Beginner testing JFETs.

[nuromantix]

I am testing a JFET whilst trying to find the problem with a guitar amp.
It's a 2SK117GR which is an N-channel JFET, and it's been removed from the board.

Connecting the ohmmeter to the 3 terminals in every combination, I got the expected results except for one:
-GATE +SOURCE - INFINITY
-SOURCE +GATE - 5.5 Mohms - I would've expected a result on the order of 50-80 Ohms
-GATE +DRAIN - 82 INFINITY
-DRAIN +GATE - 82 Ohms
-SOURCE +DRAIN - 72 Ohms
-DRAIN +SOURCE - 70 Ohms

Does this mean it's faulty or am I making a mistake to draw this conclusion?
I was really expecting to find either a working JFET or a leaking one.

THANKS!

[J M Fahey]

You are using the *ohms* scale which will only work on the DS junction, both ways, because it´s *resistive*
Indeed you found 70/72 ohm which sounds reasonable.

But both DG and DS junctions are DIODES and have to be tested on the DIODE scale.

A functioning FET will show OPEN 1 way and "1 diode drop" (about 700mV) the other way.

But I guess you are pulling all components, one by one, and testing whether they measure normal or not ... that is not the best way to find problems, by far.

You must *troubleshoot* it, which means checking what *works* or not, what controls affect what, then do some voltage measurements, and only IF troubleshooting leads you to a suspect component, you measure it.

trying to find the problem with a guitar amp

Ok, which amp?
What problem?

Pulling parts at random is the same as having a car problem and fully disassembling it, then measuring part by part checking tolerance, steel hardness, etc.
What if the car was out of gasoline, oil, water, or battery was discharged?

[nuromantix]

Thanks for the info, I will check again with diode mode.

BUT why is SG showing 5.5 Mohms?

Don't worry I'm not stripping off components one by one and testing them all! The amp is making the kind of "fireworks" sounds that I would associate with a failing transistor or FET, although I guess it could be a diode or cap gone bad too. It's a Roland JC-50 and I can hear that the noise is affected by the "bright" switch but not by the tone controls, which ought to narrow it down to a very small part of the circuit, that contains 3 of those FETs and one other transistor. So given the nature of the noise I thought I'd start with those.
It's a situation where the repair shop want to charge me more than the amp is worth so I thought I'd treat it as an opportunity to learn new skills.

[Mick Bailey]

Agreed - it's best to check the operation of a part in-circuit. The danger in using a DMM to check a part is that the meters don't test at the operational voltages within the amp. The diode test operates at maybe 2v but in the amp you'll be getting a much larger voltage to reverse-bias the GS junction. Even if the FET tested OK on the bench then you may be led to assume its OK and this gets set in your mind as not being the problem. To use an Enzo phrase, the bench test can tell you if a part is bad, but it won't tell you that it's good.

The other issue with testing FETs on the bench is with some devices the gate capacitance can hold a charge and give misleading results - even to the point of making the DS channel look open. More of a problem with MOSFETS, but not exclusively. A trick to overcome this is to stuff the legs into anti-static foam when testing, which has enough conductivity to bleed any charge.

EDIT: Ah, you posted while I was typing. You can't check a PN junction on the ohms scale - there's insufficient voltage to fully turn on the junction so when the junction is reverse-biased by the meter voltage you get infinite resistance. When you reverse the meter leads (in effect to attempt to forward-bias the junction) you may just be getting some nominal current flow that your meter reads as a high resistance. I would disregard any resistance measurements of the junctions - using the resistance rather than diode mode is like measuring weight using a thermometer.

[nuromantix]

Thanks, I'll check with diode mode tonight.

How can I check a FET in-circuit?
I would assume the effects of the rest of the circuit would make things pretty unpredictable.

I did short gate to source before testing out-of-circuit!

[J M Fahey]

Thanks, I'll check with diode mode tonight.

How can I check a FET in-circuit?
I would assume the effects of the rest of the circuit would make things pretty unpredictable.

On the contrary.
Fact is, in circuit in principle you are not testing it as a random part alone in free space but as an element in a system, in this case a circuit, where it should "do something".

Please post here the Roland JC50 schematic and indicate which one is this specific Fet, so I can show you what to do.

OTherwise explanation will be so generic that it will be unusable.

And besides you are searching/complaining about a "fireworks" (crackling/popping?) sound ... NO meter in the World is going to measure or show *that* .
does yours have such a scale or setting? ......
of course not, all it cann tell you whether it´s shorted or not.
What Enzo said and I repeat: a meter can tell you it´s "bad" (very bad as in shorted or open) , it can not tell you it's "good" (noiseless).

Just as a curiosity, not telling you you build it, (you´ll still need the specialized Lab equipment), here is how FET noise (of any kind) is measured:

this is measure "how much" noise, a single number printable on a datasheet:



this is to measure "where" is noise: low/mid/high frequency (burst/popcorn/hiss)



this is same but indicating some real Lab equipment which might be used:



and this is how it might look if you were there:



but DON´T WORRY, we won´t use any of the above setups, mainly because instead of measuring Fet by Fet to check whether it meets Manufacturer´s specs or not we´ll check for noise in the amplifier just by listening and then applying some testing logic to find *where* it comes from.

Once we found that, we´ll concentrate on the actual suspects

So please post that schematic.

[nuromantix]

The service manual is linked in this thread:
http://music-electronics-forum.com/t37939/

Because the noise is affected by the BRIGHT switch and the volume control, but not by the tone controls, I am assuming (to start with) that the problem lies between those two parts of the circuit, ie. Q5, Q7 or Q8.

I am assuming (as an initial try) that one of these transistors/FETs is open and therefore causing noise. Obviously I am not trying to test them for noise, just for a fault. If none of these make a difference I guess I will look at the diodes next, unless someone has a better suggestion.

Thanks again.

[Jazz P Bass]

I would start by limiting the tests to the Clean channel.

If the problem lies within the volume control circuit, then that limits the jfets to Q5.

The Clean signal goes from Q5 to C17, through R27 to the volume control.

The Bright circuit is C2 & R3 in parallel with the volume control.

I would start by giving the volume control a good cleaning (DeOxit)

Note: it is a possibility that Q8, which is the jfet switch for Dist, may be leaking signal from the Dist circuit.

[nuromantix]

Yes I agree, Q5 or Q8! Now I see that Q7 is going into the distortion section so probably not relevant.
It's not a dirty pot, it's a constant crackling that overwhelms any other audio.
When the problem started it was thermal, ie it took a minute to start happening after power up, but now it's all the time. Hence why I suspect it's a transistor leaking.
I will replace those 2 JFETs anyway, waiting for smoe NOS ones to come from Hong Kong, will probably be weeks!

[Mick Bailey]

Do you have a scope?

In the absence of a scope an audio probe is useful for locating noise in low-voltage circuits. Any component can cause noise. A thermal problem that gets worse (or permanent) can still sometimes be located with freeze spray, but given that you've already removed a transistor it's maybe worth waiting for the replacement to reduce any further stress on the pads if you were to find it's defective and have to remove it again.