hFE Madness!

[karbomusic]

I measured a few 2N3904s since I knew their leakage should be insignificant. I used the RG method, A crappy Pyle meter with an hFE slot and the Chinese ebay deal found here based on the "totally useful and cheap" thread:

http://www.ebay.com/itm/Mega328ESR-Meter-LCR-led-Transistor-Tester-Diode-Triode-Capacitance-MOS-PNP-NPN-/201087634115

Here are the results:

Keen Method: 280
Pyle Meter: 320
Chinese Tester: 438

Should I trust the Keen method above all others, is this measurement so difficult that no two methods can even come close?

Arggggggggggggggggggg!

[Govmnt_Lacky]

The first question that comes to mind is...

Why are you measuring the hFe on 2N3904s? 

[PRR]

hFE varies with current. What current is used in each case?

[karbomusic]

The first question that comes to mind is...

Why are you measuring the hFe on 2N3904s?  

Is measuring something with no leakage the same as not having to worry about leakage in the calculation? I'd simply like to see any two hFE meters and/or measuring methods spit out anywhere close to equal results like a 10% variance not 200%. Holding it to my head and making up a gain figure seems just as accurate

[karbomusic]

hFE varies with current. What current is used in each case?

That's a good question...

Keen method: per his method.
Chinese thingy: don't know
Pyle meter: don't know

For the last two above, the idea was with a transistor that isn't assumed to have much leakage seems like they should be close. I understand it depends on current but those are the meters, they are supposed to do that, that's what they do measure. (assuming removing leakage from the picture with 3904s).

[armdnrdy]

If you want to use the "Chinese thingy" to measure HFE...you could always check different transistors using R.G.s tester and the Mega328 tester to see if there is a consistent percentage in the different readings....and then calculate the correct HFE.

[karbomusic]

If you want to use the "Chinese thingy" to measure HFE...you could always check different transistors using R.G.s tester and the Mega328 tester to see if there is a consistent percentage in the different readings....and then calculate the correct HFE.

If I understand what you are saying, I tried that but I don't remember the discrepancy being linear; hence my frustration. I'm building a couple FFs and thought I'd do this check from an academic standpoint while I was "tooled up".  From an aggravation standpoint it "feels" like everyone is selling clocks, yet none of them keep time and all of them keep time differently. How might I ever know or trust what time it is? RG's method seems the most obvious but seems like a fairly simple measurement assuming leakage isn't in the equation speaking academically again.

It doesn't need to be the Chinese thingy, any two that are consistently close would make it feel like less of a crap shoot regardless of method. Just gimme two LOL.   I must conclude for now, that RG's method is the only one that actually has any accuracy at all.

[PRR]

R.G.'s tester forces "a touch more than 4 microamps of base current to flow in the base." We usually reference hFE against Collector (or Emitter, same-thing) current, which is hFE higher. If your transistor's hFE is 200, then Ic is 800 microamps or 0.8 milliAmps. If 100 or 800 then 0.4mA or 3.2mA.

This is designed for *matching* at "reasonable FuzzFace current", NOT precision known-current testing, with only slight math by the user. Also, of course, for leakage, which it will show super well.

The German/Chinese "cheap" tester uses a process that only a CPU could love. It puts 470K in the base, 700 (680+19) in the emitter, reads voltages, and does long division. Again it does not test at a "known" current. For a part with hFE of 200, the test current is about (argh! argh! brain-pain!) 1.5mA. (Note that this is somewhat higher than the RG tester.)

The PEAK has the test current in its sales/data-sheet. Considering the nature of your excellent exploration, I won't spoil your learning experience by telling you to look in footnote 2, or that the number is considerable higher than the others.

You can find the "typical" hFE versus Ic for 2N3904 on the data-sheet. I'm a bit puzzled because the hFE curve is quite flat from 0.1mA to 10mA.

Also note that hFE varies with temperature. Usually hotter is higher. If you throw it in boiling water it won't quite double. If one test were in summer under a 150 Watt lamp, and the other in winter in the garden-shed, significant difference is expected. If we assume 0.5% per deg C, then variations in "normal room temperature" amount to a couple percent, which is not your observed discrepancy.

[PRR]

Doh! DUH!!

hFE varies with collector voltage!! "Early Effect".

The 2N3904 specsheet test condition is 5V.

R.G.'s tester, for hFE about 200, tests near 7V.

The German/Chinese "cheep" tester gives around 3.3V for hFE=200.

It's late, I leave it to you to find Peak's test voltage.

The difference 7V or 3.3V could be quite significant. You can estimate it from the slopes of the curves on the 2N3904 datasheet.

[Buzz]

I'm hitting the same brick wall as Karbo right now. I feel ya pain mate.

Does anyone know if, on the 'Chinese' testers, the battery voltage affects the readout for hfe?

I'm thinking of abandoning the Chino tester for hfe. I think I might etch myself a board of RG's circuit and box it up. I have, for the sake of experimentation, measured pre tested Ge trans from other suppliers and come up with very close values as what is claimed by the seller. Damn close considering the environmental factors at play.

So my theory is this. If I can choose Ge Qs by hfe on RG's tester and make great sounding pedals it should work just as well to pick out Si Qs by hfe. Polarity swapping and no need for leakage testing taken into consideration.

Hmmmmm....

[karbomusic]

. If I can choose Ge Qs by hfe on RG's tester and make great sounding pedals it should work just as well to pick out Si Qs by hfe.

I have a number of low/med gain Si's for that very reason assuming the run of the mill Ge fuzz face recommendation: transistor gains <200 with Q1 < Q2.

That's not a huge gain window and some of the measurements I'm using are off enough that I could be way outside that window and not know it. And.... I have lots of old transistors, seems like I'd be interested in what their gain is regardless of how I happen to use them.

And of course all the Ge's I have (PNP and NPN Ge), it would be nice to know gain and leakage. I can certainly build a good sounding fuzz face by trial and error but that doesn't really change what I'd like to know.