Getting beat by germanium testing

[trjones1]

One thing that has consistently been a failure for me is trying to measure gain and leakage of germanium transistors.  I have no idea why, I just never get the testing mechanism to work right.  I've tried RG's testing setup and now I'm trying the one described at smallbear: http://www.smallbearelec.com/HowTos/FuzzFaceFAQ/FFFAQ.htm

I am getting leakage current readings that make sense, but when I apply forward bias to the base, it does not effect the collector current at all.  I don't even understand how this is possible.  From the diagram on the website it looks like the leakage current is measured with the DMM connected between the positive lead from the battery (I'm testing NPNs) and the collector.  Then you connect the positive lead to the base through a 1M resistor and again measure the current from the positive lead to the collector.  Is this correct?  Anyone have any idea why connected the 1M to the base has no effect on collector current? 

[jrod]

Hey,
I've not used the SB method, but have used RG's instructions. It took me a while to get it going though. Not sure why, but it did. Pluse I was confused by the math.

What I did was breadboard the circuit and pull the 2.472 resistor in and out. I conect my DMM to the other resistor. I am going to build this diagram soon, so I have a perminant tester: http://www.buildyourownclone.com/board/viewtopic.php?f=31&t=2511

I think the math is WRONG on the thread above. If not then I have a whole bus load of ge's with wrong numbers.

I hope you get it working. I'll help you anyway I can. Like I say, I had a hard time with it at first.

[ericohman]

I built one today (RG) on perfboard because I was tired of having it on a breadboard...


I used a 2.49K resistor because it was all I had, but then again, I use it with batteries so it's only to get an estimate of gain / leakage. Besides, it varies a lot with temperature so everything could be considered estimates...

I posted this link in an iPod/iPhone thread here on diystompboxes. This is what I use, if you find any errors with the math let me know, though I'm pretty sure it's correct since it's been discussed before.
http://djoman.dyndns.org/germanium/

[jrod]

Cool! This is exactly what I need to do. Thanks for the pic.

Regarding the math, I am refering to the link I posted to a BYOC thread. It appears to me that the math does not match RG's instructions. Check it out see what you think.

[mac]

This is the math behind RG test. You can scale it, ie, use another resistors if you need to know hfe and leakage at different base currents. Note that hfe is a function of collector current.

http://www.diystompboxes.com/smfforum/index.php?topic=45481.0

mac

[R.G.]

I'm always a little taken aback by the confusion which comes up over this very simple circuit.

Here's what's going on.

1. Germanium leaks. Silicon also leaks, but about a thousand times less than germanium, so it can be ignored.
2. Germanium leakage makes a collector current flow even if the base is open (that is, base current is zero)
3. DMM hfe tests can't work right for germanium because they assume no leakage. They just put a fixed current into the base and read out the collector current as HFE. The base current is scaled so that it's one unit of the collector current readout, so the collector current equals the number of times bigger than the base current, and equals HFE numerically.
4. In a stock hfe test, the collector current appears as an additional hfe that's not real - the hfe number appears artificially high because the meter ignores that some of the collector current is leakage.
5. In order to get down to the difference between leakage current and real gain - that is, current in the collector that's a multiple of the base current - you have to measure the leakage, then subtract it from the result of the base-current-multiplied current.

Everything else is just getting the numbers to scale correctly.

This is very much like measuring how tall people are if you are limited to measuring from the ground upwards. If the people to be measured are standing on the ground (that is, at the zero point on your measuring stick), you can just measure how tall they are. But if they're standing on a box, they appear to be artificially tall when measured from the ground. You have "height leakage" because some of the "tallness" comes from the constant box they're standing on. One way to figure this out is to measure the box, then subtract that measurement from the total measurement. That gives you the correct tallness.

My hfe measuring rig is not perfectly accurate. That's because there's little point in being perfectly accurate: hfe changes with operating current and temperature. but it sidesteps the gross errors of leakage. Here are some inside details about the compromises:
- the base current is not constant, it's only approximately constant. It changes with the battery voltage and the exact value of that 2.2M base resistor, both of which have some tolerance variation. But it's much more accurate than the leakage
- the value of that 2472 ohm resistor throws people in much the way that a magician's misdirection makes you look where the trick is not happening. A resistor is a current to voltage converter. It makes a voltage appear that's the current times the resistor. So we can use resistors to make any voltage we want appear from any current. That resistor's value is selected to make the output voltage come out right in numbers. The voltage across the resistor is 1/100th of hfe times the (approximate) base current in amperes.

This is the same trick the DMM guys used to get the reading to come out right on the meters - they really read either current or voltage, but the internal resistors are preselected to make the numbers come out right. In this case, I don't control the placement of the decimal point on the meter display, so you have to multiply by 100. This is pretty simple - just move the decimal point two places to the right.

And that's the underlying reason for the choice of the 2.2M resistor and the 2472 ohm resistor - it's to get the number on the meter to come out correctly so all you have to do is move the decimal point.

Done properly, the only math is to move the decimal point two places to the right, and subtract two numbers. No partial differential equations, no semiconductor physics, no second integral of elliptical functions. 

As Mac properly notes, you can scale it all you want. I did it trying to make it super easy. But other scaling factors work too. You'll just have to do the math with them to get the numbers to come out.

[aron]

I used to hate doing this. I still have a number of Ge transistors - low leakage, but they are out of the "best" hFE values for Fuzz Faces. I may put them up at some point in the store anyway.

[mac]

I prefer messing with hyper geometrical confluent eqs than measuring germs

I have a lot of low gain-leakage germs, hfe 40 or so. There must be some use for them. Maybe a 3 stage circuit a la Joe D.

mac

[jrod]

Sorry to hijack your thread trjones1  

Thanks Mac and RG for the info.

Personally I get the circuit and understand the reasoning behind it.

I suddenly became confused when I read a different explanation on a different forum.

Please, tell me if this correct.

DMM set to VOLTS.
V accross the 2.472 reads .300.
Now, button pressed, 2.2M added and I read 1.70 on the DMM

1.70 (false gain) - .300 (leakage) = 1.40*100 = 140 Hfe (true gain)
.300/2.472 = .121 so, 121uA of leakage

Bingo, Q2 for a Fuzz Face under 300uA of leakage.

Is this correct?

[mac]

Muy bien!

Leakage is much more problematic at Q1.
I mean you can use a leaky ge at Q2... unless leakage is related to time before failure.

mac

[jrod]

Muy bien!

Leakage is much more problematic at Q1.
I mean you can use a leaky ge at Q2... unless leakage is related to time before failure.

mac

Thank you, Mac!

So I am doing the math right? Woo hoo! 

[ericohman]

Look at these AC128's of mine...bought them off eBay last year

Considerer yourself lucky

[mac]

You can try the first two black germs. With some tweaks you can make a nice FF.

mac

[aron]

The low hFE germaniums actually can sound quite good in a fuzz face.

[mac]

The low hFE germaniums actually can sound quite good in a fuzz face.

making the feedback resistor 150k and increasing Q1 collector resistor to 47k might compensate the use of lower gain ge.

mac