Roland Jupiter 8 VCO1 drift

[DesolationBlvd]

Looked at the output of pin 6 on IC 3A, and switching between waveforms had no effect.

Ok, scope or DMM, Probe pins 8, 9, and 11 of the 3046, and repeat same test for each pin, looking for any changes from saw to square.

Also, not sure if you have seen this site, but it may be useful:

https://jup8restoration.wordpress.com/2 ... rd-repair/

So what should I be looking at on those pins - the DC voltage, or the waveform?

I'm having a feeling that it's not the 3046. The tuning is otherwise stable.

[Don T]

Looked at the output of pin 6 on IC 3A, and switching between waveforms had no effect.

Ok, scope or DMM, Probe pins 8, 9, and 11 of the 3046, and repeat same test for each pin, looking for any changes from saw to square.

Also, not sure if you have seen this site, but it may be useful:

https://jup8restoration.wordpress.com/2 ... rd-repair/

So what should I be looking at on those pins - the DC voltage, or the waveform?

I'm having a feeling that it's not the 3046. The tuning is otherwise stable.

Any slight change of voltage when switching from Saw to Square. Remember, 12 cents is not a lot of difference in voltage, so you may have to use your scope (DC coupled), and set the sensitivity high. Also, while probing, I'd compare VCO 1 against a known good voice, looking for any differences.

[DesolationBlvd]

Probed the pins on the 3046, switching between saw and pulse on VCO1. The voltage did not change.

I have attached a clip, each time rotating through the eight voices. First is VCO1 and pulse only. There seems to be a weird bleed on the odd numbered voices. There is no bleed if set to triangle, saw, or square. Second is a test of tuning with both VCOs on pulse, satisfyingly in tune, at least as much so as my OB-8. Third is when I flip VCO1 to saw and change nothing else. This is where the tuning gets thrown off.

[Don T]

Probed the pins on the 3046, switching between saw and pulse on VCO1. The voltage did not change.

I have attached a clip, each time rotating through the eight voices. First is VCO1 and pulse only. There seems to be a weird bleed on the odd numbered voices. There is no bleed if set to triangle, saw, or square. Second is a test of tuning with both VCOs on pulse, satisfyingly in tune, at least as much so as my OB-8. Third is when I flip VCO1 to saw and change nothing else. This is where the tuning gets thrown off.

Ok, that's just messed up! If the CV doesn't change, the pitch isn't supposed to change, unless...

Let me go look at that schematic again. Are both tempco resistors solidly against the 3046, and with thermal paste?

Perhaps the "bleed" is due to the odd-numbered voices allowing the highest overtones of a saw wave through? It sounds as if a high=pass filter is turned almost all the way up.

As tedious as this is, it may be time to probe voltages in VCO 1 and compare to any other good VCO while playing the same pitch.

[Don T]

Oh, you do have a copy of the service manual, yes?

[DesolationBlvd]

Yep, just the second edition which is missing the first edition's Theory of Operation. I think that might be very helpful to me, seeing as I used the Theory of Operation from the DMX service manual to fix its hi-hat not properly closing.

Here's another clip. The bleeding responds to the pitch of VCO2. It does not respond to the waveform of VCO2.

[Don T]

Ok, after looking over the schematics again, I've got nothing. Sorry

[DesolationBlvd]

Revisiting the JP-8, of note is that my +10V and -5V on the voice cards is closer to +12 and -6V respectively.

I tried to look for a rogue signal modulating the pulse width on VCO1 to create the bleed, but couldn't find it.

[KSS]

Did you correct the voltages yet?

[DesolationBlvd]

I'm actually stumped on the +10 / -5 adjustment. I followed the source to the mod-con, but there isn't a trimmer like for the 13.

Mine measures 11.35 at C61 and -6.1 at C62. (11.3, -6.13 on the lower board)

[KSS]

What type of resistors are R85,86,87,88? If they're carbon comp, replace them. R86, 88 set the 'constant' current for their zeners.
When you measured the too-high voltages, were the rails normally loaded?
Zeners will naturally vary, but they should be in the 5% range, and you're reporting more like 20%.

[KSS]

Do you have a scope?
If so have a look at the rails. There might be something you'll see on scope that is misinterpreted in a meter-only reading.

Edit: and of course look at the main rails too. What do they measure?

[KSS]

Just read the thread. Hadn't done that before.

Still feel like you should check the rails and zener regulated voltages -one should ALWAYS do that- but don't really think they're currently the direct source of the problem. They may still however have been the contributing factors to creating the problem in the first place. Which is why i wouldn't 'take them off the table' just yet.

Since the thread has actually two different reports, can you clarify which exactly is your current problem?
Does it go flatter after auto-tune on the single VCO of a board pair where the 2nd VCO tunes fine?
Are Saw and Pulse always different? Or only after auto-tune?

I'm presently believing something related to auto-tune, and find it potentially significant that the pitch difference between pre and post AT for the two WFs is exactly doubled in Hz. If indeed that describes your present situation.

Presently expecting this to be a cap or cracked R, or marginal trace problem.

[KSS]

Speaking to the marginal trace problem -well, the R and C too- can you give us a photo of the VCO in question? PCB both sides. Both zoomed on the problem VCO and also wide enough to see both at once.

[Don T]

If I remember correctly, a lot of Roland power supplies do not have a negative rail adjustment. I assume because the value isn't as critical.

[DesolationBlvd]

Since the thread has actually two different reports, can you clarify which exactly is your current problem?
Does it go flatter after auto-tune on the single VCO of a board pair where the 2nd VCO tunes fine?
Are Saw and Pulse always different? Or only after auto-tune?

On my JP-8, the saws on the odd numbered voices are slightly flat compared to the pulses, and this is both before and after auto-tune. The auto-tune seems to work fine for the pulse.

[Don T]

On my JP-8, the saws on the odd numbered voices are slightly flat compared to the pulses, and this is both before and after auto-tune. The auto-tune seems to work fine for the pulse.

Ah, this I did not know. I thought it was affecting only one voice, not all of them on a card. That means I had you looking in the wrong place for a voltage drop when switching from saw to square. Let me dig through the schematics again

[KSS]

On my JP-8, the saws on the odd numbered voices are slightly flat compared to the pulses, and this is both before and after auto-tune. The auto-tune seems to work fine for the pulse.

And the even numbered saws match their pulse outputs?
--------
Glad i asked. Like Don T, I thought we were talking about one vco being out. period. Details matter.

[DesolationBlvd]

On my JP-8, the saws on the odd numbered voices are slightly flat compared to the pulses, and this is both before and after auto-tune. The auto-tune seems to work fine for the pulse.

And the even numbered saws match their pulse outputs?
--------
Glad i asked. Like Don T, I thought we were talking about one vco being out. period. Details matter.

Sorry I forgot to respond to this thread. Yes, the even saws match the pulses.

[DesolationBlvd]

Bumping: I replaced the TR15 and TR16 on both upper and lower boards. No effect, good or bad.

[rich_de]

Edit: This is response to donepe's 2-year old original post, but maybe it gives some ideas for DesolationBlvd's issue.

Now that I have an account, let me chime in. Background: I've built a poly that does what the JP8 does, and a slight bit more, but when I set out to do that, I estimated the size of the effort and made it a company project, which means it's (hopefully) going to be a small scale commercial boutique run, outside of DIY scope, which is why I won't bother you with it here. But I am very familar with the signal chain .

There are three sources which can throw VCO1 off tune, aside from the CV: X-MOD and VCO MOD on the voice, and VCO1 MOD from external LM-12. I would try to selectively disable those to see if they might have an influence.

X-MOD can be suppressed by cutting off control current to IC18A (assuming IC18A itself is good). Either take out TR13A, or connect pin 1 of IC18A to V-. At your own risk, of course, but it should be safe, because for the BA662, the control current input at pin 1 is one diode drop above V-, with low resistance, anyway. Do NOT connect it to ground -> immediate classic OTA fry.

VCO MOD can easily be suppressed by grounding pin 5 of IC17 (or the same signal at R100A). Again at your own risk , but because pin 1 is fed through 10k, the 4053 should survive the worst case 1.5mA for the test.

VCI1 MOD at LM12 should not be a source of problems, because it would affect other voices as well.

Oh, and one more thing: The synth prototype of my own design also has tuning issues, but affecting all 4 voices on a layer. This goes away when the "module controller" (in jp parlance) power connector gets contact sprayed and wiggled, or when I clip a separate analog ground connection over this power input. I have to replace the connectors with good gold plated ones, but I took note that the JP oscillators are very susceptible to the slightest ground resistance. Your issue is likely unrelated, because of the precise single location where it happens, but it should be kept in mind as a theoretical source of trouble.

[DesolationBlvd]

^I've ruled out the X-MOD, VCO MOD, and external VCO1 MOD. I disconnected the bender box and the issues still persisted.

Again, it's just very weird because it's specific to waveform selection.

[rich_de]

As noted in my edit, I was focused on the original thread issue than your later waveform select one. Let me try again

The bum oscillator is much closer to in-tune when it is set to pulse/square than when it is set to saw.

The saw input to the mux (IC7 pin 14) "directly" loads the main saw signal, while the other signals are buffered through ICs 5 and 6. If I had to further debug the issue, I'd start with the mux (IC7) and its influence on the output of the main integrator (IC3A). The mux would go out and a socket be put in its place. If a quick replacement check with a new 4052 for IC7 doesn't fix it, the IC is removed again, and replaced with a socket connected 4-way manual switch that inputs pins 11,12,14, and 15, and outputs to a 220 ohm resistor (to mimic the internal IC resistance) to pin 13. I would expect the drop to happen if switched to saw, if not, I would have to think again what's up with the 4052 setup (power lines?).

If the drop happens, it is pretty much established that the slightest loading (about 110670 ohms) of the saw core causes the trouble. This could be proven without further invasiveness by buffering the signal with a TL072 on the socket/switch contraption. If that removes the drop, the load is the culprit and we have to look at the output of IC3A, even if that looks weird, because IC3A is usually loaded by the much lesser R16A. (An alternative for an opposite check is to connect pin 14 of IC7 over 100k to ground, while in square mode and looking if the drop happens). IC3A would first get auxiliary power wiring tacked on to +/-13V to make sure it has stable supply. My bet would be against it being the cause, but it's quick do do. Then IC3A would be in for replacement with a fresh TL071 (or "TL080" as the schematic says). This IC is loaded with the gate drive and runs hot, so it's a candidate for failure.

If that still doesn't fix it, one would have to look at the PCB and think about leaking or stray inductance to some weakened part...

[KSS]

If that still doesn't fix it, one would have to look at the PCB and think about leaking or stray inductance to some weakened part...

Hopefully this would have been well before anything else you've described!

[DesolationBlvd]

If that still doesn't fix it, one would have to look at the PCB and think about leaking or stray inductance to some weakened part...

Hopefully this would have been well before anything else you've described!

Hmm, that sounds like it's beyond my level of expertise...

I'll check my socket stash and get some 4052s.