High frequency counterparts of 2N2222A, 2N2907A

Not close as far as power, voltage and current are concerned, but when the ratings were adequate I used the 2N2369A, not just for its high frequency performance, but for switching speed as well.

Graham H

1. Forget 2222/2907 ever existed. They're old and underspecified.
2. If you need moderate current, use 2N4401/3. Same exact thing, but actually well specified.
3. If you don't need much current, you might as well use 2N3904/6 (40V

1. Finally, for higher frequencies, you can look at MPSH10/81 (25V 50mA >650MHz)
2. For really high frequencies, BFR92AW / BFT92A (20V 25mA >3.5GHz)

If you need more current but the same fT, low-Vce(sat) transistors are quite good. Example, PBSS303(N/P)X.

If you need comparable current (~600mA), you need RF power transistors like 2N3866/2N5160 (obsolete, $$$ if you find them), or something starting with MRF or etc.

Tim

Not the same thing. 2N2369 is a _gold-doped_ device which improves switching speed recovering from saturation... it is _not_ a very good RF device.

Just go looking for a high fT device. ...Jim Thompson

High Ft parts tend to be physically small, and their current and voltage capabilities are lower than general-purpose transistors.

What voltage, current, and Ft might you need?

I like BFS17 as a pretty-fast-but-not-too-fast NPN. 15 volts.

BFG25 is faster, but only 8 volts.

Somewhat interestingly, I just checked the spice model for this, which I got sometime ago, and its TF=5.11e-12, which is a tad small.

TF should be around the 1/2.pi.Ft, so for for a 5Ghz device, around 32e-12.

Kevin Aylward

These RF parts, BFT25 and BFG25 sorts of things, make fabulous diodes. I measured the C-B junction of a BFT25 as having reverse leakage around 10 fA, which was groveling around my measurement limits. That's way better than a PAD-1, which is way more expensive and a rotten diode in general.

Sadly gone, though. RIP.

Cheers

Phil Hobbs

I usually use BFT25's, and only used the BFG25 once. The customer has reverse-engineered my design and is making them themselves, so I'm sorta glad the BFG is discontinued.

It's all your fault then. ;)

Cheers

Phil Hobbs

There's nothing like integrating a programmable uC in the design and locking the bits.

Jamie

locking the bits. "

Who's to say they can't get around that as well ?

You can compute a checksum that includes the hardcoded processor ID (if any). They'd have to have a clone processor to get round that one.

Cheers

Phil Hobbs

...or subvert the check.

Pretty tough to do, if the checking code is included in a cryptographic checksum and is properly implemented. You certainly don't just want to check once, in one function call returning a bool. You can also encrypt the checking function.

Cheers

Phil Hobbs

If you can write to the code space, it's easy to subvert any checking.

Depends on the circumstance. I agree, physical control of the system plus sufficient time, skill, and determination equals successful hacking. The thing is to make it harder to hack than to reproduce, and there are a bunch of ways to do that.

Cheers

Phil Hobbs

Well, that's the bottom line but unless your checksum is something done by the hardware, it's useless. If you have physical control of the system, none of this is needed. The hardest problem is often to define what you're really trying to protect against.

I don't think the situation is quite that bad. For instance, Segger sells a library for this purpose, which uses symmetrical public-key cryptography. If the baddie doesn't have the private key, it's quite a bit harder.

Cheers

Phil Hobbs

All the cryptography in the world is useless if I can bypass the test. This cryptography only works for a small subset of problems and as you pointed out earlier, it requires physical control of the hardware. Again, define exactly what it is that you're protecting against before you sell the solution.