John Larkin's fast 50V pulse with a 2n7000.

The fets are pretty much infinitely fast inside, so it's just a matter of gate drive. We're driving them with a bunch (emphasis bunch) of fast TinyLogic gates. I did another circuit a couple years ago that drove a pair of 2N7002's with a gaasfet... +12 volts into the gates in about 400 ps. Turnon speed was awesome, but turnoff drive was passive so the falling edge was rotten.

This is the HV option on our benchtop pulse/delay generator, so we can make the pulse width anything. Low end, it starts to lose swing at about 3 ns width; for wider pulses, the transformer saturates at something like 1 usec at 50 volts, 10 usec at 5 volts.

Nice looking pulse, huh? It almost looks faked.

John

I assume you're switching one 2n7000 to ground. What's the load, a 50-ohm or other resistive pullup, with no coax cable, etc., just a low-capacitance scope probe?

A textbook pulse. One of the issues in fast-driving power MOSFETs is their gate spreading resistance. Depending upon the structure, some of the FET's area is quickly reached by the gate signal, for fast turn on and turn off, but the remaining area is "further away," with an RC delay. This portion won't turn on until a little later in the pulse, an aspect that doesn't have much effect on the turn-on waveform unless the FET is running at very high currents and the reduction of Ron with time is apparent. But at turnoff the "near" portion can be turned off quickly, just as it was turned on, but some of the "far" portion, which sadly is in parallel to the output after all, is RC-delayed and turns off slowly, and the waveform can have a gradual-recovery appearance.

BTW, this is another aspect that spice models fail to handle properly.

Your 2.5ns turnoff is impressive. Does this indicate the 2n7000 parts you were using don't suffer from the gate-spreading resistance problem? Which manufacturer's FET were you using? Also, don't you prefer 2n7002 SMT parts rather than 2n7000 TO-92 through-hole for your fast circuits?

The list of 2n7002 manufacturers and the variants they offer is quite impressive (A, C, D, E, F, K, L, LT1, M, T). And the datasheet output characteristics plots show even more variation than I saw in the 2n7000 families.

What's your experience with these? For example, Fairchild's 2n7002MTF type says it has "Improved Inductive Ruggedness," which sounds useful, although I'm not sure what that means. Better avalanche capability? This transistor's online datasheet features nearly unreadable plots,

but it appears it's capable of switching 2.5A, more than most other 2n7000 or 2n7002 parts. They charge more, 7.5 cents rather than 5.6 cents, qty 1000. :>)

--
 Thanks,
    - Win

Oops, I missed your explanation in a.b.s.e.

"Here's a 50 volt pulse into a 50 ohm load. The output stage is two 2N7002's (the sot-23 version of the 2N7000) in parallel, driving a transmission-line transformer for isolation. I wanted to go 75 volts, but my guys started mumbling something about exceeding the abs max specs or some wimpy nonsense like that.

"Still, that's a lot of slam for 50 cents worth of fets!"

Can you detail the transmission-line transformer connection?

--
 Thanks,
    - Win

It's just a few turns (4? I think) of micro-coax wound on a small toroid. The braid is the primary and the inner is the secondary. Probably a pot core would be more elegant, easier to wind and mount; gotta try that some day.

John

The secondary is *inside* the primary! The core's just for low-frequency response.

John

Higher leakage inductance, though.

Cheers,

Phil Hobbs

ROFLOL.

--
 Thanks,
    - Win

I read in sci.electronics.design that Winfield Hill wrote (in ) about 'John Larkin's fast 50V pulse with a 2n7000.', on Thu, 31 Mar 2005:

Obvious; it's an L of a lot better than the others.

--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk