Testing Hitachi Mosfets

[Zero Cool]

I have a number of Hitachi 2SK135/J50 mosfets pulled from various amplifiers. many of which i beleive to be bad. Is there a relaible way of testing them?

Bipolars, from what i have seen, generally short, or open and are pretty easy to spot. but none of these mosfets exhibit such activity, all have varying resistance from terminal to terminal and none are even close to a known good pair. But, what is acceptable?

How should i go about testing these? I hate to toss a perfectly good and hard to find mosfet.


Zero

[johnnyx]

I use one of these;-Component analyser

Well worth the cost, if you test a lot of components, or want to match them.

[Zero Cool]

Very cool little device and reasonably priced! now if i can find out where there sell them in the US!!!



Zero

[peranders]

It's pretty easy to make a first small test:

12 V stabilized power supply with 200 mA current limitation.

Apply, 12 between source and drain.

Apply from battery or via a second power supply, via a pot 0-6 volt between gate and source (don't forget to have 1-5 kohms in series with the gate and pot, avoiding oscillations). At 1-2.5 volts you will get 200 mA in drain current.

BTW: I have a box of those transistors....

[EUVL]

> I use one of these;-Component analyser

The problem with such component analyser is that they measure Vgs, etc with a very small current.

My experience with power mosfets (Hitachi, Toshiba and IR) is that they vary not only in Vgs (at a given bias) but also transconductance. So if you are really serious about matching, especially for power amps, you should really match under working conditions -- and that means Iq, Vds, and case temperature as in the actual application itself. IR probably has the least variations within one batch compared to the Japanese.

As I said, just my own experience.


Patrick

[MikeB]

It should be easy to sort out the bad ones with a normal multimeter,
measure "ohms" between gate to Source/drain, this must be infinite,
as gate is a pure capacitive input. Also only from source to drain
you should measure something. (in one direction)

This of course can only sort out the burned ones...

Mike

[Zero Cool]

I need a reliable way of testing the bad ones for the useable ones! a pass/fail type system.

matching i found an article on the Pass Labs DIY sight about how to match Mosfets that is very good.

Right now i just need to sort out the ones that are dead and any that can be re-used even if just for testing, or other uses then audio.

It has been a while since i put a meter to them but if i remember correctly i got around 5M ohms from any pin to any pin on a known good mosfet but the MF's in question measure anywhere from the megaohm range down to 200-300 ohms pin to pin. but none are dead shorted like a bipolar would.

I wasnt sure what i was supposed to read from pin to pin and what is an acceptable tolarance.


Zero

[Zero Cool]

BTW, the Mosfets i have that i was using that were known good, were a pair i pulled from a working amp, but they were not new so i wasnt sure just how accurate what i measured was?

The amp i was repairing, had one channel that worked and one that would blow the power supply fuses. I pulled the 3 pairs of mosfets from the bad channel, but didnt understand how to tell a bad Mosfet from a good mosfet, all of the mosfets measured differently. so i pulled the 3 pair from the supposedly good channel and all of those measured differently as well. none that were even close to each other! so thats why i started questioning if any of them were still good and how to tell?


The amp had a rough life living in a DJ rack and was quite abused when i got it.


Zero

[MikeB]

200-300ohms from any pin to pin ? That sounds like burned...
As i said, you shouldn't have any flow through the gate, but 5Mohms
is nearly infinite resistance.
I measured irfs this way, means verticalfets. Source to drain acts
like a diode, and gate is "isolated" from the rest.

Mike

[anatech]

A charge on the gate may cause the drain - source to measure resistively. Hook the fets up per Nelson Pass's method and check them. The bad ones will show up for sure, and you will want to test the good ones anyway.
Peranders has a very good simple suggestion too. That will get you an easy go / no-go test.
-Chris