Germanium transistor leakage issues

[Lücking]

Hey fellas
I've bough a load of different NOS germanium transistors to see what the hype is all about. Bought them off eBay from a Romanian seller, knowing i'd take a risk to get a bag of lemons. Bought GT308V, GT403B, GT321V and MP20A's.
I built the RG germanium tester to test them.

Suprisingly the MP20A turned out decently, not one with a leakage over 200 mA and the average hFE around 90-ish, out of a 40 pc batch. Only 3-4 lemons with 40 hFE, almost all the other ones was over 70 and the highest 128.

However, the next ones i tested was the GT403's. And oh my, that's insane. Which makes me wonder, can this really be true? Highest hFE was 40, and the AVERAGE leakage is over 1300 mA! One even tests as 2300 mA leakage, and the lowest is 700 mA... Am i doing something terribly wrong, or could they really be THAT bad?

Even when the temperature has dropped after handing (with wool gloves to prevent more heat than neccessary) and been stabilized for like 10 minutes, the ratings are the same range.

The GT321's doesnt look too good either, the first ones has been over 1000 mA aswell...

Any idea on what i could be doing wrong, or is these transistors just terrible through the roof? Thanks in advance!

[Lücking]

The GT308V's turned out pretty crap aswell. Very low leakage (highest was 40 mA), but very low hFE's aswell. 4 in the 70's, 32 out of 50 under 55 HFE... Bummer. But i'll try them and see how they sound anyways

[NFX]

When I've been testing I've have leakage in the 1000uA range and ide say 2 in every 50 I have tested have been high. I never throw any away for having odd values as u never know what they will sound good in. Early next year I am bringing out a project which utilises all yer leaky germs so hold onto them.

Bev @ Newbury Effects

[tonyharker]

Don't you mean uA microamps not mA milliamps?

[Lücking]

Don't you mean uA microamps not mA milliamps?

Well i'm not 100% sure but i don't think so, as i'm getting some very low readings on some of the other ones like the GT308's at like 8 mA. 8 uA would be insanely little.

Calculating it like this, which is the one i've built aswell. Using the excel document linked somewhere in the comments set up to calculate it from the inputs.
http://tagboardeffects.blogspot.dk/2012/08/germanium-transistor-tester.html

[EBK]

Early next year I am bringing out a project which utilises all yer leaky germs so hold onto them.

I was thinking of something similar, but you'll definitely beat me to it.  Feel free to use either of these names if you like:

Deep Ends Germanium Incontinence Control Fuzz (I guess if anyone lives in a country where Depends is not a recognizable brand, they won't get this one...)

Sunset Village Assisted Germanium Fuzzing

[TejfolvonDanone]

Don't you mean uA microamps not mA milliamps?

Well i'm not 100% sure but i don't think so, as i'm getting some very low readings on some of the other ones like the GT308's at like 8 mA. 8 uA would be insanely little.

Calculating it like this, which is the one i've built aswell. Using the excel document linked somewhere in the comments set up to calculate it from the inputs.
http://tagboardeffects.blogspot.dk/2012/08/germanium-transistor-tester.html

I don't mean to be offensive but:
The tester uses a 2.4k-ish resistor. To pump 2300mA (=2.3A) through that you need around 5500V!!!
Also putting 2300mA current from a 9V supply means it is putting out almost 20W.

It's either you are really measuring micro amps or the tester is not working properly.


BJTs tend to fail with collector to emitter short which means that you can get really high leakage readings.

[duck_arse]

Well i'm not 100% sure but i don't think so, as i'm getting some very low readings on some of the other ones like the GT308's at like 8 mA. 8 uA would be insanely little.

I was doing this very testings today. but I have an old sheet of numbers which includes a pair of GT308B's leaking 18uA and 20uA, and a pair of GT313A's leaking 7uA each.

if you're using the RG tester and you are reading current from/with your meter, you aren't doing it right/using the RG method. you read the voltages across a resistance, and let Mr Ohm and some maths do the rest.

[BetterOffShred]

Don't you mean uA microamps not mA milliamps?

Well i'm not 100% sure but i don't think so, as i'm getting some very low readings on some of the other ones like the GT308's at like 8 mA. 8 uA would be insanely little.

Calculating it like this, which is the one i've built aswell. Using the excel document linked somewhere in the comments set up to calculate it from the inputs.
http://tagboardeffects.blogspot.dk/2012/08/germanium-transistor-tester.html

I don't mean to be offensive but:
The tester uses a 2.4k-ish resistor. To pump 2300mA (=2.3A) through that you need around 5500V!!!
Also putting 2300mA current from a 9V supply means it is putting out almost 20W.

It's either you are really measuring micro amps or the tester is not working properly.


BJTs tend to fail with collector to emitter short which means that you can get really high leakage readings.

I'd wager he's talking Microamps.. and hasn't used the alt code for mu? 
alt + 2 3 0
µ   boom.    µa

[duck_arse]

[my keyboard got no "boom" key. stares back blank when I do alt 230.]

[BetterOffShred]

Sounds like a faulty device!   One thing that really does irritate me about alt codes is you have to use the number pad .. you cant use the keys above the letters..  dumbbbb!   

I tested a bunch of AC128's recently and some that were like .400 µA leakage and higher.. yikes!     all my MP20's and MP42's were under .200 µA.. 
and.. µµµ..
µ

[duck_arse]

[fun! I tried alt codes in a terminal window, returned (arg: 230), still no booms.]

however, my two MP20's show 99 µA and 117 µA. if I found some AC128's like yours, reading 0.4 µA, or 400 nA, I'd be a bit suspicious.

[BetterOffShred]

I guess technically since the number is derived from a voltage reading divided by the resistance (I'm using the Geofx tester) it's up to the operator to know where the decimal place goes..  so my .400 value would be milliamps.. or 400 microamps.   Damn operator error!

And furthermore, I'm using an embedded conversion by dividing by 2.742 instead of 2742.. the later provides correct units.

1.1v/2742r = 4.01E-4 .. or 401 microamps.  Gee whiz 

[thermionix]

µ

Hey it works!  Had no idea.

And I prefer the number pad, much faster for me.

[PRR]

[my keyboard got no "boom" key. stares back blank when I do alt 230.]

This technique is tricky.

It goes back to the IMB 5150 PC. 96 keys but an 8-bit internal representation which could store 255 characters. The BIOS interpreted alt-numpad sequences to single bytes in the key buffer. (Numpad to make it harder to do by accident.) What happened after that was up the the program which got these "keystrokes". Some barfed, some rounded-down to 7-bit, some assumed a display character in the "code page"; and while the default IBM codepage covered many european character sets, it could not handle all; and then there are the Asians.

While rarely used, this feature was used enough that MS had to emulate it as Windows bypassed BIOS keyboard routines. Windows added new codepages; 3-digit numpad was assumed BIOS and 4-digit with leading zero was assumed Windows. But ambiguity happened. Anyway old Macs and Unixes did not do it this way. HTML had "entities" instead of IBM-derived fancy fingerwork. UniCode was supposed to rescue us, but merely confirmed the story of Babel.
https://en.wikipedia.org/wiki/Alt_code

Alt-230 works on my screen, but I can not be sure it will work for ALL readers. I let "u" do my micros. That may offend Greek typographers but in electronic units it is universally clear.

[highwater]

I remember having to type the leading-'0' or else it didn't work, but I don't think I've tried on a windows box since XP. I could never remember the codes, and if I'm using Character Map to find them it's just as easy to copy+paste.

Try alt-0181.

Long story short, the alt-code behaviour apparently changed sometime between XP and 8.1 (either that, or I come from a parallel universe). Used to be that alt-0xxx was octal, and alt-xxx didn't work (this is what I was remembering). Apparently, nowadays, alt-xxx is *decimal* on one code-page, and alt-0xxx is also decimal, but on a different code-page.

P.S. - on a Mac, 'µ' is Option+M.

[duck_arse]

parallel universe here - strictly speaking, using linux, I don't have an "alt" key either, I have "meta". either way, it ain't booming. the top-row numbers give a different reaction to the keypad numbers, but fires blanks either method. with or without the leading 0.

[BetterOffShred]

I've learned so much in this thread! 

[antonis]

using linux,

They live among us...!!!

[PRR]

> using linux

Did you see and read the link I gave? There is a section on "Linux". It needs more fingers than I can manage; also you want a Unicode Table (which is linked in the link). I could boot a Linux, but unix hates me, so why try?