The beta petering out at 50% or so would suffice in my case. It just feels iffy because nobody really guarantees this.
Yes, it does increase noise but it wouldn't matter in this case. Fixing it is easy. A few volt negative is needed so a zener plus a diode would do but it's a design change of a running product with all the hassle it entails.
-- Regards, Joerg http://www.analogconsultants.com/
What do you mean by useless? Beta dropped to a fraction of its former self?
The gain seems to drop of
Did things plateau after that week or did they degrade farther, just mosre slowly?
How much beta drop did you see?
It's easy to fix but not so easy on a running design.
-- Regards, Joerg http://www.analogconsultants.com/
Thanks, that is a very interesting device. I wonder how they managed to obtain such high hfe for a non-Darlington. I stored the datasheet in case there is a case where I could use it.
In this case I have to do the zener-diode combo if the client wants to because the circuit architecture relies on the breakdown.
-- Regards, Joerg http://www.analogconsultants.com/
Sure looks like it from the 2N2222 data. AFAICT there's no reason that hot carrier damage would be self-limiting. More recombination sites, lower beta, lather rinse repeat.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
I use them a lot as low-current cap multipliers. With an extra RC in the collector, the large interelectrode capacitance doesn't matter, and with the high beta, I can get long time constants without huge capacitors.
Yikes.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
+1 if they'd cut open the package they'd have seen the light being emitted!
60mA continously applied is nothing like the intermittant 1mA Joerg is worried about.
piglet
Yes. The 2N2222A, 2N3904 and 2N4401 reverse breakdown are all 6.0 volts.
1 mA is hardly in the avalanche breakdown mode. It's only 6 mW, so the base junction might well be able to dissipate the heat without damage. The only way to tell is to measure it.The zener diode could be a reasonable precaution if normal circuit variations could produce much higher currents. However the base-emitter breakdown could be much sharper than the zener, so the type of zener and the zener voltage needs to be carefully selected to ensure the zener is actually protecting the base junction.
Here is some more info on reverse junction breakdown:
AN1628/D Understanding Power Transistors Breakdown Parameters
Emitter Base Breakdown Voltage of NPN Used as Zener
gain dropped off to about 10% or so of the original gain (IIRC - in any event it was quite dramatic and took the gain way below the minimum specs for any version of the transistor). I only *suspect* gain would continue to drop off but very slowly. However, what I saw looked like the gain was stable, but the short term (days to a week or two) of my testing may not have been able to spot a continuing slow degradation after the initial 'settling' period. I wasn't actually looking for gain over time. I was looking for reverse breakdown noise (which also degrades btw). The gain thing was just a curiosity.
-- Using Opera's mail client: http://www.opera.com/mail/
Thanks for the info. A 90% drop in gain is pretty much a destructive process. So that would mean just a few hundred uA can gradually destroy a BJT.
-- Regards, Joerg http://www.analogconsultants.com/
That is also what David's experience shows and his was in the sub-mA range. IOW a mode of operation where -Vbe max is exceeded is not kosher.
-- Regards, Joerg http://www.analogconsultants.com/
They did it at a variety of levels, and interestingly the lower levels were worse. I expect that was because the die got hot enough to self-anneal at 60 mA.
CHeers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
Are you talking about the paper Joerg linked to? I didn't see any different currents. Can you point to a page or graph?
Thanks George H.
Never mind, I was misreading Fig. 2. Thanks for spotting that.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
What pdf are you looking at? In the 053209.pdf discussed above, they applied a constant -8V to the base emitter junction, resulting in 60 mA current. They varied the length of time the voltage was applied. Figure 2, "Current gain variations versus VBE for different stress times" shows the degredation in beta vs application time.
The minimum time resulted in a loss of beta. Longer times simply increased the reduction in beta. For example, the original beta at 0.4V forward bias is approx 68. As soon as they applied reverse current, the beta dropped close to zero at 0.4V.
There are some problems with this test. Did they use the same 2N2222A throughout, or did they get fresh 2N2222A's for each run. If they used the same transistor, then each run accumulated the degredation from previous runs. If they used different transistors, how did they get ones with identical beta's?
And how did they get 60 mA current each time they applied -8V reverse bias?
The paper is not helpful. The data is useless.
Or are you looking at a different paper?
That brings back memories, the zener diode oscillation was a hefty discussion back then
Bob Wilson had a lot of good posts. Whatever happened to him?
Cheers
Klaus
So a voltage just below the zener threshold will do about as much damage as zener current?
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
I remember Bob. He was very good. He just suddenly disappeared one day and never posted again. I don't recall anyone ever mentioning what happened to him.
You are right - the paper is strange. Measuring beta at a different Vbe values is unusual. In practice one is mostly concerned with operation around the familiar 0.6-0.7V room temp Vbe. Measuring Vbe at 0.4V implies very low current operation.
The whole premise of subjecting a BJT to extended continous reverse breakdown and then testing forward behaviour is not representative of the real world. In the kind of situation where Veb breakdown occurs it is usually of limited short duration, typically as a result of capacitive coupling. The BJT spends (far) more time forward biased than reverse biased.
CRT line driver transistors were sometimes coupled by electrolytic capacitors to their drive transformer. The drive levels could be enough to zener the b-e junction but the majority of reliability problems were associated with the reverse bias on the electrolytic capacitor not the transistor.
piglet
You are right. The curves for the 2N2222A, 2N3904, 2N4401 series show the gain vs collector current. They start at 100uA and go up. The graphs include curves for different temperatures, such as -40C, 25C, and 125C. This is much more useful than comparing against base-emitter voltage.
One big problem is oscillator design, such as multivibrators, and LC and crystal oscillators. The amplitude may be so high that the B-E junction is broken down on each cycle. LTspice will not show this unless you add a variable, Bvbe=XX to the model where XX is the breakdown voltage in volts.
Crystal oscillators are a special case. The Q of the crystal may cause the oscillator to take many milliseconds to start. This makes it virtually impossible to view the oscillator waveforms to see how close the amplitude is to breaking down the B-E junction.
I have devised a simple method to start the oscillator instantly at the full amplitude so you can see the oscillator waveforms immediately instead of having to wait. Unfortunately, both versions of LTspice may require changing the LTspice.ini file to work with this procedure.
It means replacing the existing .INI file with a copy of my .INI file. It appears to be identical to the original file, but there is some difference that does not show up in the most careful examination to show why my .INI file works and the original does not.
For this reason, I am reluctant to post my procedure since I am certain it will result in numerous questions that will take too much of my time to answer. Sorry.
Getting an oscillator to start in SPICE is easy. You put in a voltage step of a millivolt or so. That's small enough that the transistor stays in the small-signal regime, so you don't wind up fooling yourself that it'll start when it won't in real life.
Good tip about BVBE.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.