Testing AC187K / AC188K germanium transistors - a learning experience

[1980s_john]

As part of my repair to a Grundig Concert Boy 1100 radio, I suspected one of the two output transistors had failed. These are both germanium, AC187 is NPN and AC188 is PNP, act as a push-pull pair (nb K variant just means transistor has a fitted heatsink block).

I removed them and tested them on my DMM, a UNI-T UT30A (not a recommendation, happened to be under £10 in Maplin one day). I checked the PN junctions using the diode test, and all four measurements were 0.2V in the forward direction and OL (open circuit) in the other so all good.

DMM has a basic hfe test socket, so I tried this, gave AC188 hfe=95 and AC187 hfe=140 (I'll come back to hfe later). This appears to be good at first sight, but I thought the pair ought to have similar hfe so I went out and bought two NOS transistors from different sources (Langrex / yitry on eBay and Fuzz Dogs Pedal Parts on the web).

I tested the new AC187K first on the DMM, PN test was fine, but hfe gave a very odd result. The meter gave a very high reading (over 900) then flashed OL meaning greater than 1000. I googled for help and found comments that basic DMM hfe testers are designed for silicon low leakage transistors, and gave give wildly wrong hfe measurements for germanium types especially if leaky:

https://www.tdpri.com/threads/hfe-testing.391898/

I looked around and considered getting a commercial transistor tester, either a vintage one (such as a Heathkit) or a modern one (such as a Peak Atlas DCA55 or 75). In the end I couldn't justify the cost so followed the basic tester shown here:

http://www.geofex.com/Article_Folders/ffselect.htm

which tests basic DC (static) performance of a transistor. I used a 9V supply, a 2.7kohm resistor for the collector leg and a 680kohm one for the base. The article gives the method to measure Iceo (current from collector to emitter with base open circuit) and gain (Ic-Iceo)/Ib.

Using this set-up (with wires and croc clips) I checked the 4 transistors, 2 old and 2 new. I started with the new AC187K. With the base disconnected the voltage across the 2.7kohm resistor (Vrc) was 3.75V. Therefore: Iceo=3.75/2700 A = 1.4mA. Seemed very high!

I connected the base resistor and measured voltage across it of 8.75V (Vrb), so base current = 8.75/680k A = 0.013mA or 13uA. This was pretty much the same for all 4 transistors. Collector current = 3.0mA.

So increase in collector current due to base current = 1.6mA, gain = 1.6/0.013 = 123

This is much better figure than my DMM gave, showing how unsuitable it was to measure gain / hfe.

I repeated the above for the other 3 transistors, results were:

Old AC187K Iceo=0.5mA, gain = 76
Old AC188K Iceo=0.26mA, gain = 60
New AC188K Iceo=0.5mA, gain = 107

I looked up specs for the AC187/8:

https://pdf1.alldatasheet.com/datash...TC1/AC187.html

and this gave figures for Icbo (leakage from collector to base with emitter open) as up to 35uA @ 10V, but no figures for Iceo.

I went back to the new AC187K and disconnected the emitter and base resistor, and connected the negative side of the PSU to base. I measured the voltage across Rc (33mV) and calculated Ibco as 12uA.

I figured that Iceo=Icbo x gain, this ties up with the measurements above ie:

1.5mA= 12uA x 123 (approx)

In conclusion although the Iceo measurements of the new transistors showed more leakage than the old, the newer ones had about 70% more gain and were within spec for Icbo. The old transistors hadn't failed completely, so could probably work but at much reduced performance.

I haven't checked Vce limits or reverse breakdown, needs further study to do safely.

Regards,
John
PS
Note that hfe is a dynamic measure of a transistor, and is (delta Ic)/(delta Ib) for a given Vc ie you increase the base current then measure the increase in collector current. Therefore hfe should ignore leakage - my basic DMM hfe measurements clearly did not.

I found this reference (for a vintage Tektronics curve tracer) of interest:
https://web.mit.edu/jhawk/tmp/Using_Tek_575_opt.pdf

[dave_n_t]

John,
Thanks for a really clear exposition.
I agree with your assumption that Icbo x hFE = Iceo (but can't recall it having seen that in print).

A couple of extra points, if I may:

i) if the current passed during testing is enough to warm the junctions up, then the leakage will increase dramatically. So it's best to do things quickly, if possible (as the Peak & generic Chinese testers do)

ii) my multimeter has a 'transistor test socket', which feeds a fixed base current in + measures collector current. I know (from having used a second meter!) that a gain of 100 corresponds to 1mA of collector current (and so the fixed base current is 10uA). Knowing this, two 'tests' can be made - with base connected, reading a 'gain' of g; and with the base disconnected, reading a 'gain' of h. The true gain is then about g-h; the Iceo is 10h uA; and the Icbo is 10h/(g-h) uA.

dave

[Boulevardier]

"...basic DMM hfe testers are designed for silicon low leakage transistors, and gave give wildly wrong hfe measurements for germanium types especially if leaky".

So, does anyone know if the cheap chip-based Chinese testers are similarly useless for germanium semiconductors?

Mike

[Refugee]

I have got a Chinese TC-1 and it is unreliable at testing germanium transistors.
It works fine on all the other ranges including silicon transistors.

[Keith956]

hFE varies widely with Ic, the datasheet figures for AC187/188 are at Ic=300mA. At low currents it's likely to be less; I'm assuming that the simple DVM testers assume small signal transistors not power transistors so may be testing at maybe 1mA or so.

btw it's Iceo = Icbo x (1+hFE), a small difference except for low gain devices.

[Refugee]

Back in the day germanium I built a transistor with fixed VCE and several IC settings from 2 to 100ma. It has one leakage setting.
The meter was in the base with a pot in series. HFE was obtained with a slide rule and later a calculator.
Much later a bench power supply got me up to a couple of amps.
I must build a new one with a the op-amp wired like a servo and a second to indicate if the transistor is either short or open.

[John_BS]

I have a BBC transistor tester dating (I'd guess) from about 1960. It measures Ico as well as beta, as you'd expect at a time when germanium was prevalent.

If anyone would like the circuit please pm me.

John