What's this J-FET doing ?

Thread Starter

DarthVolta

Joined Jan 27, 2015
499
I'm new to JFET's and I think 1 is out of whack in my Keithley197. So I'm looking at some lecture notes today, but I still don't know if this means the JFET is gone bad. I think it's a N-ch Vishay 2N4117A
http://pdf.datasheetcatalog.com/datasheet/vishay/70239.pdf
https://www.utwente.nl/en/tnw/slt/documentation/Equipment/GeneralEquipment/keithley197.pdf

An op-amp drives the gate low with -8.9823V, across a 1Meg to the gate, but it's at -7.4885V, so that's 1.5uA exiting the gate, isn't that huge?
Also the 330k drain resistor has the 2.00184V Vref on top, and the drain is at 1.88762V, so that's 346nA, if it's actually entering the FET. The Source is at around 10mV on the buffer/amp, and it varies a lot, 10s-100s iirc of mV on Voltage ranges) and it's not 60Hz or harmonics, or it has to be subtracted out, right now this junk signal is there, nearly 1:1 matching the bad readings, and throws stuff of, . But it's the rouge voltage on that JFET Source (but 3 others too), that I think is the problem.

I haven't tried to calculate any JFET eqn's yet, but isn't that bad ? And it's supposed to be a high Imp. JFET, so really high.

It's Q113 on page 66, and on page 50 you can see the 4 JFETs of the sampling multiplexer, Q113 is to sample the 2Vref, and the other 4 JFETs are for the range dividers.

Here's Q113, w/ the Grn probe on Drain, blue on source, red on gate, yellow on gate drive, that was on 200V rng, in testmode with only the Q113 network running.

I'll have to check the othre JFets, what do u folks think ?

Ok in the last pic, it matches at any freq, as the gate voltage drops, the drain voltage increases, as the current into it from R111 drops. So indeed this JFET is a lot on-er than it should be.

I could have had current going into the shield pins, I guess there's some leakage and AC too, but I'm sure there's DC leakage/short problems
 

Attachments

Last edited:

KeithWalker

Joined Jul 10, 2017
1,142
You really have not defined a question. I read you post and have no idea what the J-FET is doing. You are the one with the circuit and measuring equipment. What is it doing?
 
Careful that multimeter loading can affect readings, say 10MEG input resistance with a 1MEG feed resistor gives -9%, so defective JFET "leakage" current may not be there. Scope input resistance and capacitance can also cause problems with stability. If you are seeing stuff at 200nsec it's likely the scopes SMPS or AM radio interference. I'm saying try not to go on a goose chase. The MCU seems to be constantly switching all the JFETS too. I would expect Q113 to just switch in the 2Vref but any JFET of the set stuck on would upset all.

Searching for bad switching JFETs I would look at both sides of the 1MEG gate resistors (R117) as there should be almost nothing for voltage drop- unless the JFET is biased way off and the gate diode is on. I think there should be -9V to shut them off.
 

Thread Starter

DarthVolta

Joined Jan 27, 2015
499
Yeah, Hi I have it on the test modes, so they aren't switching. So far it seems that the range JFET's all have an off state -8.8830V gate voltage , w/ -8.9828V drive, mostly across 1Meg, so those are really super off when needed.

3 of the 4 multiplexor JEFTs have -8.1658V, w/ -8.9828V drive from the LM339, across 1 Meg, thats still a lot but a matches

Where as Q113 is at -7.4955V, w/ -8.9828V drive from the LM339, across 1 Meg. It's gate current really does match it's Ids, I can zoom in/out anywhere on the wave form and they match.

I'm going to cut it's gate trace and see if the junk signal goes away. Hmm that turns on a JFET, so I removed it and I still have signal on the input divider, that either isn't supposed to be there, or isn't getting cancelled out properly (yes it's not working w/o that JFET). So back at it.


When 1st got this, it seemed to work fine, but I never really tested it, or compared it, I just started using it on another repair, and the numbers matched what I was used to from my other DMM, so I knew it was pretty close.

But now it can be off by 5%, 20%, across a whole range. So for an experienced pro working on this thing, they'd already have found/fixed it probably.

The mystery continues. I'm looking up Johnson–Nyquist noise too
 
Last edited:

Thread Starter

DarthVolta

Joined Jan 27, 2015
499
Yeah I want a 1Gigaohm I/P imp. meter

On any voltage range, and the I/P shorted, what voltage should be on pin 2 of the 10Meg R105A? I'm getting 100's of mV in the past, like 150mV after it settles overnight, and to an offset, it matches and moves with the LCD reading, and it's upstream/higher voltage, than U102-NI, .even in test modes when nothing's switching.

The offset between U102-NI and OP is about 3-5mV, anytime I checked on whatever setting it was

On Ohms, the % readings really drop off once I get over 1Meg, I've made some tables, but they're cluttered looking

If 100kHz was around the BW of the 10Meg in operation, that's a noise voltage of around 128uV, if I'm applying that right.

What voltage does anyone think should be on pin 2 of the 10Meg R105A with the I/P shorted. on volts ?
 
Last edited:

KeithWalker

Joined Jul 10, 2017
1,142
You don't appear to be using any logical approach to your trouble shooting. If you continue to poke around with your test probes in the very high impedance J-Fet circuits you are going to permanently damage them. I have worked in the electronic instrument service environment and have seen untrained technicians destroy thousands of dollars worth of professional instrumentation by using methods similar to yours.
The logical approach to trouble shooting is:
Define the problem up front.
Look at the block diagram and decide how to test the functionality of each section. Don't overlook any sections or presume that they could not contribute to the problem.
Start with the simplest section (e.g. the power supply) and with carefully planned measurements, test the functionality of that section. Don't make any more measurements than are absolutely necessary to test it. If it works, remove it from the list of possible causes and move on to the next section.
Using this technique, you will quickly narrow down the possible causes of the problem.
Regards,
Keith
 

Thread Starter

DarthVolta

Joined Jan 27, 2015
499
So don't probe J-Fets ???

The PSU works fine. I am keeping logs of sort, but yeah I'm jumping around a lot, because I'm also just learning how these types of things work in general.

And what do you think the voltage should be on pin 2 of the 10Meg R105A with the I/P shorted on volts ? I know what I expect is not there, and is way closerto what LTspice says w/ a simplified version.
 
Last edited:

KeithWalker

Joined Jul 10, 2017
1,142
I don't know what the voltage should be on pin 2 of the 10Meg R105A with the I/P shorted on volts because you have not given me any reason to take the time to consider it. The only thing I can get from your posts is that you multimeter (?) "can be off by 5%, 20%, across a whole range".
That is a really vague definition of a problem.
I very much doubt that knowing what that voltage should be would contribute to fixing your meter. No wonder you haven't repaired it yet.
You appear to be just wasting your own time and the time of anyone who is willing to try to help.
Try to define your problem concisely and let us know which sections of the instrument you have found to be working correctly so that we can possibly help you with your trouble shooting.
Regards,
Keith
 

Thread Starter

DarthVolta

Joined Jan 27, 2015
499
I never added my DMM to the voltage/current measurements. I really really underestimate the signal level, and how my working DMM, attached to such a circuit is effecting it. And yeah that's why to understand more how something works, then setup a condition, and observe.

I don't know much small signal analysis, and yet for some reason, it was not occurring to me, that THIS IS super small signal analysis. In the past when I used +1M resistors in say a 555 timer circuit, I also never even considered my DMM/scopes effect on it, and never looked close enough to notice the difference.

Well I got a decent sim running in LTspice with actual JFets, the buffer op-amp, the multiplexing, and op-amp that in my limited understanding, is sensing the voltage on the shielding, the noise, and adjusting the JFet's gate biasing a tiny amount to make it either cancel, or have both sides move in unison, I have not calculated yet and can't say off the top of my head, I'm at that level.

Now I have new clues after taking out the switches again and hard wiring the 20Vrng. It's still not right, I haven't checked much farther forward in the circuit past the front end tho, for the sake that it seems to be working so good in ways, I figured it's just all solder brideges in the SW's, and/or traces around them..........and I was hoping that they should be easy to pin point, like some are. But I hat already cleaned the SW's, scrapped around the traces, and still had problems, so I gave up and did the real heavy duty de-soldering, and took out the SW again. And have a different problem, to figure out.



But I don't want to find a simple solder bridge, I want to find something cooler (but cheap) , like at the end of a level in a videogame.
 

Attachments

Last edited:

shortbus

Joined Sep 30, 2009
7,841
Now I have new clues after taking out the switches again and hard wiring the 20Vrng. It's still not right, I haven't checked much farther forward in the circuit past the front end tho, for the sake that it seems to be working so good in ways, I figured it's just all solder brideges in the SW's, and/or traces around them..........and I was hoping that they should be easy to pin point, like some are. But I hat already cleaned the SW's, scrapped around the traces, and still had problems, so I gave up and did the real heavy duty de-soldering, and took out the SW again. And have a different problem, to figure out.
So in short you have totally trashed you meter?
 
Top