Transistor testing methods - anyone compared?

[CodeMonk]

I'm wondering if anyone has done any comparisons on transistor testing methods and compared them to each other.
Like for hFE, leakage etc.

I'm wondering what, if any different results you may have gotten when testing the same transistors
when using RG's method ( http://www.geofex.com/Article_Folders/fuzzface/fftech.htm#Picking transistors ) versus using something else (in my case a Peak DCA 55).

Maybe difference in terms of a percentage or something?

Anyone done any comparison testing?

[R.G.]

This is made harder by the fact that any one transistor has gain, leakage, etc that varies with conditions, including how you test it. The problem with testing transistors with different testing methods is how hard it is to get any consistent results at all. To get consistency, you have to go redesign the testers so the test conditions the transistor sees is identical.

And in fact, as I've said before, no transistor can boil down to a single number or set of numbers when it varies with time, temperature, current, phase of the moon. What a transistor test is is an indication, perhaps relative to other transistors you're testing at that moment on that same tester.

[LucifersTrip]

I did a quick test when I bought the 55 a while ago...

http://www.diystompboxes.com/smfforum/index.php?topic=91115.msg776253#msg776253

[sshrugg]

I can vouch for what RG is saying in his post above.  He's always right, but this time I UNDERSTAND!  When I used his testing methods building my fuzz face, I had to switch from handling the transistors directly to using calipers, because my own body heat from the tips of my fingers was causing the readings to change as the transistor cooled down.  I decided rather than worry about what temperature the transistor is at, I would avoid handling them.

Transistor checking (especially with old leaky germanium transistors) is an approximate measurement.  I pick the most ideal candidates from the batch I ordered, but I don't know to this day what the exact reading would be.  Likely it is not even the same as when I built that pedal two years ago.

[CodeMonk]

Thanks for the info RG and everyone else.

I was mainly curious when testing, if the environmental conditions are the same, if there is some way that the differences between the different methods are consistent.
Regardless of what methods were used as long as the same methods were used and environmental conditions were the same.

For instance, if you tested a transistor using RG's method and got a leakage of 300uA, then used the Peak and got leakage of 330uA
In that case, the Peak got a leakage value that was 10% higher than with RG's method.
Then you tested another transistor and got 100uA leakage with RG's method and 110uA with the Peak.
That's still 10%

I know temp during testing can also make a huge difference.
Hell, in the last few weeks, I tested one transistor and got a leakage value of around 660uA when my workshop was about 100F
The following week, when it was raining, and the temp was about 80F, I got a leakage value of 380uA.
With the same transistor.
Both using the Peak.
Due to the temperature differences, I expected a difference.


That's what I was asking about.
If you can get that kind of difference consistently during tests under the same conditions.

But I guess I'll have to sacrifice a goat on the 3rd Tuesday of the month during a Blood Moon...again.

Thanks again.

[electrip]

This is made harder by the fact that any one transistor has gain, leakage, etc that varies with conditions, including how you test it. The problem with testing transistors with different testing methods is how hard it is to get any consistent results at all. To get consistency, you have to go redesign the testers so the test conditions the transistor sees is identical.

And in fact, as I've said before, no transistor can boil down to a single number or set of numbers when it varies with time, temperature, current, phase of the moon. What a transistor test is is an indication, perhaps relative to other transistors you're testing at that moment on that same tester.

I did (compare).
I own a Peak Atlas DCA55, a DCA75pro and a MK-168 (new cheap chinese one from ebay).
Both PEAK Atlas testers measure Hfe with an quite high collector current for our puposes.

DCA55 (FW: R2.
HFE @ Ic= 2.5x mA
Vbe @ Ib=5.xx ma
Leakage @ unknown voltage (have to look up)

DCA75 (FW: 0018)
HFE @ Ic= 5.0x mA
Vbe @ Ib=5.0x ma
Leakage @ unknown voltage (have to look up)

Both testers use a fairly fixed test point and if temperature is constant
and test probes have a good contact to the transistor leads (corrosion is an issue!)
test results have only small fluctuations. No moon phase :-)

Constant collector current means different Ib and a different resulting Vc for HFE measurement.
Useful collector currents for Fuzz Face Q2 would be around 0.5 mA (@ ca 5V Vc)

Example AC153 V Siemens:
Tester DCA75pro
Test Result:
PNP Germanium BJT
Red-C Green-B Blue-E
HFE=73 at Ic=5,01mA
Vbe=0,246V at Ib=5,00mA
IcLeak=0,151mA   

from the curve tracing data (Vce= 5V):
HFE=43 at Ic=0,2 mA
HFE=48 at Ic=0,5 mA
HFE=55 at Ic=1.0 mA
HFE=62 at Ic=2.0 mA
HFE=71 at Ic=4.0 mA
HFE=81 at Ic=7.2 mA

This is an extreme example but it shows the  problems.

Vbe is measured at a very high current (sometimes not useful).
Leakage is relative independent from voltages above Uce > 1-2 V
For the germans here (idomatic): Wer misst misst Mist. ;-)
("Who measures measures rubbish!")

Both the R.G.Keen-Method and the chinese tester are using a constant Ib for measuring HFE
resulting in HFE-dependent Ic and Uce (R.G.: Ub= 9V Rc = 2.472k Rb= 2.2M / MK-168: Ub=5V Rc/Re= 700 Ohm Rb=470k)
At moment I cannot comment on the accuracy of the MK-168.

R.G.s method is as accurate as resistor values and voltmeter.
And it is the fastest method for no/low leakage transistors
(beside a multimeter if you get the transistor leads in the right positons fast enough)
One could find the R.G. test point within the curvetracer data of the DCA75pro.

I don't know a tester which could test Hfe and leakage in one second and at once.

electrip

[Atodovax]

Sorry to bother but do you think that to get the "usual" recomended gains for FF should i curve trace both transistors with VCE 5v and VCE 1.2v and try different currents for IB. Or should i just sorted all my transistors using 9v VCE and 9ua on Base? Im really getting crazy with all this parameters. Because if i get a a transistor that has 70 HFE at 9ua Ib and 9v VCe it will clearly show less HFE in Q1 position conditions of a Fuzzface. Should we aim for 70/120 hfe´s considering both conditions or just meassure them all at smallbears conditions?
Also i would like to know if at the curve trace graph is leakage considered or if i have to substract that from the gains showed there. Sorry for so many questions!

[MaxPower]

I guess you need to ask them how they measured their transistors. Just use one method to measure your transistors and then try them out. Play it by ear (ha!), once you find the transistors you think sound the best make a note of the hfes/leakage measurements and which method you used to take those measurements for future reference.

I tried R.G.'s method and Steve's(?) method to see if my meter was accurate. My meter's measurements tend to agree with the measurements using R.G.'s method up to about 500 hfe and then they start to drift apart. Since I'm concerned with general low, mid, high hfes for the most part, I'm satisfied with my meter's accuracy.

[Rob Strand]

I guess you need to ask them how they measured their transistors. Just use one method to measure your transistors and then try them out. Play it by ear (ha!), once you find the transistors you think sound the best make a note of the hfes/leakage measurements and which method you used to take those measurements for future reference.

One thing that can change the tone is the transistor's collector-base capacitance.  It can be significant for high gain circuits.  The old GE transistors had quite large capacitances so it is very likely to affect the tone.  It is also quite variable (perhaps 1/2 to 2 times nominal).

If you look at some boutique fuzz faces they add capacitances across the base and collector.