need fast switching transistor

How much current?

The DEI fets come close, if you can drive the gate hard.

The Nitronex GaN fets are really fast and relatively easy to drive.

Avalanche transistors maybe?

Relay contacts can do that, too, in the closing direction.

What's the application?

John

Lots of current? If so you could take a look at these:

Not exactly cheap, and it'll be pushing it. Digikey has them but the 2nd one carries a lalaland notice.

Check out the NTE64. Silicon NPN high-speed switching transistor. UHF. For some reason I can't get to the datasheet at the moment, but it might be worth looking at.

Dave

Nevermind. I finally got the datasheet to open, and it only handles 15V collecter/emitter, or 25V collector/base, and 3V emitter/base. Sorry. I had forgotten those parameters.

best...

Dave

Well, it works out to be 5mA

The application is ultrawide band pulse generator for UWB radar.

I looked at the data sheets, thanks, but I could not see any figures relating to the switching speed; on to off times. I am going to drive the device hard on and off with a digital square wave.

You won't find that kind of data for an RF transistor, got to buy one and try it. Often they are remarkable switches, just not marketed as such.

Same in the other direction. On my very first project in the corporate world I used a BSS123 as an RF amp. Da boss was a bit uneasy about that because there were only Ton, Toff and such, no decibels or things like that. We've cranked over 30,000 units per year and it worked great.

That could make life a lot easier. Can't you use a regular lower voltage device and follow that with a step-up transformer? 10:1 BW should be easy and sufficient for UWB.

That's a problem, maybe. If only 5 mA is available to switch, there's not enough current to charge circuit nodes when the transistor turns off. Essentially, high impedance nodes can't exist at nanosecond speeds.

Plugging into C = I * T / V

and using 80 volts, 5 mA, 1 ns, C = 0.06 pF.

Is the load an antenna? If you want a fast 80v impulse into an antenna, a step-recovery diode is the easy way to go. Well, relatively easy: nothing is really easy below 1 ns.

Both of these gidgets use SRDs to make their spikes:

The T250 uses, I think, three SRDs in series, which shouldn't work but does.

A Zetex avalanche transistor will give you a couple of hundred volts in about 1 ns, and is fairly easy to use if you can tolerate KHz-range rep-rates.

John

Hmm, the initial design value was 68 pF. The charging path is through a 16.5K resistor. When the transistor closes, the charge is dumped into 50ohm load; the antenna

I have heard of this before. I will look into these components. The original design is not mine, I have been handed the project and the original engineer is not available for comment; figures right! So I am trying to improve the design and looking into other options.

This is very interesting stuff.

The BLF6G... series is in full production.

robert

However, only 64 pieces of the BLF6G22 in stock at Digikey, the rest is all non-stock. Ok, they are seldomly needed but non-stock doesn't instill much confidence here.

Yeah, NXP is really bad at stocking "generic" distributors in small qtys. You can either buy 3000-piece reels or nothing whatsoever. The BLF series ist mostly sold directly to cell phone base stattion manufacturers I guess. Should still be a fun part if you can get hold of one.

robert

I wonder when companies will learn that a concentration on key accounts can lead to trouble quickly. European companies seem especially exposed here yet that's exactly the region where Senor Lopez fought his battles. That should have been a good lesson but I guess it wasn't enough.

Old rule out here in the US: If Digikey doesn't have much stock chances that it gets designed in are slim, close to zero.

Who is Senor Lopez?

robert

It starts to become interesting on page 1445.