fascode

Potentially interesting, but it looks like it would require a fair bit of c are once you get out of the small signal regime. IME beads are fairly disap pointing at making slow transistors look like fast ones, e.g. in the collec tor of a current source. IIRC you do that a fair amount, but I've yet to fi nd the secret sauce to make it a win.

Cheers

Phil Hobbs

The problem I have is the disappearence of the NXP fast power PNPs, BFQ149. I can get SOT23s to replace them but they can't handle the power disssipation. The fascode reduces the voltage on the upper PNP.

The cap, which can be giant, is the fast signal path; there is no cascode voltage gain here. The slow PNP just does the long-term stuff. This would be about 40 mA switched on/off, and waveform cosmetics don't matter much.

I have a perfectly beautiful pulse generator output stage, and the NXP PNPs and the Avago phemts are both EOL. Ten of each per board!

I feel your pain, at least on the pHEMT front. OTOH with the lucre from selling your building you could buy several years worth of each. ;)

Cheers

Phil Hobbs

For what HF compliance range?

The "slow PNP" just looks like an oscillator. You'll probably get off okay using s***ty FBs, but woe be unto anyone who accidentally uses "real" inductors. ;-)

What's wrong with stacking inductors? It worked for Picotronics. You've opened one before, you know exactly how they did it!

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

A long time ago when I was a student I was teached that to accelerate transistor switching you have to match drive impedance and real input impedance (dependant on transistor) in order that Zd / Zi is real. I never had the opportunity to test.

That sounds like RF, where things are generally linear and small-signal. We work in time domain, where everything is nonlinear. The concept of impedance matching doesn't usually mean anything to us.

It's switching, not RF. I should trickle a little current into the cascode all the time to keep it alive, I guess. The upper PNP will swing on and off.

All transistors want to oscillate! The upper FB could be a resistor maybe. Maybe even both. I've had good results adding FPs to bipolars, especially base of collector. Haven't tried emitter so far.

You'll probably get off okay

FBs are designed to be cruddy. The cruddier, the better.

Too much work. The conical inductors are great, but too expensive. But they sure make bias tees easy.

I've done some very improbable-looking things along these lines lately, that work. Given that I have no way to accurately simulate this stuff, namely using RF parts in switching circuits, I lay out a

Gosh, my typing and spelling are terrible before the coffee kicks in.

Unlike a cascode, this has the full Miller effect on the low-V transistor. Other than that, it's just a level translator scheme, not really much different than using an output coupling transformer.

The high-V transistor in a cascode doesn't hurt the high frequency response enough to matter; I fixed a monitor's video bandpass issue once by using a better speed of low-V transistor, though.

It works OK with the EOL NXP transistors. The ultimate load is low impedance.

Well, except for being fast and DC coupled and cheap.

Eventually, at some frequency (namely Ft) beta is 1, and the cascode input gets lost to base current.

Termination?

Sometimes, on long runs.