Ge Transistor leakage question

[boogietone]

I have recently begun to test a batch of PNP Ge transistors using R.G.'s method (see fuzzface discussion on geofex). I believe I've got the idea behind both transistor leakage and the tester. Transistor leakage is the leakage current across the reverse biased collector-base junction. This current then crosses the base-emitter, which, of course, causes a current from collector to emitter equal to the leakage current times the Hfe of the transistor, just as any current from base to emitter does. As I understand it, the leakage current is a positive function of temperature, i.e., an increase in temperature increases the leakage current.

However, the testing I did last night was a bit odd. When first hooked up to measure leakage with a ~2400 ohm resistor from collector to the negative of a 1-spot, emitter at +8.81 volts, and base open I measured voltages ranging from 0.7 to 2 volts for various transistors. What I thought was strange was that the measured voltage (and thus leakage current) over the course of minutes went down by 25 to 50%. The voltage did stabilized at some unknown time later. I would have expected the leakage current to increase due to resistive heating in the transistor and in fact putting my finger on the transistor did immediately cause an increase. Removing my finger, the voltage went back down.

Is this normal? Is there some capacitive charging taking place? In essence, what gives? I assume that I should wait for the stabilization of the voltage/current before noting the leakage measurement.

Note that the room temperature is fairly stable, the transistors had had plenty of time to be in thermal equilibrium with the room, and adding the 2.2M resistor at the base did increase measured voltage as expected.

Thanks.

[jrod]

In my experience...Yup, that's normal. The measurements will go up when the transistor is heating (from touching, room temp, breathing near the transistor  ) and will go down as the transistor cools. You will need to let the transistors settle to get the best measurements. You can use tweezers when placing the transistors in the socket to keep the transistor from heating up.

You can see how sensitive germanium transistors are to temp by taking measurements this way. Pretty crazy how the measurements fluctuate!

[DavenPaget]

Transistors were used as a "diode"-esque temperature sensor .
I have them in a kit i recieved to build in 20mins . I did it in 10mins .
Brag aside ,
germanium transistors make the fastest temperature sensors , and if you notice temperature sensors using diode style connected transistors make use of leakage .
But yeah ... there's 2 diode junctions in a transistor . But really , in a NPN you get base and emitter as a single diode . base and collector being another diode . But honestly unless it had two bases it won't work .

[Earthscum]

I was measuring Germs last night and was waiting for the first to stabilize... Forever. The gain kept going up and up... I finally realized the incandescent desk lamp was shining directly on the transistors black case.

[boogietone]

I did some more testing this afternoon and will continue tonight. My surprise was not that temperature changes the current or that leakage current went up when I purposely heated the transistor but that the current went down after I hooked up the power and had current flowing through the transistor to test the leakage. I was expecting it to go up due to resistive heating. The current fell at a relatively slow monotonic pace for well past 20mins and maybe even an hour.

I am now hypothesizing that enough heat was absorbed while handling the transistor before adding power that the temperature of the transistor was still on the way back down to room temp + resistive heating for that length of time. It is also possible that the room temperature changed over that time. It would not have changed much more than a degree or two , however.

Does that make sense? Is there another explanation?

[.Mike]

I don't have an explanation, but...

When I tested a big batch of germaniums, the gain seemed to change based on which direction I faced when exhaling. Ok, maybe not that sensitive, but still, it was going to take forever to let every single transistor assume room temperature.

I pre-labeled all of them, lined them all up on a breadboard-- you can fit a bunch-- and then went to bed. They had certainly settled by morning. Then, instead of moving each transistor to the tester, I moved the tester to the transistor using test leads. No need to touch them until after they were measured. Very fast, so your readings won't go wonky when the furnace kicks on.

Mike

[Earthscum]

I pre-labeled all of them, lined them all up on a breadboard-- you can fit a bunch-- and then went to bed. They had certainly settled by morning. Then, instead of moving each transistor to the tester, I moved the tester to the transistor using test leads. No need to touch them until after they were measured. Very fast, so your readings won't go wonky when the furnace kicks on.

Mike

Good idea!

[jrod]

I am now hypothesizing that enough heat was absorbed while handling the transistor before adding power that the temperature of the transistor was still on the way back down to room temp + resistive heating for that length of time. It is also possible that the room temperature changed over that time. It would not have changed much more than a degree or two , however.

Does that make sense? Is there another explanation?

Yep, when you touch the transistor it heats up enough to change the gain and leakage parameters. I don't know how much heat current produces in this circuit, but I think what you are seeing is the effects of body heat from touching the transistor and room temperature. I know RG Keen has suggested using tweezers or gloves to handle the transistor because the heat from your fingers is enough to give you false readings. Plus once they heat up it can take forever for the reading on the DMM to finally settle down.

[electrosonic]

I put ten mediocre AC176 transistors on a breadboard and measured leakage and gain at three temperatures (following the method on the small bear site)
1) At my workbench in the basement 19 C (66 F), upstairs in the dining room 22 c (71 F) and in the kitchen with the door closed and the heat turned up 25 (77 F).
I left the transistors for 15 minutes to stabilize their temperature before measuring them, and since they were breadboarded I did not touch them while measuring them.

I put the results in a spread sheet...



From 19 C to 22 C, the leakage increases an average of 60%, and then it more than doubles when increased 3 degrees C more.
Gain increased by about %20 with each 3 degree C increment.

I read about using a germanium diode to offset the change leakage - the old geocities site "Plate to Plate" has been resurrected somewhat here...

http://www.reocities.com/sunsetstrip/studio/2987/britface.html

I will have to try that out next.

Andrew

[Pyr0]

Interesting, now I just need a minature heater, fit it in to my fuzz face enclosure and I can use some low gain Ge's and still get a good sound. 

[seedlings]

Interesting, now I just need a minature heater, fit it in to my fuzz face enclosure and I can use some low gain Ge's and still get a good sound. 

http://www.heatersplus.com/nichrome.html

1.5 amps of 16 gauge NiChrome 60 should give you 100F in the enclosure!  Just need to cram 23 feet of wire inside.  Alternately, use a 6.3V heater lamp.


CHAD

[brett]

Hi
when it was first turned on, the first power transistor in a transistor amp heated up, it's gain went up, so the temperature went up more, and the gain went up more, .... until it blew up.
So some bright lad or lady added an extra transistor - also fitted to the heatsink - so that as the temperature goes up, it diverts base voltage *away* from the power transistor, and avoids "thermal runaway".
cheers

[PRR]

> At my workbench in the basement 19 C

You are apparently measuring Ice, which is Ico times hFE.

And your hFE measurement is probably not at the same current as the leakage. And hFe does change with current.

However do the math pretending the hFE values are ball-park.

Tc_leak_hfe_Ico
19 0.31 69  0.0045
22 0.48 80  0.006
25 1.00 100 0.01

Ico doubles in 6 degrees C.

The "right" answer is double in 10 deg C.



I suspect you are reading leakage so far down the slope of hFE/Ie that you have the two effects factoring together. Hot increases both Ico and hFE. Ice increases faster than double/10C.

Can you read 5 micro-Amps? Some DMMs can. Wire 1.5V battery, 100 ohm protection resistor, current meter, and transistor C-B junction in series. If you get 12milli-Amps, turn the transistor around. Values below 20micro-Amps are expected.

Do you have ice? An hour on ice under a box gives you a far-out data point.

[electrosonic]

OK this is making my head hurt.

I am connecting 9 volts from collector to emitter, with an ammeter in series. I measure the current and call it leakage.  You are saying that this is the  collector to base leakage  multiplied by hfe. Ok that makes sense.

Using a 1 M resistor I inject a small current into the base (less than 10 uA) and measure the  resulting current  emitter current. I subtract off the leakage from the first measurement and divide by the injected base current to get the current gain. You are saying that since the hfe varies with current and I am taking measurements at two different currents (leakage + w/ small base current),  my calculation for hfe is not accurate (though probably good enough for sorting my transistors)

My cheap DMM lowest range for dc current is 200uA, 5uA is near the limit of what I expect to be able to measure. So if I could measure such a low current accurately I could measure the collector base leakage directly. I guess way I am measuring leakage a higher gain transistor will show more leakage, because the leakage is being amplified.

Andrew.

[PRR]

> higher gain transistor will show more leakage

Yes; and in many circuits this is relevant.

I was just curious why your numbers were near but not-quite double/10C, and was kinda thinking out loud.

That plot from -25C to 100C is a thing of beauty, but not worth the effort to duplicate.

[brett]

Thanks lads, this has been an interesting read.