mosfet driver needed for logic-level mosfets?

You need to do some (gasp!) engineering. You need to know how quickly the FET needs to turn on and off, you need to know the FET's total gate charge, and you need to know how much current the processor pins can supply.

For slow-enough PWM, no, you don't need a driver. Otherwise, yes, you do

-- and you want to be careful to get a driver that has close to rail-rail output. Some of the oldy-goldie ones have bipolar drive sections with considerable (up to 1.5V) drop from supply voltage to gate drive. So -- even more engineering.

Given that you are also asking about BJT vs. FET, note that in order to switch at it's fastest, a BJT needs a base driver that can push charge in quickly and actively pull it out quickly. The difference in recovery from saturation when you pull the base below ground vs. when you just pull it _to_ ground can be dramatic.

And don't use a grotty old TIP31 if you want speed -- Zetex developed some pretty impressive modern BJTs before they got bought by Diodes, Inc., and turned into a cash cow. Those transistors are still sold, and have some pretty impressive HFE_sat and VCE_sat numbers.

Ok, thanks for the information. You've given me lots to think about.

For higher-speed switching of BJTs, you recommend yanking the base to *below* ground, then?

Which EE classes are these topics discussed in? Can you recommend a book or two?

Thanks,

Michael

Yes. But not too far... ~3V maximum to avoid VBE breakdown.

...Jim Thompson

Oh cool. Thanks.

Michael

Yea verily.

None that I know of, and no, in that order. That doesn't mean that someone else may be more helpful -- just that I don't know.

I learned the technique from Zetex applications notes about 20 years ago; I used it in a prototype and never used it again.

You can get a nicely controlled negative-going spike at the leading edge of your base drive with an RC network from your driver, or, if you want nice behavior over a wider range of duty cycles, a resistor-diode-cap network. It's all complercated and I can't remember the details and by God I hope there are still app notes floating around.

The basic idea is that if you just pull the base to ground then the residual charges that are in the base making it conduct have to recombine, and you have to wait. If you yank them out by force then -- they're not there any more.

circuit is something called the Baker clamp

"Practical Electronics for Inventors" is a good basic electronics book, (and the price is right.) after that go to AoE3.

George H.

Huh. I have to check out Zetex. Thanks.

You might have to write another book! I took a peek at your Controls book and I had no idea Scilab was free. A free Matlab clone, huh?

Michael

Cool! Thanks!

Michael

Impressive! Thanks!

But by avoiding saturation, Vce,sat is not minimized, is it..?

Michael

A Baker clamp minimizes recovery time by preventing hard saturation and charge storage... "Vsat" is now typically equal to Vbe.

But, as shown by "default", it also prevents quickly sucking charge out of the base... more complex arrangements are required. ...Jim Thompson

Check out all the variations... Seems to me I remember another Baker switch that returned the base to a negative voltage to suck the voltage down faster - but it was Fred Baker, from Motorola, not Richard, and this was all state of the art stuff in the 70's, to drive plasma panel displays at very high refresh rates. For merely driving a motor with PWM chances are you don't need to get fancy at all.

I had an application (triggering a camera to power up take a pix then go back to sleep) running on flea-power batteries. I used one I/O pin of the controller to run a simple 20 KHZ inductor flyback booster, to raise the 3V battery supply to 12V, to just provide mosfet bias - so I didn't need a logic level device and it worked very well.

I clamped the voltage at 12V but it would eventually get to ~50V if I didn't have a load or clamp the bias power supply.

BETTER than Matlab. For a certain meaning of "better" (it's not compatible), but I like it.

People keep trying to tell me to use Python for scientific computation, and I keep stabbing at learning it, only to gag at learning yet another interpreted language (and, I'm sorry, indentation should NOT tell the compiler what the program structure is -- that's simply nuts).

That's not at all what I was referring to. Zetex (or somebody) figured out that you could go ahead and let the transistor saturate, but still get significantly faster turn-off if you actively drew the carriers out of the base.

As a heavily Algebra-oriented guy I find that I can simply write the equation in a PSpice behavioral block and solve faster than I can think bass-ackwards programming ;-) ...Jim Thompson

On Mon, 15 May 2017 12:37:45 -0500, Tim Wescott wrote:

Just for chuckles, here's a demonstration circuit (very much demonstration only). Set the "inital" voltage on V2 to 0 and compare the transistor turn-off time: I get about a 10:1 difference.

Version 4 SHEET 1 880 680 WIRE 240 16 -160 16 WIRE 240 48 240 16 WIRE -160 112 -160 16 WIRE -32 144 -64 144 WIRE 80 144 48 144 WIRE 160 144 144 144 WIRE 240 144 240 128 WIRE 368 144 240 144 WIRE 592 144 432 144 WIRE 656 144 592 144 WIRE 768 144 736 144 WIRE 240 176 240 144 WIRE 592 192 592 144 WIRE -160 224 -160 192 WIRE -64 224 -64 144 WIRE 32 224 -64 224 WIRE 160 224 160 144 WIRE 160 224 112 224 WIRE 176 224 160 224 WIRE 768 240 768 144 WIRE -64 256 -64 224 WIRE -64 368 -64 336 WIRE 240 368 240 272 WIRE 592 368 592 256 WIRE 768 368 768 304 FLAG -160 224 0 FLAG 240 368 0 FLAG -64 368 0 FLAG 592 368 0 FLAG 768 368 0 SYMBOL npn 176 176 R0 SYMATTR InstName Q1 SYMATTR Value 2N2222 SYMBOL ind 224 32 R0 SYMATTR InstName L1 SYMATTR Value 120? SYMBOL voltage -160 96 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 5 SYMBOL voltage -64 240 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V2 SYMATTR Value PULSE(-2 5 0 10n 10n 7u 10u) SYMBOL cap 576 192 R0 SYMATTR InstName C1 SYMATTR Value 1? SYMBOL res 752 128 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 50 SYMBOL zener 784 304 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D1 SYMATTR Value BZX84C10L SYMATTR Description Diode SYMATTR Type diode SYMBOL res 128 208 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R2 SYMATTR Value 220 SYMBOL schottky 368 160 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D2 SYMATTR Value MBR0530L SYMATTR Description Diode SYMATTR Type diode SYMBOL diode 144 128 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D3 SYMATTR Value MBR0530L SYMBOL res 64 128 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 10 TEXT -82 392 Left 2 !.tran 100m

~kapow~

Is the app note somewhere in here? It's not exactly organized by Zetex, but maybe you remember parts of the name of the app note..?

It's too bad that companies don't always keep the old website when they get bought out...

Michael