Close

Testing Germanium transistors (2nd part)

A project log for Clockwork germanium

A retro version of Yet Another (Discrete) Clock, with vintage parts

Yann Guidon / YGDESYann Guidon / YGDES 04/08/2016 at 22:052 Comments

Before I can proceed with the debouncing circuit, I have to evaluate the characteristics of the transistors and choose the lousiest (debouncing is not a critical circuit). Fortunately I had built a board for manually Testing Germanium transistors.

Remark: the gain and leakage are tested at 10V but this clock works betwwen 4 and 5V. I had expected leakage to be exponential but it appears to be logarithmic vs voltage... so the values will not change much in practice. The leakage moves (drops during the measurement) quickly too, and it very sensitive to temperature, which is disconcerting. I suppose that the large variations are due to the varying time to manipulate them with warm fingers, varying times to measure, etc.

Oh and gain seems to decrease with temperature... spooky...

I have no experience yet to deal with leakage so I don't know how to proof my circuits against it (and low gain). I would expect a minimum gain of 20 so I can have some margin with a 1:10 resistor ratio for the feedback.

Additionally, leakage seems to be a major obstacle to low power consumption and I hope to get some skills with this design.

Let's now pick 10 MP13 and 10 MP26 at random and test them:

МП13Б (MP13B) :

Off (mV)On (mV)Leakage (µA)Gain
108, 83
516, 470
43, 33
40.8, 38.7
310, 260, 220, 200
770, 705, 640, 623
125, 105, 88, 81
46, 44.5, 42, 42,3
220, 200, 180, 160
640, 608, 580, 543
88, 81, 73, 64
42, 40.8, 40, 38.3
330, 280, 200
740, 680, 580
133, 113, 81
41, 40, 38
230, 200, 180, 160, 100
555, 517, 490, 460, 372
93, 81, 73, 64, 40
32.5, 31.7, 310, 30, 27.2
250, 210, 180
750, 690, 650
101, 85, 73
50, 48, 47
250, 100, 85, 66
(i've left it cool a while)
660, 392, 372, 339
101, 40, 34, 26
41, 29.2, 28.7, 27.3
350, 280, 220
790, 700, 620
141, 113, 88
44, 42, 40
220, 200, 160
645, 613, 560
88, 81, 64
42.5, 41.3, 40
230, 200, 180, 130
615, 575, 515, 472
93, 81, 73
38.5, 37.5, 33.5, 34.2

МП26А (MP26A):

Off (mV)On (mV)Leakage (µA)Gain
350, 300, 270
560, 500, 470
141, 121, 109
21, 20, 20
600, 500, 400
880, 780, 660
243, 202, 161
28, 28, 26
230, 210
407, 385
93, 85
17.7, 17.5
360, 320, 250
590, 550, 470
145, 129, 101
23, 23, 22
210, 190
450, 427
85, 77
24, 23.7
240, 210
465, 428
97, 85
22.5, 21.8
300, 250, 200
500, 420, 363
121, 101, 81
20, 17, 16.3
220, 200, 190
413, 390, 378
88, 81, 77
19.3, 19, 18.8
430, 400, 300, 250
630, 600, 480, 430
173, 161, 121, 101
20, 20, 18, 18
470, 400, 300
725, 646, 531
190, 161, 121
25.5, 24.6, 23.1

As you can see, I don't use spreadsheets. I am THAT oldschool. Though I might try to import that in gnuplot, who knows.

Now the most interesting part : МП26А's gain is about 20 and МП13Б is about 40. From the reference number I would have expected the reverse... Once again, it was important to measure !

Alexander confirmed: "without suffix it's 12, with suffix 20...60"

So I wanted to pick the "lousiest" and the МП26А (about 75pc) is selected for the debouncer.

At "room temperature", the leakage is not excessive but 100µA is still much more than what's needed to run a quartz oscillator or other things we take for granted in the CMOS era...

Question : is gain correlated with leakage ? Should I select high or low gain transistors then ? Since the temperature varies widely, it's hard to sort parts by leakage, but I should see if the gain has a positive influence...


Alexander Shabarshin provided more insight:

(I don't understand russian, BTW :-D)


I also got 10×GC117C. The leakage seems to be lower. 2 parts are dead and the gains are all over the place: 37, 62, 50, 60, 50, 75, 36, 60. Now I understand why "matched pairs of Germanium transistors" was such a thing.

They seem to be a better fit for a Xtal oscillator but I still wonder if the gain is high enough...

For the record:

GC117c low noise PNP RFT

Ge PNP 20V 50mA 70mW F < 10dB

According to http://www.radiomuseum.org/tubes/tube_gc117.html:
Introduced in 1963, made by RFT/HFO/VEB (germany), similar to OC817 (300KHz max)

Discussions

add_ocean wrote 09/26/2017 at 22:10 point

I know reverse base-collector current, it rarely reaches 5 microamps for all those МП at room temp., so putting 10K resistor between base and emitter makes transistor to be reliably closed, zero leakage. Are we going to leave base floating in the air?

Secondly, gain should be measured under rated collector current, and less current renders much lower gain in case if Ge.

Amplification Ge rated at 1 mA usually, but UHF or switching transistors can be rated at 5 or even 25 mA. МП26А rated at 2,5 mA, btw. (i previously mentioned here 25mA but sorry, checked again, its МП20 and 21 amplification rated at 25 mA). Most trusted book for these is by Goryunov, nothing else.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 09/26/2017 at 22:16 point

that's good to know, spacibo :-)

  Are you sure? yes | no