R.G. Keen's "Picking transistors for FF Clones" question

[bushidov]

Hi All,

I am trying out some Germanium PNP transistors (and some Germanium NPN ones too) that I got from Small Bear a while ago and wanted to try R.G. Keen's article on measuring leakage and calculating the true gain.

The article is here: http://www.geofex.com/article_folders/ffselect.htm

So, firstly, I did a PNP Germanium transistor.


Got a leakage of about 101.5uA, and when the 2.2M resistor was applied to the base, I got 0.842V, subtracted the 0.251V I was getting from the leak, so 0.842V - 0.251V = 0.591V or gain of 59.1

So far, so good.

Like the article suggested, I tried an NPN Germanium Transistor next. Flipped the 9V and GND power supply as is mentioned.

Got a leakage of about 29.1uA, and when the 2.2M resistor was applied to the base, I got 0.241V, subtracted the 0.072V I was getting from the leak, so 0.241V - 0.072V = 0.169V or gain of 16.9

And that's about what Small Bear wrote on the baggies, so that all seems good.

Then I did an NPN Silicon Transistor, (2N3904) just to sanity check myself. I shouldn't see any leak. So...


And indeed, I got 0 leak, or atleast less than 1uA of leak, as my Fluke multi-meter couldn't see it. When I applied the 2.2M resistor to the base, I got 2.170V, so a gain of 217, which for a 2N3904 sounds okay to me.

Now, the question. How do I do this with a Sziklai pair, where it would be the NPN 2N3904 into the PNP Germanium?

I did this:

But this can't be right, as when I measure the "leakage" by not having the 2.2M resistor connected to the "base", I see the leakage from the Germanium PNP transistor. I thought there would be less or 0? Also, when I apply the 2.2M resistor to the base, I get almost 0 gain. So, I must be doing something wrong, or this doesn't apply to Sziklai pairs.

Help, please.

[bushidov]

Nevermind, all. I figured it out. I got bit by one of the oldest tricks in the book: breadboard failure. My 2N3904 wasn't making good contact.

So, now on my NPN Sziklai pair, my leakage did drop the original PNP Germanium's 101.5uA leakage down to 72.8uA. And the gain definitely went up when I added the 2.2M resistor at the base to about 866, but when I subtracted the leakage, it came out to a gain of 848 (temperature, obviously fluctuates it, but at the time of measurement, that's what I got)

So, here is what I measured:
2N3904 Silicon NPN.
Leakage: 0 (or less than 1uA, cause my Fluke 117 multi-meter couldn't see it)
Gain: 217

Germanium PNP.
Leakage: 101.5uA
Gain: 59.1 (after leakage is taken out)

Sziklai Pair of 2N3904 going into Germanium PNP.
Leakage: 72.8uA
Gain: 848 (after leakage is taken out)

I thought gains multiplied together? 217 * 59.1 does not equal 848. Did I do something incorrect?

[bushidov]

Ok, I thought I had it, but no, that second time must have been when my breadboard was bad. Tried again and am back to the first post results.

So, if I do a Sziklai pair with an NPN silicon transistor to a PNP Germanium Transistor, the leakage from the Germanium transistor still follows as if the NPN didn't exist. I thought Sziklai pairs with silicon/germanium pairs fixed that?

Now, if I add a 100K resistor on the collector pin of the NPN silicon transistor pulled up to 9V, then the leakage goes down, but I didn't think that was mandatory to get rid of the leak.

[willienillie]

Keep in mind that the RG tester can only measure gain theoretically up to 900 (9V supply x 100), practically it's probably a fair bit less than that where the readings become unreliable, I would guess but don't actually know.  Darlington and Sziklai pairs usually have gains in the thousands.

[bushidov]

Keep in mind that the RG tester can only measure gain theoretically up to 900 (9V supply x 100)

Yeah, I caught that after I posted, and you are correct. I think they just roughly multiply, so that would be 217 × 59.1 = 12,824.7. So I'd get walled around 9V - some couple hundred millivolts of non-headroom or a gain of 866. To "sanity check" myself, I swapped out my 2N3904 with a 2N3903, which has around half the gain stats of a 2N3904, and I still got a "gain of 866". So that definitely proves that to be correct.

As of the 100K resistor and when I need it versus when I don't, like most things with Germaniums, I discovered the "actual" culprit was heat. Which I knew was a thing, but I didn't know this was the exact symptom.

So, in the mornings, my garage is pretty cool. Somewhere in the low 70's F. If I don't include the 100K resistor on the collector pin of the NPN silicon transistor pulled up to 9V, I get leakage, but less than the pure leakage of the PNP Germanium it is paired with. If I add the 100K resistor, the leakage drops even further.

However, at around noon, the garage gets pretty hot, somewhere in a high 80's F. Now, if I don't include the 100K resistor, I get the exact same amount of leakage from the Germanium PNP as if I didn't make a Sziklai Pair with that Silicon NPN and just used the PNP straight. If I add the 100K resistor, the leakage goes down a lot, about the amount it went down without it, but when the temp was in the 70's earlier in the morning. The gains didn't seem to vary between either temperature, or with or without the 100K resistor, but I am guessing that was mostly due to the fact that this Sziklai Pair has a gain in the 1000's and is being brick-walled by the 9V power supply (So, I couldn't see a change that well, if I wanted to).