DMMs and hFE and transistors. WHat is the reliable number?

[Thecomedian]

I can plug a Q into my new DMM and it provides me with accurate looking hFEs, and the fun part has been watching the value skyrocket when I keep my finger heat pressed against it, but what's the correct way to gauge the FE? IF you dont touch it, it will usually have a counted FE that starts decreasing over time, and, if I wait long enough (5 minutes), hFE will eventually equalize and stop shifting.

Is that the "accurate" reading, or is there just some general rule of thumb to apply about how fast the change in the reading drops or rises, and how much gain the Tx has?

[greaser_au]

I assume that these were germanium transistors?   I'd refer you to R.G. Keen's site* geofex.com and in particular the article on the fuzzface (Technology of the Fuzz Face).  The last few lines  of the 3rd last paragraph of  'picking transistors' answers your question most concisely!

The thermal aspects of semiconductors has been a bugbear for solid-state designers since the beginning. In the early days when hFE's were low, it was a bit more difficult to design a useful circuit with adequate thermal compensation - especially in power stages.  A device would warm up due to passing current, the warming up would increase it's gain and leakage, which meant that the device would pass more current, which would warm it up still further. Hopefully the stage design was such that at some point the operating conditions would settle, but if not, the current (and temperature) would increase to saturation or possibly even destruction. it's called 'thermal runaway'.

Silicon does it too, but it's much less severe. As an example the ETI-480 amplifier from the '70s used a BD139/140 driver pair and a 2n3055/2955 output pair. The 'Vbe multiplier',  a TO-92 BC549  was mounted in a hole in the heatsink with thermal grease, to keep the output stage bias point sane.

david

*many thanks for a very informative site, R.G.!

[LucifersTrip]

if I wait long enough (5 minutes), hFE will eventually equalize and stop shifting.
Is that the "accurate" reading

yes, if your dmm accounts for leakage. if not, use RG test:
http://www.geofex.com/Article_Folders/ffselect.htm

actually, even if your dmm does account for leakage (Peak, for ex), it's still cool to use the above test to compare.

btw, different Q's take longer to settle, of course...

[Kesh]

whatever temp you intend to be using the pedal at is the one to test it at.

DMM tests are kind of rubbish though, not only because of leakage which they won't detect (or rather they will, but think it's hfe), but also because there's no accepted standard of what current or voltage you test at, and some fix base-emitter current, others collector-emitter current.

[brett]

Hi
just 2c worth...
In most (most, but not all) designs hFE isn't an important parameter. In most designs (esp. common emitter), there is some negative feedback offered by an emitter resistor, and the gain is determined by the ratio Rcollector/Remitter. In circuits like the BMP, there is collector feedback, and the BJT gain stages operate in a similar way to an inverting op-amp (gain = Rfeedback/Rin).

If there's no emitter resistor, then you might want to check hFE. Especially for low-gain Germanium transistor (hFE=50 to 150). If it's a modern Si device and circuit and hFE > 300, then it's very likely that the circuit doesn't run it "flat out".
cheers

[R.G.]

Yep. In general hfe is NOT to be relied on. It varies. A lot.

Hfe is a useful SORTING characteristic to separate non-useful from useful parts, especially for germanium.

I was taught to design for a3:1 and be prepared for a 10:1 variation in transistor gain. The circuits are generally designed so if you have enough gain, you're OK.

Germanium comes from a time when that was not necessarily true.

[PRR]

hFE should not "skyrocket" from finger heat. Even if you are burning-up with fever.

This suggests (as said) you are testing Ge parts on a system which does not account leakage.

(hFE _does_ generally rise with temp, but only a few percent from room to finger temp.)
__________________

> I was taught to design for a 3:1 and be prepared for a 10:1 variation

Under a different prof, I was taught to assume only that hFE was >50 (10 for power devices). Yes, to infinity! Or to avoid awkward calcs, say 1,000 (sorta high then, not absurd today). If your circuit plan burned-up (or just slammed) with very-high hFE, that was 2 points off the quiz.

While you can get parts with 3:1 spread, it's always been cheeper to take "unsorted" parts with 10:1 spread.

At one time, it was cheapest to design for a MAX hFE. Parts over hFE=20 went into computers. Parts under hFE=20 were not worth using in computers, or much else, and were dumped for just a few bucks each (when a "good" transistor was $20).

Since Si released us from leakage worries, and made hFE>50 pretty cheap, it has been quite easy to design for hFE=50-1000. You might save one part in a multi-stage amp if you spec hFE>250. Hi-hFE has a broader sweet-zone for Noise Impedance. But any design that WON'T work hFE>50 is just obsolete.

Yes, the obsolete stuff is sometimes more interesting.

[pinkjimiphoton]

i may be totally off base here, but i usually take transistors and plug 'em in my meter to read the hfe, and then reverse them. transistors that show a reading backwards always seem to be problematic if silicon.

i figure if i see more than about 30% backwards of whatever they read when inserted "right" are probably crappy. the lower the second the number, the better they seem to sound.

does that mean my cheap harbor freight meter is testing for leakage, or is it just an accident and i've been lucky?

[PRR]

Reverse hFE actually means other things. I don't think it is a good figure of merit (ie, you been lucky).

[R.G.]

Yep.
(@ other than PRR)
If you look at it this way, an NPN has one of the Ns as collector and the other N as emitter. In the original transistors, there was little difference (intentional, anyways) between the two junctions between an N and the middle P-type section, and the transistor worked equally well (or poorly, depending on how you look at it) either way round. In effect, it had the same general gain no matter which N region was the collector or emitter.

But semiconductor researchers found out that the collector-base junction does different things than the  base-emitter, and you could get better gain, frequency response, voltage, and current characteristics by specializing the two junctions. So making the collector-base be higher voltage and more lightly doped than the base-emitter made a higher gain, faster, higher voltage device, and transistors became specialized.

That didn't wipe out the "reverse beta" where the emitter is more positive than the collector. It just made it much smaller than the forward beta, which is what we would expect. And the reverse voltage ability of the base-emitter became much lower, generally about 5-8V  for most modern silicon devices.

If you just reverse the emitter and collector leads of a modern NPN in a circuit which has a low enough voltage on the "collector" not to break it down, then the transistor still amplifies, but it may have a reverse gain of less than 10, compared to the forward gain of 100 or much more.

So in a way, you're right. The more unbalanced the gain reading you get on your test, the better the transistor is in the theoretical sense. However, it's not leakage that makes it that way. A higher unbalance usually means a higher forward gain, and that generally makes ordinary circuits work better.l

But it's not leakage.

[pinkjimiphoton]

ok, i actually understood that now, RG and Paul...
thanks!!!

seems the bad ge's always seem to be more prone to be close when reversed in my meter.

the best "sounding" ones may be a forward hfe of say 200 or so... turn it backwards, and i've found any reading over 10 or so just plain won't work right in most circuits...those are the ones that (like the nte's i bought when i started) would freak out if they got hot OR cold.

the only silicon ones i've seen read are just plain bad.

i guess i've been lucky, and now it's time to build the ge testing jig. (i just bought 40 nos ge pnp's off a surplus site i found for .35 a piece... the data sheets said average hFE 20-28 i think ... 2N2273)
i know they're probably too low for most uses, but i figure they may sound really good in some fuzzes.

[reveal]

Small Bear has an article on uses for low gain germanium.  https://www.smallbearelec.com/HowTos/GeDarlingtons/GeDarlingtonPairs.htm