Maker Pro
Maker Pro

Emitter Base Breakdown

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Sorry, I read that as "I didn't test at all for beta..."

I am awaiting independent verification. Or should I just take your word for it?

All of the explanations I can figure out (again, can you show me any others documented anywhere) would suggest that any damage is likely to be related to current, yet you say it isn't.

I have spent quite a bit of time on several occasions looking for something published that either demonstrates the effect or explains the cause.

Someone else is doing some experiments and I'm setting up a high voltage (200V) constant current source (1uA, 10uA, 100uA 1mA) so I can do my own tests. (And after I've read the next couple of posts I'm going out to build it...)

You suggest that the effect is very easy to measure and I've got bazillions of transistors I can try it on.)
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Here are some results I have measured:

Transistor MPS2222A
Test current 55uA

Initial measurements:

HFE 127 @ Ic=2.50 mA

After a few seconds of reverse current (Veb= 6.87V)

HFE 126 @IC=2.50mA

After about a minute of reverse current (Veb= 6.88V)

HFE 127 @IC=2.50mA

After a few minutes of reverse current (Veb= 6.90V)

HFE 123 @IC=2.50mA

There is some evidence of reducing HFE and increasing reverse voltage. However I will wait and see if this can be attrivuted to thermal issues.

Here is an interesting wikipedia quote to read while I wait...

The emitter-base zener diodes can handle only smaller currents as the energy is dissipated in the base depletion region which is very small. Higher amount of dissipated energy (higher current for longer time, or a short very high current spike) will cause thermal damage to the junction and/or its contacts. Partial damage of the junction can shift its Zener voltage. Total destruction of the Zener junction by overheating it and causing migration of metallization across the junction ("spiking") can be used intentionally as a Zener zap antifuse
OK, if we assume this is an avalanche breakdown, it is clear that heating wil increase the voltage:

Above 5.6 volts, the avalanche effect becomes predominant and exhibits a positive temperature coefficient.
OK, after a few minutes wait...

HFE 120 @Ic=2.50mA

HFE increases at about 0.5%/Cdeg, so heating is going to ncrease HFE, and that explains the decrease in HFE with time after this test.

Cearly, at 55uA on this single sample, an effect on HFE can be seen that appears to be related to the duration of the reverse current. (So the effect on HFE appears not to be instant).

Interestingly, the avalanche voltage does NOT seem to reduce as the junction cools. It is still at 6.90 volts as I start the next test (which wil be for a much longer duration).
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Now after 35 minutes or so, Veb is 6.93V

HFE = 113 @Ic=2.50mA

After a few minutes HFE = 110

I'm going to leave it on overnight now...
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
OK, over 10 hours later...

Veb = 7.02V

HFE=73 @ Ic=2.5mA (falling to 72 after a few minures)

Next thing woud be to try one at a lower current...

New transistor

HFE=123 @Ic=2.50mA

After a few seconds of Veb of 6.85V @ 2uA

HFE=123 @ Ic=2.50mA

After about 30 seconds of Veb of 6.84 V @ 2uA (seems to be falling!)

HFE=123 @ Ic=2.50mA

After a few minutes of Veb=6.845 V @ 2uA (not falling so much....)

HFE=122 @ Ic=2.50mA

After about 15 minutes at Veb=6.85V @ 2uA (seems steady now)

Rose to 6.855V after 15 minutes.

HFE = 121 @ Ic = 2.50mA

OK, I'll leave it on for a long time now...
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Page 30 of this document gives a good explanation.
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
Great stuff. So now we know how to get a lower Hfe transistor if we need one :)

Bob
 

john monks

Mar 9, 2012
685
Joined
Mar 9, 2012
Messages
685
I finished my test. A 2N2369A transistor started out with a beta of 30.3 last Friday.
I ran about 7.35 milliamps through the base emitter junction in the zener mode until Monday morning.
The beta went to 23.2 So I am surprised.
Being skeptical I tried another 2N2369A transistor.
The beta started out at 44.4 After very briefly pulsing the base emitter junction at about 55 milliamps in the zener direction for less than a second the beta went down to 41.3.
Letting the transistor sit for a while the beta increased slightly.
So my conclusion is that EinarA is correct, that the beta is decreased by running the transistor in the reverse direction.
Now what I would like to know is if running a transistor in the avalanche mode does any damage.
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
At 2uA, after 20 hours or so the Veb rose to 6.92 volts and the HFE fell to 102
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Read the link I gave above. My suspicion is that the effect would be (at the very least) much reduced.

The BC junction is less highly doped and much larger, two factors which apparently lead to the effect at the emitter junction.

edit: does anyone have access to this?

edit2: This is interesting.

(Knowing the correct term to google for -- hot carrier base emitter -- leads you to all sorts of interesting places).
 
Last edited:

john monks

Mar 9, 2012
685
Joined
Mar 9, 2012
Messages
685
Thank you steve. I read the articles and now I am wondering how I've been able to ship hundreds of pulse generators going back to the early 90's with avalanche transistors that are still being used and still coming in for their annual calibration without failure. And the old 1970,s Tektronics sampling oscilloscope used transistors in the avalanche mode, I suspect are 2N2369A's and gets regular use by me and has never failed. I suspect that the transistors degrade to a point and then begin to stabilize but the charts you linked to tend to paint a different picture. I have noticed that the 2N2369A transistor, in the avalanche mode, tend to have a very sharp and well defined avalanche characteristic when starting out and then then to avalanche about half-way on minutes later when the avalanche is a rather high current.
The ZTX415 is designed to be an avalanche transistor and I wonder if it also suffers from the degrading effects like the other transistors. What is different about them?
Thanks EinarA
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
In one of the links I gave, failure was determined to be a 10% decrease in HFE.

I will also note that one of them also suggested that the 2222 I chose is likely to be more affected than other transistors.

From that it it logical to conclude that other transistors are less affected, and that it is possible to operate an appropriately designed transistor in this mode for an extended period. Is the ZTX415 such a transistor?

The answer is that yes, the ZTX415 is clearly designed to be operated with CB avalanche, but I'm pretty sure, NOT with EB avalanche.

It would be interesting to repeat my tests with the 2222 in CB avalanche mode.
 

john monks

Mar 9, 2012
685
Joined
Mar 9, 2012
Messages
685
The ZTX415 is designed to operate for an extended period of time in the avalanche mode.
I typically run them at 8 amperes for 25ns at 2khz for weeks. They almost never fail.
I have never tested one for beta but I certainly can and I will.
About 21 years ago I was using an 2N4237 in the avalanche mode and they are still out in the field working reliably.
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
I have tested a 2N2222 for effects on HFE of reverse biasing the CB junction at 3uA and 55uA.

Surprisingly there is some effect on HFE even at 3uA (but not a lot).

I am seeing more at 55uA (no surprise perhaps), but these are pretty low currents.

The 2N2222's I have break down at about 140V if anyone's interested. (c.f. datasheet Vcbo of 75V)

edit: at 55uA the dissipation is under 8mW, so heating should not be an issue.
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Latest results, 2N2222 tested @ 3uA and 55uA CB avalanche.

3uA (at around 145V)

Initially HFE 127
+10s hfe=127
+1m hfe=126
+3m hfe=126
+25m hfe=127
+1:20 hfe=128
+5:00 hfe=125

(All times are with respect to the previous test, so total was around 6:49)

All changes up to the last reading are within the error of the testing device. HFE of 125 does suggest a reduction in HFE.

The same transistor was then tested at 55uA avalanche current. (Interestingly it initially measured 124, which indicates that +/- 1 in the reading from the test is not significant.

initial hfe=124
+10s hfe=125
+1m hfe=125
+3m hfe=123
+35m hfe-119
+1:20 hfe=117
+8:45 hfe=98

(And this transistor is still on test while I'm at work, so expect more results in a little over 8 hours).

It seems that there is evidence that avalanche in any junction of a 2N2222 causes a reduction in HFE. It would be interesting to see if people have similar results with other transistors, especially those designed for avalanche operation.

It would also be interesting to see what happens if a jfet is subject to this, and if so (I can't see they would be immune to hot carrier injection) which parameters will change. And what about mosfets...? (ds avalanche)

+
 

john monks

Mar 9, 2012
685
Joined
Mar 9, 2012
Messages
685
I'm not sure about jets but Mosfets blow out right away. And I think this is because the gate capacitance is typically around 0.001uF. And the voltage required to punch through the metal oxide insulator is around forty volts or so. And I thing this stores enough energy that when an internal arc takes place permanent damage is done to the insulator shorting it out. What I no for sure is that whenever I punch through barrier potential the MOSFET shorted out,
However I did run 3N120 up to about 20 amps for about 10 milliamps source to drin without damage. This is several times the rated specifications.
But because you mentioned it I will try jfets.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
I was thinking about drain/source avalanche for the mosfet...

But anyway, the last reading for a while on the 2N2222.

After an additional 12 hours and 24 minutes at 55uA, the HFE fell to 89 (from 98)
 

john monks

Mar 9, 2012
685
Joined
Mar 9, 2012
Messages
685
That is interesting that damage continues to occur.
I tried the ZTX415 in the avalanche mode.
The beta started out a 55.4
After the test the beta was unchanged at 55.4
So something is peculiar about the ZTX415.

I noticed that Fairchild did a white paper on mosfets in the avalanche mode.
Apparently they are switching an inductive load with a 800 volt mosfet and watching for the characteristics of the avalanche condition. From their chart it appears the avalanche condition occurs a little over 1000 volts. I think I'll try this but this is going to be more difficult because the gate must be brought from the on state to the off state within a matter of a few nanoseconds. I'm afraid to predict the results. I was off on the bjt's.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
It would indeed be interesting to know what is actually different about that transistor.

I'm going to give a mosfet a try later on today.

Something with a 30V Vds should avalanche below the limit of my 200V supply...
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
OK, so I repeated the same thing with a mosfet.

The one I chose happened to be an IRF7220 P channel mosfet.

Rather than leaving the gate open (and risk damage, I tied it to the source with a 270k resistor.

Voltage was applied across the drain and source.

Interestingly, for a 14V device, it started to conduct at just a fraction over 14V (compared with over 140V for the 40V 2N2222)

The simplest measurement I could take was Vgs, and for the unused device, it was 0.78V for 2.5mA Ids

After about 1 second of avalanche at 55 mA, it rose to 0.82V -- Cool! I'm going to see something.

And then it remained the same after a total of 9 hours.

The initial change (I feel) was too much to be measurement error, especially since all subsequent readings were spot on 0.82V

OK, sorry for the delay... I just went and checked that I hadn't accidentally been passing 55mA through the body diode :D I hadn't :)
 
Top