Testing transistors

[jaytee]

I've been testing some germanium transistors for leakage and hfe using the resistors specified on geofex. I noticed he is using a 2m2 resistor to supply base current. In the article I read at Smallbear he is using a 1m resistor. Thing is, hfe is not constant and depends on collector current. It rises with collector current to a peak then drops off again. Usually hfe specs are given along with the collector current.
I tried this out with an oc44 and got gains of 71.8 with 2m2 and 79.5 with 1m. Wouldnt it be better to test at a constant collector current and measure the base current?

[R.G.]

You can certainly get there by holding collector current constant and measuring Ib to make it be so. An opamp servo like I used for the JFET matcher would do that in a jiff.

But there are a lot of considerations in coming up with a canned measurement process for beginners.

First, there is the concept of hfe itself. As you note, hfe is not constant with collector current. Neither is it constant with temperature. If I remember right, there's a slight change with Vce. And it's by no means constant with frequency.

The idea of measuring the result of many independent variables with great precision is itself flawed unless we simultaneously measure all of the independent variables and specify the result. Then, it only applies to that specific set of conditions.

It's better to look at it as these tests producing just an estimate of hfe, not a number which is exactly hfe. And frankly, for most effects use, an estimate is just fine.

Then there's what to do with the data. I picked my resistor values to make it easy to compute the result for a beginner. For a measurement where Ib varies by 2:1, I'm kinda happy that the measured hfe varied only a little less than 7%. I can make it do worse than that by leaving my finger on the transistor being measured.

As a last comment, in testing germanium, you want to separate leakage from gain. Leakage current is what happens when base current is zero. You certainly could measure leakage by measuring collector current, then flip over and measure base current at a specified collector current, but then you have to decide - at a fixed collector current, or a fixed current plus leakage? Because the next thing you're going to do is to try to seperate out the real gain from the leakage. In addition, most cheap DMMs are not all that accurate down in the microamps of base current. Measuring collector current lets you use the instruments where they're more accurate.

For effects purposes, the exact number is hardly ever necessary, and measurement to three decimal places implies much more information about the device than we actually have.

[jaytee]

I get what you're saying, it's not so straightforward and great accuracy is not needed. I don't usualy record gains using decimal places. However I think the difference in gain I got with two methods is significant. I'm guessing the leakage would have to be  taken into account so if the collector current was 1ma then it would be 1ma+leakage. If it's not too complicated I wouldn't mind setting something up to do this. It would give me more of a true comparison. Another thing is if I decide to buy pre-tested transistors I don't really know what I'm getting unless I'm told how they are tested.