2N2369 reverse Vbe avalanching, consequences?

LSTTL did that with a Schottky imposed right over the B-C junction of the bottom transistor in the "totem pole". ...Jim Thompson

And I doubt the 1N4148 claim... never heard of that... and I'm been around this biz for a large fraction of a century ;-)

My MC4024 voltage-controlled multivibrator was built on a gold-doped TTL process, to accommodate a TTL-compatible output stage. I had to do various beta compensation tricks to keep the analog portions well-behaved.

Later we were able to do selective-area gold-doping to make mixed-signal chips, but CMOS and, particularly, BiCMOS processes pretty well eliminated the need for that game.

BTW: Gold-doping kills lateral PNP's. ...Jim Thompson

Plenty of google hits and I think Pease wrote about it too. Wasn't that the only way sub 5ns trr could be achieved back in the day?

I've seen people complain that some makes of 1N914s and family are and some aren't.

piglet

There seems to be quite a variation of performance in different brands of the "small signal" diodes... but I've never heard of them being gold-doped... seems like an invitation to large reverse leakage. ...Jim Thompson

Yup. Hence the combination NPN/Schottky diode symbol in the datasheets. I was teasing Joerg about using gold doping, as though it were 1968. (I understand about not wanting to do a board turn, of course.)

Cheers

Phil Hobbs

Well, they're half-dead to begin with. ;)

Cheers

Phil Hobbs

I won a coffee bet with Tom Frederiksen that I could make a reasonably accurate current source as long as Beta_LPNP > 1 (as in the MC1494/1594 ;-) ...Jim Thompson

I still occasionally use some variation of the Baker Clamp concept when I have to use bipolar output devices.

Most of the more economical I/C processes only have Aluminum Schottky's, so are not robust enough for high current outputs. ...Jim Thompson

Yes but I wanted to not have to roach on stuff but just switch a transistor. The 3904 with 10k in the collector gets us by for now.

Do you think you could put a BAT54L (SOD882 package 1.0x0.6x0.5 mm) onto the base and emitter pads "between" the pins of the transistor?

According to the MMBT2369 datasheet, the SOT23 package has 1.5 mm between the pins (on the "inside", edge-to-edge). A reasonably sized pad should be big enough to extend at least 0.7 mm outwards from the SOT23 package. If your pads are also large enough in the direction parallel to the package (that is, the shortest distance between the B and E pads on the PCB is not larger than 0.35 mm) then that would make it possible to place a SOD882 directly next to the SOT23, onto the existing PCB pads, between the pins, without touching either the pins or the package of the SOT23 (in a sort-of "normal" machine-placeable way).

Regards Dimitrij

I could do that but the machine placement guys won't accept such rework. But it's ok, the 3904 now gets us through. Not nice but works.

I've seen tons of discussion about HFE degradation from zenering b-e, but I've never seen numbers.

There have been studies but many were done when we were still kids and those haven't made it into cyberspace. I remember one from RCA Labs but can't find it. Here is a more recent one from France:

OK, a few coulombs of base zenering current dopped the beta in half.

But they pushed 60 mA into about -8 volts Vbe, so they basically cooked the transistor.

Yeah, no mention of what TJ was during the test. It's not obvious if there's a rate dependency either. Interesting to see the low current hFE degredation, which suggests it's a defect recombination effect, which suggests that... well hey, 2N2369 should be pretty shit to begin with, and not really get much worse, right? ;-)

('Course, in the present case, it looks like capacitance and bias current are the limitations, so killing storage with gold doping or electron bombardment will make little to no difference.)

Tim

That's getting on for half a watt. Quite a bit for a 2N2222A, but

At more moderate currents such as 10mA, I've not observed much change for modest exposure like 15 minutes. Almost less than I can measure without controlling the temperature of the transistors.

Eg. C8050 NPN transistor TO-92, using DMM hFE function The C8050 is a popular general purpose part similar to this one:

Before 15 minutes @ 10mA Bf = 157, Br = 31 after Bf = 152, Br = 31

Just handling the transistor increased initial Bf from 157 to 161 so it's a pretty subtle change- a bit over -3% in this case.

Tried another sample at 45mA (about the maximum in a TO-92) for 30min

Before 30 minutes @45mA Bf = 127, Br = 21 After Bf = 95, Br = 21

That's about a 25% decrease in hFE.

--sp

That's about 80 coulombs of reverse current.

Now we have some numbers!

nd

Thanks for the data. (Vrev. was the typical 7-8V?)

At long times (low power levels) I'd sorta expect heating to be heating, whether forward voltage of Vce or reverse of Vbe.

George H.

Yes, in that range.

I sure would not expect to see any (permanent) change in hFE at all from dissipating a few hundred mW in the collector for half an hour or half a year.

--sp