How to spot a bad transistor...

[flintstoned]

The bad tube thread was pretty interesting and got me thinking about a transistor thread. So I was wondering how you experts determine bad transistors while they're in circuits, are there any tricks to use multimeters to find issues?

[edvard]

I can usually spot bad transistors by the skull tattoos and dragon bandanas.
If they're smoking, you can bet it ain't tobacco...









(sorry, couldn't resist...)

[jrod]

Here is an article by R.G Keen: http://www.geofex.com/fxdebug/trantest.htm

I am not aware of a way you can look at a transistor in a circuit and tell it's bad.

[Paul Marossy]

Here is an article by R.G Keen: http://www.geofex.com/fxdebug/trantest.htm

I am not aware of a way you can look at a transistor in a circuit and tell it's bad.

That's a good page. Here is another couple of pages to look at:

http://www.reprise.com/host/tektronix/workbench/transistor.asp
http://www.reprise.com/host/tektronix/workbench/transistor_test.asp (out of circuit testing)

[LucifersTrip]

I can usually spot bad transistors by the skull tattoos and dragon bandanas.
If they're smoking, you can bet it ain't tobacco...

Exactly...always easy to spot a bad one...

[Scruffie]

I use the transistor tester at the bottom of this page, works pretty well for sorting through transistors quickly -
http://www.aronnelson.com/gallery/main.php/v/Richard-Boop-RLBJR65/album81/test.gif.html?g2_imageViewsIndex=1

[R.G.]

The bad tube thread was pretty interesting and got me thinking about a transistor thread. So I was wondering how you experts determine bad transistors while they're in circuits, are there any tricks to use multimeters to find issues?

It's hard to do a quantitative test (i.e. "gain = 124, leakage = 106nA) in circuit. But you can do some qualitative tests. That is in fact, part of the basis of "What to do when it doesn't work".  I suggest you read "What to do when it doesn't work" with an eye to what it tells you about operating voltages, and think about why this is such a remarkably useful set of tests. I don't even count any more the number of times we've debugged a bum effect by simply looking at the voltages from this test. It's not perfect, by any means, but it's really, really good for sorting out the "nothing works" cases.

In a small signal NPN working as an amplifier and not as an on-off switch, the emitter will always be 0.4 to 0.7V below the base, and the collector will be at a higher voltage than the base. When the collector gets lower than the base, the device is saturated and can't be amplifying. How much above the base the collector stays is dependent on the circuit, but the Collector -> Base -> Emitter voltage cascade must be true, or it's not amplifying.

If the base and emitter voltages are the same, the device may be internally or externally (i.e. bad soldering) shorted base to emitter, or the base may be open, no drive applied. If the collector and emitter are the same voltage, either the collector load is open, or the device is internally or externally shorted.

The same is true of PNPs, but in the opposite direction: Ve > Vb by 0.4 to 0.7V, Vb > Vc by some amount. For both NPN and PNP, the "some amount" the collector is away from the base is most often a significant fraction of the power supply, 1/3 or more. There exist special cases where this is not true, but mostly.

Sometimes the resistors on the base are so large that you can't directly measure Vb and Ve then subtract and get sane measurements. In that case, measure with one probe on the base and one on the emitter directly to get Vbe.

[LucifersTrip]

I use the transistor tester at the bottom of this page, works pretty well for sorting through transistors quickly -
http://www.aronnelson.com/gallery/main.php/v/Richard-Boop-RLBJR65/album81/test.gif.html?g2_imageViewsIndex=1

Do you know what criteria is used to determine good or bad?

[R.G.]

Just like we used to use to determine that computer systems were passed final test and ready to packge and ship:

--- it has the correct weight ---

[Scruffie]

  Yeah... not a clue how it works... it just does, it's not 100% efficient i'll give it that but for a quick flick of an on switch and poking it in (Oh the inuendo  ), it's fairly reliable, Strong Red or Green light for working, no light or both lights (some times faintly) for dead/ crappy, i've not even checked the CD4001 datasheet to be honest, I built it when I was new to electronics and tested it with known bad transistors against good to see if they worked and it's just been one of those things that stuck around cause it served a decent purpose.

[petemoore]

  I've done pretty good spotting good ones, the bad ones get a leg bent over the head and then tossed. The good ones always seem to come out of a bag, but I've gotten good ones from many varied sources.
  For NPN, testing is easy, have a circuit set up with a socket for the NPN, try it with a good transistor, try the suspect transistor/compare.
  Alternately the DMM will tell if there is good liklihood of working transistor, every times it's produced a 'likely' Hfe reading they worked well [not so true for Ge, leakage isn't measured and it matters to performance and throws off the Hfe reading.
  All the ones out of the bag worked except the SMT's that accidentally arrived.
  Some types are easier to get to work good than others.
  Hand picking the Jfet for a 9v booster [see Fetzer article and testjig] may find more suitable Jfet.