Juno 106 - wierd distortion fault

Pulling out your hair? Don't know what to do or where to go? Ask in here.
Forum rules
READ: VSE Board-Wide Rules and Guidelines

If your Help request has been solved, please edit your first post in order to select the Image Topic Icon to let others know your topic has been solved.
polysixer
Newbie
Newbie
Posts: 56
Joined: Mon May 09, 2016 11:54 am

Re: Juno 106 - wierd distortion fault

Post by polysixer » Fri Oct 26, 2018 9:10 am

Mooger5 wrote:No problem. OK, I was indeed thinking about clipping distortion, and was about to suggest testing the power rails near the opamps, decoupling caps and also the non polarized blocking caps, as any of those could cause asymmetrical distortion.
Can you post an audio file? Is the distortion you hear present on both channels? If so, any possible culprit must be common. For instance, if there´s some leakage from the wet to the dry signal when off is selected, the suspects would not be the fets tr11 and tr12, as both would have to fail at the same time, but a leaking tr4 instead, that controls them. Anyway such situation I think would cause just a very faint chorus, audible when disengaged.
I dont know about tr9, doesn´t look like a suspect, but tr10 otoh is regulating the positive supply of the final opamp stage IC6. For some reason they found it necessary. I call IC5 the final vca btw.
Neither TR9 or TR10 are failing, neither is TR3 back in IC3. TR7 and 8 were also ruled out previously. There are no ICs left to check in the audio path either. I have discarded the diodes in the post-chorus section as well. I think I can rule out completely the chorus section as there is no leak from it when off, as you suggested. All polarized electrolytics in the PCB were replaced and this leaves me with just two bipolar caps C7 and C12 left in the VCA path to replace (C14 bipolar was also replaced), so I'll do that today. If this is not it I'm out of ideas besides recapping mylars and ceramics. I'll keep you posted and grab an audio file later.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Fri Oct 26, 2018 2:38 pm

Lost my reply. Later
Herrare umanum est.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Fri Oct 26, 2018 8:05 pm

OK.

If you lift r89, r72 and r74 the chorus will be isolated and you should definitely rule it out. Unless there´s some capacitive coupling caused by flux stains, a stray solder bridge or whatever, but that should have been already ruled out as well by this time. It´s like the first thing to check out for...

You said it´s too loud. So a bad solder joint at r39 perhaps?

Most of the amp stages are set for unity gain: r28 etc with r29... r36 with r16... So most of the signal´s amplitude stays more or less the same, and should be fairly traceable with a scope or an improvised audio probe. The key to find the clean vs distorted waveform is to set the unit to unison mode. A chord or simply a bunch of keys won´t do. If the 80017a modules are socketed, you could try removing two or three so the waves, such as a square wave, appear as hot and as uniform as possible. A distorted square wave should be noticeable despite the phasing caused by unison mode.

Check Points: r39... ic3 pins 1, 2, 3, 4, 5... ic4 pins 1 and 2... ic5 pins 1 and 8... ic2 pins 6 and 7... and that´s about it.

Good luck!
Herrare umanum est.

polysixer
Newbie
Newbie
Posts: 56
Joined: Mon May 09, 2016 11:54 am

Re: Juno 106 - wierd distortion fault

Post by polysixer » Sat Oct 27, 2018 5:49 pm

Mooger5 wrote:OK.

If you lift r89, r72 and r74 the chorus will be isolated and you should definitely rule it out. Unless there´s some capacitive coupling caused by flux stains, a stray solder bridge or whatever, but that should have been already ruled out as well by this time. It´s like the first thing to check out for...

You said it´s too loud. So a bad solder joint at r39 perhaps?

Most of the amp stages are set for unity gain: r28 etc with r29... r36 with r16... So most of the signal´s amplitude stays more or less the same, and should be fairly traceable with a scope or an improvised audio probe. The key to find the clean vs distorted waveform is to set the unit to unison mode. A chord or simply a bunch of keys won´t do. If the 80017a modules are socketed, you could try removing two or three so the waves, such as a square wave, appear as hot and as uniform as possible. A distorted square wave should be noticeable despite the phasing caused by unison mode.

Check Points: r39... ic3 pins 1, 2, 3, 4, 5... ic4 pins 1 and 2... ic5 pins 1 and 8... ic2 pins 6 and 7... and that´s about it.

Good luck!
Thanks, your suggestion for isolating full sections is a really good one. So yeah, lifted r89, 72 and 74 to leave out the chorus and the problem still happens, as expected earlier. I have also lifted r24, 28 and 26 to bypass the HPF low pass filter boost and the higher 2 positions and leave the signal clean, and the problem persists, so that leaves out all the caps in the HPF.

Of course I have tried to scope the issue but it's incredibly hard because unison doesn't get distortion unless you push the VCA higher. For example at position VCA 0 unison won't distort, as 1 or 2 notes won't either. On the third note it will distort easily. It doesn't seem like an amplitude problem but a phase or frequency problem (that's why I wanted to isolate from the HPF first). IC tolerances on the VCA IC5 don't seem to matter either, as I've swapped it with the one from the good chorus board sample with the same result. I would have to check now all the resistors controlling the VCA IO and the following low pass filter composed of IC2b (already swapped and ruled out) R16 and C5 cap.

For good measure I’ve also bypassed the HML switch in case that’s inteoducing noise.

This is pretty challenging and a bit frustrating so far. Will post more updates later.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Sat Oct 27, 2018 8:05 pm

You can isolate the vca by removing ic5 from its socket and bridging pins 1 and 8 with a 33k resistor.
It´s too bad the datasheet is in japanese, but it´s possible to see it´s a bit like the ca3080; it´s an attenuator rather than an amplifier. If you bypass it with a piece of wire from pin 1 to pin 8, it will pass the signal at its full level, like it was set to +5 on the front panel. The "default" value for the vca is at 0, so in reality the signal is "normally" at half level. IC2 is not a lopass filter, btw, it´s a current to voltage converter, as the vca outputs current. The cap is there to protect the opamp from going into selfoscillation. Now, with the vca isolated, just a wire from 1 to 8 and ic2 gain is set for unity because of r36 and r16 having equal values. By adding a resistor of 33k in series with r36 we´re halfing the the voltage output of ic2. So it will be like the vca being set to zero, If IC5 responds linearly of course.
If isolating the vca kills the problem, then it´s something in the voltage control. R31 is of a low value. If it failed, better replace with another of higher wattage...

The more frustrating it is, the more rewarding it will be when fixed heh
Herrare umanum est.

polysixer
Newbie
Newbie
Posts: 56
Joined: Mon May 09, 2016 11:54 am

Re: Juno 106 - wierd distortion fault

Post by polysixer » Sun Oct 28, 2018 12:40 am

Mooger5 wrote:You can isolate the vca by removing ic5 from its socket and bridging pins 1 and 8 with a 33k resistor.
It´s too bad the datasheet is in japanese, but it´s possible to see it´s a bit like the ca3080; it´s an attenuator rather than an amplifier. If you bypass it with a piece of wire from pin 1 to pin 8, it will pass the signal at its full level, like it was set to +5 on the front panel. The "default" value for the vca is at 0, so in reality the signal is "normally" at half level. IC2 is not a lopass filter, btw, it´s a current to voltage converter, as the vca outputs current. The cap is there to protect the opamp from going into selfoscillation. Now, with the vca isolated, just a wire from 1 to 8 and ic2 gain is set for unity because of r36 and r16 having equal values. By adding a resistor of 33k in series with r36 we´re halfing the the voltage output of ic2. So it will be like the vca being set to zero, If IC5 responds linearly of course.
If isolating the vca kills the problem, then it´s something in the voltage control. R31 is of a low value. If it failed, better replace with another of higher wattage...

The more frustrating it is, the more rewarding it will be when fixed heh
This is a brilliant idea, I bypassed IC5 with a 33k resistor and indeed it's like VCA range halfway... but the problem remains. I have also replaced C5 in the current converter (thanks for the heads up) to no avail. However, this VCA test saves me from checking all the resistors surrounding IC5. Unfortunately I'm out of ideas now, as I've made the path as simple as possible. Will check voltage supplies at IC3, 4 and 6 tomorrow.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Sun Oct 28, 2018 9:34 pm

I guess all that will remain is to isolate IC3 by connecting pin 3 to pin 1, then to pin 5, etc... a piece of wire will do.
Herrare umanum est.

polysixer
Newbie
Newbie
Posts: 56
Joined: Mon May 09, 2016 11:54 am

Re: Juno 106 - wierd distortion fault

Post by polysixer » Mon Oct 29, 2018 9:47 am

Mooger5 wrote:I guess all that will remain is to isolate IC3 by connecting pin 3 to pin 1, then to pin 5, etc... a piece of wire will do.
Yes, I already did this method to "manual mux" but the problem persists. However, following your suggestions, I replaced the VCA IC5 with a 100k resistor and the problem goes away! So yes, definitely an attenuation problem. Either something is not properly setting the gain on IC5 correctly or a previous resistor is not doing its job. Encouraging at least.

Edit: IC5 uPC1252H2 datasheet in English http://www.proaudiodesignforum.com/imag ... H2_VCA.pdf

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Mon Oct 29, 2018 10:56 pm

Thanks. TBH I can´t make much of it even in English.

As you can see in the graphics (page 5) it says with a control voltage of -300mV the maximum gain is 50dB, and with a CV of +300mV the minimum gain is -50dB. So the more negative the CV goes the louder it gets. The response being linear, with zero volts the gain is also 0dB.

Read the voltage at pin 3 of IC5 while moving the VCA slider. I can´t remember what voltage comes out of the main board, so you should read that too (pin 7 of the chorus board connector). You have a voltage divider leading to IC5 pin3, but the +15V is constant, so the network of r30, r32, r165 and r31 is arranged to change the law from linear to logarithmic, I believe.
All this to say that if it sounds louder than normal, then there should be an excess of negative voltage across r31. This could mean the +15V rail is deficient or r165 is open.

A long long shot, but have you also checked the line out selector? The LMH switch? Use contact cleaner and see if it sounds better, in case you didn´t already of course!
Herrare umanum est.

polysixer
Newbie
Newbie
Posts: 56
Joined: Mon May 09, 2016 11:54 am

Re: Juno 106 - wierd distortion fault

Post by polysixer » Wed Oct 31, 2018 10:15 am

Mooger5 wrote:Thanks. TBH I can´t make much of it even in English.

As you can see in the graphics (page 5) it says with a control voltage of -300mV the maximum gain is 50dB, and with a CV of +300mV the minimum gain is -50dB. So the more negative the CV goes the louder it gets. The response being linear, with zero volts the gain is also 0dB.

Read the voltage at pin 3 of IC5 while moving the VCA slider. I can´t remember what voltage comes out of the main board, so you should read that too (pin 7 of the chorus board connector). You have a voltage divider leading to IC5 pin3, but the +15V is constant, so the network of r30, r32, r165 and r31 is arranged to change the law from linear to logarithmic, I believe.
All this to say that if it sounds louder than normal, then there should be an excess of negative voltage across r31. This could mean the +15V rail is deficient or r165 is open.

A long long shot, but have you also checked the line out selector? The LMH switch? Use contact cleaner and see if it sounds better, in case you didn´t already of course!

Ok, measurements are in: pin 3 of IC 5:

VCA=-5, 90mv
VCA=0, 30mv
VCA=+5 -30mv

So clearly the range is shifted (to positive voltage, not negative however?). The rails are all a perfect +15V in R165 and R17 and -15 in R35. All resistors across the VCA level line measure perfectly, as well as all in the signal path and surrounding IC5 except one: R165 measures 4K in circuit, although it's 15K out of circuit. Replaced with a new one, same behavior. Why this happens I have no idea, the trace is short and it looks good, goes into pin3 and connects to R31 and R32 which are not failing and giving correct measurement, and R31 grounding is good. I guess the problem could be easily fixed replacing R165 with a trimpot to measure 0mv at pin3 in 0 VCA position. It doesn't explain what causes R165 to drop the resistance though.

LMH switch bypass was one of the first things I did before focusing on the VCA ;)

Edit: TR5 is in the same +15V line, could a failure here explain a parallel resistance to R165 and hence a drop?

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Wed Oct 31, 2018 2:47 pm

Hi. Good observations. You should have included the voltages between R30 and the connector. But for now from your readings we can see the response seems linear. Zero at +30mV then +60/-60. We should remember that zero on the front panel doesn´t have to mean zero volts in the control circuitry. It´s only an abstraction. If it´s positively biased, I guess it´s because of the highish output of the voice modules that needs to be tamed down. So my initial observations were wrong: no signs of a logarithmic response. R161 is there for the positive bias, r30 and 32 limit the negative current, the np cap smooths out any sudden changes, r31 turns them back into a voltage that the chip can sense, and that´s it. Not really a volume control, anyway. Not that it would matter much. So all seems well and I´m officially clueless right now.

IDK what you mean by tr5 in the +15 line. The emitter is tied to the -15V line. Note the down arrow is the negative supply.

In the chorus schematic, on the bottom right, there are all the symbols for the different voltages and grounds. Note the Analog 5V and the Digital Ground are different from Digital 5V and the Analog Ground, for instance.

There is also one thing: there were several revisions of the main board and possibly of the chorus board as well. My FrankenJuno has one of the early main boards. There is a ground black wire from the chorus board that the old main board doesn´t provide a connection for. If this wire is left floating there will be some background noise that would be mistaken for voice modules in need of replacement. By connecting it to ground the noise disappears.
There is also another connection related to the Tune function which is missing. Leaving it unconnected the analog 5V is drained and the synth pitch drops when the lfo is triggered by the bender. It´s weird and took me a long time to fix it.
So what´s the story of that Juno? Is it an early model? Is it made of salvaged parts?
Herrare umanum est.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Wed Oct 31, 2018 3:45 pm

Forget it! I dug out my Juno from storage, and my readings don´t match yours! The extremes are identical, -5 = 90mV and +5 = -30mV, but at halfway it reads +75mV! It reads Zero volts only at +2.5. The readings are an aproximation because they lag due to that capacitor, but we can figure the control voltage is actually logarithmic. So I was right :mrgreen:

I´m about to post some more readings. Stay tuned ;)
Herrare umanum est.

polysixer
Newbie
Newbie
Posts: 56
Joined: Mon May 09, 2016 11:54 am

Re: Juno 106 - wierd distortion fault

Post by polysixer » Wed Oct 31, 2018 4:30 pm

TR5 - emitter is indeed to -15 but collector goes into +15 through R46 and R50, so just wondering if a failure in this transistor explains the resistance drop of R165 in circuit. That could explain the scaling fault I guess. Furthermore, why does R17 connected to the same +15 line measure correctly and not drop resistance? My Juno seems like an early unit judging by serial number. I did make many more measurements and will post later when I get home. As always, many thanks for your input.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Wed Oct 31, 2018 4:42 pm

IC5:
pin 7 and pin 6 = +11.96V
pin 5 and pin 6 = -2.7V

Look for hairline solder bridges near pin 5 as some pads there are too close. Don´t use RoHS solder.

Voltage across pin7 of connector for VCA Level and ground:
-5 = +3.84V
0 = +2.74V
+5 = -6.13V

Voltage across C7 is around +1.2V at Zero.

R165 in circuit reads 1.5K (!)
Measured voltage across R165 should be slightly less than +15V.

Hope this helps!
Herrare umanum est.

Mooger5
Expert Member
Expert Member
Posts: 1458
Joined: Tue May 06, 2008 5:11 pm
Location: Lisbon

Re: Juno 106 - wierd distortion fault

Post by Mooger5 » Wed Oct 31, 2018 5:23 pm

polysixer wrote:TR5 - emitter is indeed to -15 but collector goes into +15 through R46 and R50, so just wondering if a failure in this transistor explains the resistance drop of R165 in circuit. That could explain the scaling fault I guess. Furthermore, why does R17 connected to the same +15 line measure correctly and not drop resistance? My Juno seems like an early unit judging by serial number. I did make many more measurements and will post later when I get home. As always, many thanks for your input.
I see your point. Well TR5 is in the Chorus On/Off command path and the chorus is working, so TR5 is behaving correctly. It wouldn´t draw much current anyway, because of the resistors you mentioned. That +15V line should be the same everywhere, give or take a few mV, and there should be continuity between every +15V point. So if TR5 or any other component caused a drop in the voltage, that drop would be noticed at the other +15 points just as well.
Anyway resistors measured in circuit aren´t very good for reference, as they are always in parallel with some other components with their own internal resistance which will cause misreadings.
Whatever the misread on r165 though, mine and yours should match but 4k and 1.5k are nowhere near so it must mean something´s not right. If anything, 4k being a higher resistance means a lesser positive bias, favouring the ratio of the negative supply...
Herrare umanum est.

Post Reply