Adapting this NSC app note for my +150V regulator

Virtually all modern transistors will self-oscillate if applied incorrectly. I'd play my money on bad design.

...Jim Thompson

Well we never did before and after that batch! :)

"Jigs" often exacerbate the problem.

...Jim Thompson

Well, think what you want, the units were tested in 2 different jig circuits and failed at aprox the same current in common emitter test and emitter follower test. Playing with different biasing and voltages didn't seem to matter. Only the Ice seem to be a factor. If we attempted to suppress the noise it would place an attenuated dip in that region of linearity of the circuit. Basically, it was almost acting like a tunnel effect at that point.

They do seem to work well for switching how ever. We send back a few to the supplier for them to look at before they were going to replace them. We had made a large order of these which were fairchild made. They were suppose to send us a different vender of the product but we ended up with Fairchild again with a different date code. These were fine.

I kind of think that many look at the 2222 as a switching transistor and not much used in linear applications with this family. Maybe the fab house also thought this and didn't care. Just correct the problem and call it a day.

I agree with Jim. I have used the MMBT2222 in a lot of analog circuits and never had any oscillate, if the rest of the circuit was right.

As far as your jigs: Do you have the specifications for them? Lead inductance? Inter electrode capacitance? Did you compare the gain of that batch against the ones that work, as well as the fT? Sometimes a bad circuit works great with a part that barely meets specs. Replace it with a better quality component and everything goes to hell.

An obvious candidate would be like so. Nothing shows up in simulation but there's something about it that disrupts LM317 stability, too much gain at current limit: View in a fixed-width font such as Courier.

. . . V+ . | 1N4004 . .----+-------| LM317 | . | | | --------- | . | | | | | | . '-----+-[7.5]--+---|Vin Vout|---+- - - . | | | . | Adj | [150] . --------- | . | | . '--------+ . | . . .

There's no problem with this wasteful thing though: View in a fixed-width font such as Courier.

. . . V+ . | 1N4004 . .----+-------|

To analyze the case better it would have been good to check the suspicious lot for hfe and ft. Chances are that one or both may have been better than in the datasheet. Maybe too much for the jig, but I don't know the jig.

Sounds like a bad die, with some kind of current dependent Vce breakdown, not worth investigating further, you should dispose of them. Oh wait! Sell them on ebay...

Now that's a great idea! :)