JFET Testing - ARP 2600 Synthesizer

[Erik S]

I'm trying to test some JFETs in-circuit and make sense of the readings. These are part of the keyboard for a synthesizer I'm working on.

I'm working from a lot of guesses and limited understanding, so if I'm saying something that makes no sense, please call me out! I'm also hoping that taking the time to type this out will help me organize my thoughts a little.

I included a section of the schematic showing the two pieces highlighted in yellow. Each is a metal can dual package JFET (IMF3958). The keyboard generates a different Control Voltage for each key, and in the case that two keys are pressed, a second voltage indicating the difference between the two keys. These CV values are then stored in capacitors (C2, and C4) so the synth can continue to play those notes after you release the keys. The JFETs are working along with some op amps to buffer the voltage from those capacitors. (The manual has an explanation of how that works, I certainly didn't figure that out just from the schematic).

I understand testing anything in circuit is not ideal, but given that these have 6 legs, and are out of production, I was hoping to collect some data instead of yanking them out to test them, especially since I'm not at all sure these are the source of my problem. I tried shorting each Gate to it's Source with my meter lead, and tested the Drain to Source resistance, expecting the channel to open, and resistance to go from high to low. One out of 4 gave me about what I was expecting, but the other three are still showing pretty low resistance in the "closed" condition.

From looking at the schematic, I would guess that each pair would test the same.

Part	"Open"	"Closed"
Q2-1	400 ohms	20 M
Q2-2	400 ohms	27 k
Q5-1	400 ohms	27 k
Q5-2	400 ohms	24 k

Some things I would love to know, if you know them and have made it this far, and are taking pity on me:

Does my test procedure make sense?
Did testing these in-circuit yield any useful data?
Would you assume one or both of these is faulty?
Would you assume that each pair would test the same?
Would you assume that each of the 4 individual JFETs would test the same?
What the heck does the dotted line oval symbol mean between C2 and Q2?
Should I stick to assembling pedals since I clearly don't know what I'm doing? Probably yes on that one.

 

[bowanderror]

I think the Q2/Q5 dual JFET is just a high-impedance input stage for the opamp (which itself is just a buffer). In terms of testing in-circuit, you're best bet is probably testing it in-use, monitoring voltages with a scope as you trigger keys on the keyboard. Maybe someone else can chime in on an easier way to test the viability of the FETs without removing them. If it came down to it, you could probably replace the dual JFET & opamp with just a FET input opamp.

That oval symbol intrigued me though, never seen that particular one before but it reminded me of a shield termination - e.g. from a shielded cable. Turns out it's called a guard ring and it's there to prevent the held voltage from leaking out and/or leakage from adjacent components on the PCB. You can see the funky shaped trace on the PCB, it surrounds one side of the 0.47uF cap & the 2nd JFET's drain (in the IMF3958).



If you wanna learn more, I found a good application note that talks about electrical guarding. Figure 6 shows an example layout of a guard ring on high-z inputs. Apologies for resurrecting a 2 month old thread & keep us posted on your progress.

 

[Erik S]

Apologies for resurrecting a 2 month old thread & keep us posted on your progress.

Thanks so much for reading/ thinking about it/ taking the time to respond! You’re right of course - the best troubleshooting would be to understand what should be present at each node, and test in use. I’ve been lucky with shortcuts/ finding dead components, but sometimes you just have to do the homework.

That guard ring makes sense! I think they went to a lot of trouble to try and avoid the stored voltage (and the corresponding note) from drifting. Nice to have one less mystery in this project.

As far as the JFETs I posted about, I found some NOS ones somewhere, ordered them, tested them, then pulled the ones from the synth, verified they tested the same, then put the originals back in. I could have avoided all that with some testing, but at least now I have some spares.

I’ve since worked out that I don’t think there’s anything wrong with the stored note voltages, the issue seems to be with triggering.

The symptom is: hold a high note (plays), key a low note (switches to the low note), release the low note, and there it should switch back to playing the held high note, but it keeps playing the low note.

I think I understand how that was accomplished originally, but at some point in the last 50 years someone hacked in an aftermarket midi conversion, and I need to work out if/ how that’s related.

Today I got some reprieve from that problem when I was playing the synth and one of the sliders crapped out. Got to pull one of the main boards, desolder and clean the slider, and put it back in. Nice to have a problem I know how to solve!

Here’s a couple pics I took during that process.

 

[bowanderror]

Wow those hand-taped PCBs are GORGEOUS! Looks in decent enough shape, although the presence of flux & heat marks on only some pads makes me wonder if there are undiscovered mods & component changes (other than what you've had to do). Same goes for the bodge wires, are those part of the MIDI retrofit you were talking about?

Glad to hear the JFETs tested out OK, at least you can banish that though for now.

That guard ring makes sense! I think they went to a lot of trouble to try and avoid the stored voltage (and the corresponding note) from drifting. Nice to have one less mystery in this project.

Such a neat little PCB technique. I'll have to keep my eyes out for it on other old boards.

I’ve since worked out that I don’t think there’s anything wrong with the stored note voltages, the issue seems to be with triggering.

The symptom is: hold a high note (plays), key a low note (switches to the low note), release the low note, and there it should switch back to playing the held high note, but it keeps playing the low note.

Have you checked the caps in the voltage storage & triggering circuits? Depending on the type, they're probably the components with the shortest shelf life. A bum cap or leaky board could defo mess with logic thresholds & time constants, and would be more terminal in a trigger circuit. You could try injecting triggers directly starting where it interacts with the voltage memory and working back through the earlier parts of the trigger circuit. Like a reverse audio probe, you'd at least get to ID which circuit blocks are functioning.

I'd be interested to peek at the triggering circuit & how it interfaces with the stored voltage section if you've got the schematic for it. I might not be able to help you fix it, but it will be interesting either way!

 

[Erik S]

Wow those hand-taped PCBs are GORGEOUS! Looks in decent enough shape, although the presence of flux & heat marks on only some pads makes me wonder if there are undiscovered mods & component changes (other than what you've had to do). Same goes for the bodge wires, are those part of the MIDI retrofit you were talking about?

It definitely is nice to look at. It’s also pretty great to work on compared to modern stuff. Access is easy, and the documentation is excellent.

Pretty sure all the flux is either original or from stuff I’ve replaced. The spliced in wires are from the midi retrofit. Looks a little ugly, but installed correctly as far as I can tell.

I'd be interested to peek at the triggering circuit & how it interfaces with the stored voltage section if you've got the schematic for it. I might not be able to help you fix it, but it will be interesting either way!

I’ll try and get an update together tomorrow and link a copy of the schematic. I still haven’t solved the problem yet, but I’ve made some progress and eliminated some more stuff.

Always nice to have another set of eyes on this stuff, and writing out what I think I know usually helps me figure out what to look at next.

 

[Passinwind]

It definitely is nice to look at. It’s also pretty great to work on compared to modern stuff. Access is easy, and the documentation is excellent.

Pretty sure all the flux is either original or from stuff I’ve replaced. The spliced in wires are from the midi retrofit. Looks a little ugly, but installed correctly as far as I can tell.

I’ll try and get an update together tomorrow and link a copy of the schematic. I still haven’t solved the problem yet, but I’ve made some progress and eliminated some more stuff.

Always nice to have another set of eyes on this stuff, and writing out what I think I know usually helps me figure out what to look at next.

The customer repair tech from ARP back then is still around, are at least was in 2020 when I had a nice chat with her at NAMM. I'll see if I can scare up some contact info.

Unfortunately I never got to work on any 2600s when I was a QC tech there, but I played the two in the employee studio quite a bit. IIRC we only built one run of those while I was on the assembly line, but I moved to test tech after only 3 months as an assembly droid.