Question about Atlas DCA gain/leakage calculations for Germaniums

[culturejam]

I saw this asked a few times here, but no definitive answers yet (or any that seemed to answer the question in a way that  made sense to me):

Does the Atlas tester's gain measurement of Germanium transistors take the leakage measurement into account, or should I be back-calculating the real gain myself?

If so, is there a quick formula for working out the adjusted gain?

I read through RG's article on leakage/gain testing, and I came up with a simple formula. However, I apply that formula to the values here.......
http://www.muzique.com/news/russian-germanium-transistors/

.....then it seems there are some transistors with negative gain. I doubt the esteemed Mr. Orman would say those trannies were "consistent in performance, low leakage, and a good value for the money," if they were super-low gain or negative. 

I feel like I'm missing a critical bit of info here to arrive at good numbers. I just ordered two sets of NPNs from Russia (actually, one from Bulgaria), and if I'm calculating everything right, the adjusted gains are really not in a usable range for most fuzz applications, and this runs contrary to the general experience I've read here. My numbers are in the same range as the AMZ link above, yet my calculations show adjusted gains in the low teens or even negative, so I must be doing something wrong.

Thanks!

[culturejam]

Hmm. I just looked up the manual online (should've done that first), and got this:

"DC current gain is simply the ratio of the collector current to the base current for a particular operating condition. The Atlas DCA measures HFE at a
collector current of 2.50mA and a collector-emitter voltage of between 2V and 3V."

I'm assuming this means that the gain measurement is the same as in any DMM: just IC/IB

So then, can anyone suggest a workable formula for converting micro-amps of leakage into current gain to be subtracted from the measured hFE?

Thanks.

[culturejam]

I got it all figured out. Nevermind. 

[gmr1]

What was the final conclusion? I assumed that the Peak meter took leakage into account.

[gwpt]

Hi,
I wouldn't mind knowing how to work it out too!
Thanks

[wokeupfalling]

I also would like to use my Atlas DCA to measure gain and leakage for germaniums...

Does anyone have the formula I need to calculate "real gain" from the Hfe and leakage mA readings I get from the Atlas?

RG's setup has worked for me in the past, but I also would like a better understanding of the situation by doing the math with these readings.

I really appreciate it!

[R.G.]

I also would like to use my Atlas DCA to measure gain and leakage for germaniums...

Can you provide a link so I can figure out what the Atlas DCA (??) does to test gain?

Does anyone have the formula I need to calculate "real gain" from the Hfe and leakage mA readings I get from the Atlas?

RG's setup has worked for me in the past, but I also would like a better understanding of the situation by doing the math with these readings.

No problem, if I can figure out what the Atlas does.

[wokeupfalling]

Hi R.G.!

http://www.peakelec.co.uk/acatalog/jz_dca55.html

There's a link to the user guide there as well...

Thanks for checking!

[IvIark]

Yes it does give you the gain and leakage.  It's a highly recommended buy.

[John Lyons]

But is the gain figure calculating in the leakage?

[R.G.]

http://www.peakelec.co.uk/acatalog/jz_dca55.html

There's a link to the user guide there as well...

Nifty device. However, I could find no information in the user's guide to tell me the critical stuff - how leakage current is measured, other than "current which flows when there is no base current", and how forward gain is measured. I'll look for more data.

Until I figure out how the forward gain is measured, there's no good way to relate the leakage measured and any correction to the Hfe.

There's not a lot of mysterious understanding to be had.

1. In a BJT, any current going into the base results in a current HFE times greater coming through the collector, for simple DC conditions. Any current.
2. The BJT collector-base junction is a reverse-biased semiconductor diode junction.
3. Diode junction leakage is approximately constant for a wide range of reverse voltages. Pretty much if you have more than a few volts across it, the leakage is constant up until the voltage is enough to start breaking it over. That includes the reverse-biased collector base junction.

So if you put a reverse bias on the collector-base and open circuit the emitter, a current flows through the collector and base. That is the leakage, and is about constant with voltage changes, but is quite temperature sensitive.

If you also connect the emitter to a voltage lower than the base, the leakage from the collector into the base acts exactly like you injected that same current in from the outside, and is multiplied by the DC HFE.

Here's where the testers muck it up. Most testers simply inject an external current into the base, measure the collector current, and then divide the collector current by the injected base current, and call that the value of HFE. This works fine for silicon because the collector base junction leakage is so massively smaller than the external base current. Any errors from leakage are negligible.

Not so for germanium. Germanium leaks about 1000 times as much as silicon at equal temperatures. So a silicon device may leak 10nano-amperes, and germanium devices may leak ten micro-amperes, just to pick out representative numbers. Two hypothetical transistors, each with a gain of 100 (for argument's sake) should be read as an HFE of 100. Most testers would inject maybe 100uA into the base, read the collector current, which would be 10ma for this made-up example, and get an HFE of 10ma/100uA = 100. If the device was silicon, that's as far as you need to go.

If it was germanium, the leakage from collector to base would be 50uA to 300uA all by itself, no transistor gain needed. So the tester would put in 100uA, and the transistor's internal leakage would leak in another - say 75uA just to pick a number out. The transistor would then multiply the external 100uA and the internal leaked 75uA and get 17.5ma of collector current. The tester would do the math, 17.5ma of collector current divided by the 100uA what was all it knew about, and come up with a "gain" of 175.

That's all there is to it. Most testers ignore the leakage current from collector to base. It is possible to make a tester like the Atlas open the base, read the current leaking OUT of the base with the emitter open, then connect the emitter, put in a test current so it can measure the collector current and do the math to get the correct answer. But I've never seen a tester that would do that, even though it's possible.

The problem with fancy integrated testers is that unless they're clear about what they measure, or unless they have a special mode taking leakage into account, you don't know what they're telling you.

Testers usually don't measure the collector-base current as "leakage". They measure the collector-emitter current with the base open, and call that "leakage". It's a valid thing to do, but it's not the same leakage, and it confuses people. Both are - or were - specified on datasheets. The collector base leakage is spec'ed as I_cbo, the current from collector to base with the emitter open. The collector-emitter leakage is I_ceo, the current from collector to emitter with the base open. I_ceo is the HFE times the I_cbo. It's not all that useful, since you don't know the HFE that makes I_cbo into the I_ceo.

For silicon, I_ceo is so small that it is simply ignored. For germanium, perfectly good working devices will leak tens to hundreds of microamps with no current into the base at all. Most testers simply assume that leakage and gain are unrelated, and silicon devices let them get away with it.

For a germanium device, you have to measure I_ceo (my tester does this), then apply a fixed base current (my tester does this), then measure the collector current with base current applied (yep, my tester does this too). You then have to subtract out the I_ceo from the collector current measurement before dividing by the base current to get real gain separated from the leakage current.

The Atlas certainly has the capability to do the process: measure collector-emitter leakage (I_ceo), insert a fixed base current, measure total collector current, subtract out the measured I_ceo, then divide the remainder by the fixed base current to get real gain. But they don't say that they do, and so it's probably safer to assume they don't, unless you independently measure some test transistor, run it through their tests and figure out backwards from those measurements whether the Atlas does or does not do the right computations.  Until you know they do the right computations -> there is no possible way to say that any formula using their measurements is correct or not <-.

Like many things in life, getting the right answer is A Simple Matter Of Programming (ASMOP). 

But is the gain figure calculating in the leakage?

Near as I can tell, they don't say. It's possible to sub in a device that you have previously tested and figure out what they do, but the user guide doesn't say.

I thought about programming up a whizzo microcontroller which would do all this automagically, but didn't because (a) it's a PITA to do the programming although not at all mysterious, (b) there is a very small market for it, and (c ) seeing this set of questions on a commercial device that does some of the same thing reaffirms my belief that magical testers and calculators that simply tell you the answer inhibit your ability to learn what's going on, perhaps permanently.

Frankly, figuring out how to get an Atlas to tell you the right answer is as much work as doing the measurements to get the right answer in the first place.

[PRR]

> a whizzo microcontroller which would do all this

Hmmph. Couple switches, meter, and slide-rule does the deed.



Bigger page-scans:
http://i.imgur.com/hLS6F.gif
http://i.imgur.com/RbwlH.gif
http://i.imgur.com/sQUbC.gif

> open the base, read the current leaking OUT of the base with the emitter open .... I've never seen a tester that would do that, even though it's possible.

Position 4, read meter directly.

If you know the book, fig 15.25 is even simpler and will tell both leakage and increase-of-current-above-leakage, which sorts duds from workables.

> inject an external current into the base, measure the collector current, and then divide the collector current by the injected base current, and call that the value of HFE.

Another way, not the same but very applicable: Let it leak, or bias it, then inject a microAmp of AC into the base, read the AC in the collector. "AC" can be the wall-power, filament transformer. 6.3VAC into 6Meg (or divide 5K:1K then into 1Meg) is near-enuff 1uA. Put 1K in the collector, if you see 0.1VAC there then h_fe is 100. Modern digi-meters make it trivial to read 200mV AC.

On the average workbench, that may want to be shielded. Alternatively there's a plan in the book for any half-awake transistor plus something like Radio Shed's 1KCT:8 transformer to whine medium-pitch, so a coarse bandpass (bass-cut and radio-cut) would read only the test tone.

[darron]

good question....

i've always assumed that it does not factor leakage. i've noticed that the trend is for leakage to almost always measure proportionate with repoted hfe.


for example if i get a tranny that reads hfe as 190 the leakage might be 0.4mA. if something reads zero leakage gain may be around 80.



VERY handy tool....

[John Lyons]

I asked Stuart Castledine and he said that the leakage is figured into the gain reading.

[mac]

I also would like to use my Atlas DCA to measure gain and leakage for germaniums...

Does anyone have the formula I need to calculate "real gain" from the Hfe and leakage mA readings I get from the Atlas?

RG's setup has worked for me in the past, but I also would like a better understanding of the situation by doing the math with these readings.

I really appreciate it!

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

In my gallery you can find a HTML page you can use to automate RG calculations.

mac

[R.G.]

I asked Stuart Castledine and he said that the leakage is figured into the gain reading.

Next question: Does Stuart know, and if so how?


My Golden Retreiver "Ducks" thinks it does too. 

The Atlas could do this, and it would be nice if it did, but the actual details matter.

Hmmph. Couple switches, meter, and slide-rule does the deed.

Of course that's possible. The problem I find here is that most of our Gentle Readers are math averse. They'll go to almost any amount of time flogging a search engine looking for the answer on the internet to avoid a paper and pencil, pocket calculator, or God forbid, a slide rule. I strongly suspect that only a tiny percentage of the GRs here have ever seen a slide rule, much less know how to do simple math on one. I keep a K&E Log-Log Duplex in the desk drawer just for sentimental reasons.

It's not just that math is hard. I think the real problem is fear of understanding. To actually bang the calculations, you have to understand what numbers to add, subtract, multiply and divide. So the prospect of a little box that not only tells you what this device is, but tells you all you want to know about it, even odd, obscure stuff, it is almost irresistable.

[darron]

if it is calculated in, then my question is does the likely hood of higher leakage go up with a higher (true) HFE? because that's always what i see with the PEAK/Atlas?

[PRR]

> does the likely hood of higher leakage go up with a higher (true) HFE?

WHICH "leakage"? Ico? Iceo? Ices?

Ico does not depend on hFE.

Iceo is Ico times hFE.

WHICH is the "leakage" we want to know?

Both. More or less. Depends on the circuit.

> It's not just that math is hard. I think the real problem is fear of understanding.

Oh, fah. Banging the calculations is less work than chipping, slicing, and throwing snow. Which many of us are a little tired of. Curl up beside the fire with a transistor theory manual.

I suggest:
Transistor Circuits and Applications
Lawrence G. Cowles
ABE.com has 20 copies under $10 delivered.

Also
GE Transistor Manual
General Electric Company, Cleary, J. F. technical editor
Several offers under $15.

[petemoore]

  I got by without it, but RG's tester's the way to go.
  I had pretty good batch of Ge's, some were SB tested, others had been salvaged from old stuff.
  It's pretty easy with a tweekable FF circuit [adjustable Q2C bias resistor needed] and a DMM, and a handful of Ge's to tell which ones leak.
  The DMM'll say X gain + leakage, when compared to something similar that doesnt leak so much the noise, and "Gain doesn't add up and = what the DMM misread"...they get sorted out anyway.
  If not in the know about GE's with all unknown Ge's...tester is pretty much a must to determine if any are suitable candidates/to hear what a 'good set' sounds like and use as comparator.

[R.G.]

Oh, fah. Banging the calculations is less work than chipping, slicing, and throwing snow. Which many of us are a little tired of. Curl up beside the fire with a transistor theory manual.

Yeah. Where is Al Gore when we need him?