Germanium Transistors - testing Hfe and leakage

[Stickman393]

A'ite:

Am I correct here in presuming that the cheapie amazon/aliexpress TC-1 multi-function tester is essentially useless for testing germanium transistors?

Some it seems to do OK on, some it's just like...way off vs the RG Keen method. An AC128 on the TC-1 reads 320uA leakage and 142 gain, but the RG keen method yeilds 298uA leakage and 182 gain (corresponding to measurements of 1: 0.738Vdc and 2: 2.12Vdc).

I'm using a trimmer to dial in exactly 2.472k ohms as measured by my Fluke 87V. I'm inclined to believe that this setup is going to be more accurate, especially considering that some of these transistors seem to take some time to stabilize once under test: a quick two second test doesn't seem like it would be able to accurately account for this.

This is especially true for the AC128's and OC75's I've tested: these take a solid 5 minutes to stabilize. The USSR era MP41's seem to stabilize after about 30 seconds or so.

So...hey, sounding board, yeah? Whatcha'll think?

 

[Brett]

Last I looked, the TC-1 (and others in the same vein) do not factor leakage into gain calculations. The RG Keen method does. Your leakage numbers are reasonably close to each other, but this could account for at least some of the discrepancy you're seeing.

Your results with the RG Keen method will be more accurate.I still believe that the inexpensive testers like the TC-1 still deserve a place in the toolbox because they're useful for quick preliminary testing and rough sorting.

 

[Brett]

but the RG keen method yeilds 298uA leakage and 182 gain (corresponding to measurements of 1: 0.738Vdc and 2: 2.12Vdc).

It's also worth noting that your calculations are incorrect here.

It's (#2 - #1) * 100 = gain

That'd be (2.12 - 0.738) * 100 = 1.382 * 100 = 138.2 for gain, not 182.

Much closer to what the TC-1 is telling you.

 

[Stickman393]

It's also worth noting that your calculations are incorrect here.

It's (#2 - #1) * 100 = gain

That'd be (2.12 - 0.738) * 100 = 1.382 * 100 = 138.2 for gain, not 182.

Much closer to what the TC-1 is telling you.

Mmm, that's not the way I understood the article. From this section here in the section referring to leakage testing, before the 2.2M resistor is connected:

A 2472 ohm resistor is 2.472 volts per milliamp, so a milliamp of leakage will cause 2.472 volts to display.

What Keen refers to here is ohms law: if the meter reads 2.472Vdc, and the resistive load is 2472 ohms, then the current passing through the load is 1mA. With the equation V/R=A, as long as we know two of the three values we can calculate the third.

In my specific case, actual leakage current will be 0.738 (V) / 2472 (R) = roughly 0.000317A=0.317mA

Subtract that from the measured gain value: 2.12 - 0.317 = 1.803 x 100 = 180, roughly (though the numbers I provided here were rounded a bit from what I was working with).

That's what surprised me about how far off the TC-1 is. It has a tendency of veering way off the Keen methodology.

That said: I don't know the math behind his HFE equation, and why 2.2Meg and 2.472K with a 9v power supply will yield a direct gain measurement.

Although...huh...I really should measure my power supply voltage: that's gotta have an impact. 9.12Vdc. That could be throwing off my measurements...how to compensate for that though...

 

[Brett]

The previous comment was based on this excerpt from: http://www.geofex.com/article_folders/ffselect.htm

"Let's say the device really leaks 93uA, and has a gain of 110 - a prime specimen. What happens when we test? We chuck the thing in the socket, and read (93uA)*(2472) = .229V. Then we press the switch, and read 1.330V. To get the real gain, we subtract 0.229V from 1.330V and get 1.101V. The true gain is just 100 times the reading."

I'm basing my calculations on the voltages you provided and the excerpt above.

I don't know the math behind his HFE equation, and why 2.2Meg and 2.472K with a 9v power supply will yield a direct gain measurement.

I'd like to better understand the math myself. An equation would go a long way in understanding how to compensate for resistor inaccuracies.

9.12Vdc. That could be throwing off my measurements...how to compensate for that though

I'd expect some level of deviation if the supply voltage is different from the ideal testing conditions. Once again, having an equation for the calculations would prove useful.

 

[Stickman393]

"Let's say the device really leaks 93uA, and has a gain of 110 - a prime specimen. What happens when we test? We chuck the thing in the socket, and read (93uA)*(2472) = .229V. Then we press the switch, and read 1.330V. To get the real gain, we subtract 0.229V from 1.330V and get 1.101V. The true gain is just 100 times the reading."

Ah ha! That makes a bit more sense. For some reason I had gotten it in my head that it was the actual leakage in mA rather than the measured voltage that had to be subtracted.

Woof. Appreciate the heads up there. That saves me some head scratching.

 

[Big Monk]

It’s not useless but it is generally more accurate the lower the leakage is.

I’ve had it compare favorably to the RGK method up to 100microamps of leakage.

 

[Stickman393]

Word, Monk-ey, word. Don't seem to have too many under 100uA at the moment.

I've had COVID over the last few days here (figures, 4 day union holiday, OUR union holiday, god's a prick.) The most I've been able to muster yesterday and the day before while isolating from the rest of my family was going through my stacks of these things and documenting. Thankfully I only got through the PNP pile, now I gotta go and revise all those gain figures.

My wife asked me what I was doing. When I told her...she couldn't begin to fathom.

"...That sounds so...*Boring*..."

"sho is, sweetie. sho is."

 

[Fartman]

Hey if iceo is leakage and a transistor measures at 3.3ma, is that 3,300ua of leakage? Sorry for bumping with a question I didn't think it warranted a new thread

 

[Coda]

Hey if iceo is leakage and a transistor measures at 3.3ma, is that 3,300ua of leakage? Sorry for bumping with a question I didn't think it warranted a new thread

Yes. Converting mA to uA moves the decimal three to the right…

 

[Stickman393]

I found the TC-1 doesn't quite measure accurately up around that range.

That would be way too much leakage though. That guy would go in my pile reserved for horror scenes re-created with electronic components.

I'm inclined to trust the RG keen method the most; theres something to being able to watch a transistor stabilize over a couple of minutes vs having an instantaneous result...

 

[Locrian99]

The peak works really well for this. If youre sorting 100 its worth the investment.

The dca55 will do hfe and leakage.

If you want to do jfets too you need the 75

 

[Big Monk]

I found the TC-1 doesn't quite measure accurately up around that range.

That would be way too much leakage though. That guy would go in my pile reserved for horror scenes re-created with electronic components.

I'm inclined to trust the RG keen method the most; theres something to being able to watch a transistor stabilize over a couple of minutes vs having an instantaneous result...

If you have a known set of transistors with leakage of < 100 microamps, the TC1 is fine.

Lots of Russian and Mid 60s/Early 70s American Ge transistors will fall into this category.

But most of the rest of the shit out there has been picked over and DEFINITELY needs the RG Method.

 

[StompBoxParts]

Breadboard it. Throw a meter on it. RG Keen is the way.

TC1 does not do well with germanium stuff. DCA55 is better but still will not catch an unstable transistor. When testing germanium stuff, a snapshot tester is really inadequate for the job. You gotta see how the numbers change over at least a few seconds, ideally more.

 

[Ginsly]

I'll also add that the TC1 is oddly not great for testing silicon transistors, I've found. Any other testers (DCA55, ports on most multimeters, RG Keen) show a much lower hfe that is likely more in line with what a pedal builder would count on as "accurate". I realize the TC1's readings for silicon are technically accurate as well, but the testing conditions produce results that are way different than others I've tried.

 

[Big Monk]

so i've just read this statement:
"for a transistor to function well in an original fuzz face circuit is not the correct gain, but rather the correct amount of leakage to bias the base."

honestly i hadn't really paid a lot of attention to the voltage bias at Q1 base.
seems to me, since these MP38s are so low gain (~30), perhaps the leakage may not be enough in some to bias Q1 base properly?
(and maybe that's why that super leaky transistor in Q1 i first tried sounded alright...)
View attachment 82645
i did notice Q1 base was about 0.1xx V at one point (after inserting the 'properly measured' MP38s)

- so, it looks like i need to something in Q1 that will leak enough to get the bias at Q1 base up to at least 0.2V.
am i on the right track?

Leakage is only a “requirement” if you are dead set on using the original circuit values.

The circuit doesn’t need any leakage at all if you modify the component values.

Leakage is only a part of the design of these old circuits because savvy engineers made lemonade back in 60s

 

[Stickman393]

Somebody check me on this?

My understanding is that leakage presents several issues in a circuit:

1) Correctly biasing leaky transistors is a bit trickier than biasing non-leaky transistors (I need to understand this a bit more myself)
2) Noise? (electrons flowing when they should not be flowing=raised noise floor?)
3) Heat: a steady stream of current through the transistor will cause a leaky transistor to heat up more than a non-leaky transistor, which could have various other effects on it's operation.

 

[Big Monk]

Somebody check me on this?

My understanding is that leakage presents several issues in a circuit:

1) Correctly biasing leaky transistors is a bit trickier than biasing non-leaky transistors (I need to understand this a bit more myself)

Yes.

2) Noise? (electrons flowing when they should not be flowing=raised noise floor?)

To be fair, I’ve had plenty of NOS Silicon transistors that were just as noisy and hissy as Germanium but in general, yes to this to.

3) Heat: a steady stream of current through the transistor will cause a leaky transistor to heat up more than a non-leaky transistor, which could have various other effects on it's operation.

I don’t know the exact mechanism by which temperature induced voltage shift happens but I imagine it’s as you describe.