Huh, does anyone put 2N3904 dies in a TO-126 or so?

Zetex puts some small chips into SOT223 packages. Have a look over there maybe. But 5 watts is probably too much for that bit of silicon, I'm guessing.

What do you need, spec-wise?

John

Dissipation isn't just limited by the package; it's limited by the size and construction of the silicon as well.

You may gain some ground putting a 2N3904 die into a better package, but chances are low it'd be that much.

As suggested, Zetex has done some _impressive_ things with little bipolars.

When I was ~10 y.o., I had the similar idea about 2N273, as it was in the nice metal can. Easy to dissipate heat.

Yea... The bonding wires just burned off.

Another ingenious idea could be connecting few dozens of 2N3904 in parallel.

Who cares what Zetex does. I fall into the nostagia. Where can I get

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

I'd hazard a guess that the 2N3904 is a 25mil x 25mil die. The power density thru the die attach media will limit the power capability. ...Jim Thompson

eBay. I've got some Philco 225's and 227's from an uncle's estate (I'm too young for them -- I started doing electronics stuff with bought components circa the 2N3904. My Radio Shack 100-in-1 kit did have some germanium transistor or another, however).

Nat Semi 15x18mils Fairchild (was)11x18mils

RL

I was _guessing_ ;-)

So it could be even worse a thermal disaster. ...Jim Thompson

Jim Thompson a écrit :

Even smaller. According to National's databook the 3906 is a 13x22mils die and the 3904 is 18x18. Both have a tiny 4x4 base region with a 3 fingers emitter.

Just try 5W with that...

Too bad parts like the 2N5641 aren't made any more. It may have done what you needed.

Well, there weer idiots that put 100mA 2N2222 chips into a TO-3 package and made a false datasheet rating it to 2 amps.. ..so why not? Who cares about internal IR losses?

They have, but they don't make any power packages (SMT doesn't count ;-) ). It's nice having a 1A, 30V transistor with hFE > 100 at Ic =3D 1A, but it's kind of annoying only being able to use it as a switch. And if you don't switch it quickly, that e-line won't take it for long. Makes breadboarding awfully fickle. Last time I ordered a few ZTX651/751, for gate drive emitter followers, I burned half and went with 2N4401/03 instead. Half the current, four times as robust...

Tim

el.

Heh. The capacitance would go up a lot, though. Okay, Ic does too, but still, on a PCB there's more stray than if it were done hybrid or monolithic (more emitter fingers?).

What I was thinking about was something like the C2682 (which Mouser currently sells), or 3502 or 3953 (doesn't), or 3595 (which I've pulled from old Trinitrons).

But really, the thought occured to me because I was simulating a circuit, and not having a 2SC2682 model, I thought, meh I'll just put in a 2N3904... the model is bad enough (no 2nd breakdown, avalanche, no falling hFE at high Ic) that it works pretty well just as if someone made a '3904 that did 5W and 100V. And with fT =3D 300MHz, it'll do better than a 2SC3502 (I was getting 10ns edges from the output :-) ).

Tim

How about a 2n5682 instead?

Nah, fT =3D 30MHz min.

Impressive power rating for a TO-39 though. Most top out at 3 or 5W.

Tim

Not sure what datasheet you have but the ones i was using 25 years ago had an f(t) of 150 MHz or better. I was getting about 10 ns 10 volt edges.

ST Micro says 30 min.

It does say "minimum".

Tim