·energy in 1/4 of the·area than·the 350nm technology predecessor could·... this is purely a function of IR drop across the S/D transistor junction.· From 350 to 180 you have·about a 49% reduction in·gate channel·width, and because of this, the transistor only needs to be about half as long reducing it about 50% further.· This results in a 180nm process·only needing to be about 24% the size of a 350nm process for an equal drive transistor between the two technologies.·· As an added benefit, the gate capacitance is also reduced by about 1/4 (<-in this example).


▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 6/19/2008 5:48:40 PM GMT

parsko

Graham,

According to the following statement, one should use a driver when used with a ucontroller (or am I reading this wrong??):

http://www.best-microcontroller-projects.com/logic-level-mosfet.html said...
If you disable the gate control voltage e.g. if you turn the microcontoller pin that controls the logic level MOSFET gate to high impedance (i.e. turn it into an input) then you can have a problem. Because of the extremely high input impedance any induced charge at the gate will build up a voltage e.g from noise generated by cross talk signals. Since The charge can not leak because of the ultra high input impedance of the gate at some point the voltage will build up enough to turn the FET on! The 'standard' solution is to use a 1MOhm resistor to ground to leak the charge away.

Note: This could also be a problem for critical circuity at power up where most microcontrollers hold their pins as inputs i.e. safe until configured internally. There are some chips available that ensure fail safe operation

Leon

I use a pull-down resistor on the gate.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Mike Green

This doesn't say that you should use a driver, only that you need some kind of pull-down resistance if the MOSFET gate is connected to a high impedance driver some of the time (like during initial conditions).

Phil Pilgrim (PhiPi)

The only time a driver is needed is either 1) if you need a higher gate voltage than your logic output provides, or 2) if the gate charge requirements exceed what the logic output can provide at the driving frequency. The idea is to minimize the percentage of time that the MOSFET spends in its linear region (i.e. with R_DS greater than minimum and less than infinity).

-Phil

Post Edited (Phil Pilgrim (PhiPi)) : 6/19/2008 7:59:54 PM GMT

Graham Stabler

In that case as I'm only flashing some leds at 0.7A and 4khz so I'll probably be OK I suspect.

Thanks for all the help.

Graham

Phil Pilgrim (PhiPi)

Graham,

The proof is to touch the MOSFET after it's been running awhile to see if it's hot. Empircism trumps theory any day! (Or, put another way, nature makes a great analog computer.)

-Phil