Vintage Synth Repair

I have an MKI here that's been driving me mental for the last little while. Basically any note you hit on the keyboard leads to a sweep at the highest possible audible pitch. You can actually play the synth normally, but the moment you let go of the key, it picks up from that note and sweeps to a higher and higher frequency until it's inaudible. I can slow the sweep down with the portamento slider.

I've done thorough tests in the AR / ADSR section and they all check out. I thought that perhaps the caps were dead and were not discharging properly on release but they did. I even removed the keyboard assembly thinking it might be a dead cap somewhere there. I applied just cv to trigger the synth and it did the same.  I then started looking to the LFO but you can control it with all the sliders normally. It behaves as expected while this sweep occurs.

I think I'm going to owe someone a case of beer here. :)

TIA

I suspect the error to be on board A1; check the whole vicinity of the CV
memory circuit.

I've been leaning to board A over and over again too. I'll look again now.

I suppose you have the service manual?

Check the main keyboard CV out (pin 8 on the connector). I believe this
should even work without the backplane. If this voltage changes after the
key release it's definitely a keyboard CV memory issue. Since you mention
that the pitch changes over time (not immediately) I believe that the
memory cap is actually working. Check Q7.

If all this fails I have a partly working MKI here that I could use to
compare measurements.

Of course there's also the faint possibility that a key contact has gotten
stuck.

Which leads to a number of additional questions:

You mentioned it is a MKI. does it have the original three-buss keyboard or
a two-buss keyboard? You also mentioned applying a CV; I take this as
"external CV". The MKI did not come with CV ins or outs originally. So this
is a retrofit?

Hi Malte,

Thank you for the help so far! I figured I should reply back when I had something useful to add.

I actually got sidetracked with the original issue above. The main output began to fail. I tracked it down to a faulty diode and transistor on board C. (The manual showed a 5172 but this had a 2SC828 in it).

This is a three bus black MKI but there are no CV I/O mods on it yet. I plan to put them in after this is working again. In order to apply CV I've often disconnected the keyboard and have applied a voltage directly to C6 on Board A-1.

As it turns out, you were correct! Q7 was failing. I replaced it managed to get rid of that problem. Unfortunately that revealed an underlying issue :(

Now when one releases the key, it hangs and tends to lift up by a semitone or so. If I set the output to the main VCA - AR - it will remain audible endlessly. If I set it to ADSR, it can fade away (though I believe the signal is still passing though up to that envelope. The ADSR just happens to hide the signal by fading it out).

Ugh.

> Now when one releases the key, it hangs and tends to lift up by a semitone or
> so.

This is not uncommon. I don't know offhand if this could be a calibration
issue (probably not). However I have observed that the whole CV memory
circuit is very sensitive to dust, dirt and other contamination. A very
thorough cleaning of this area might already do the job; in addition I tend
to exchange the opamps if there is even a slight problem.

On to the next issue:

> If I set the output to the main VCA - AR - it will remain audible endlessly.
> If I set it to ADSR, it can fade away (though I believe the signal is still
> passing though up to that envelope. The ADSR just happens to hide the signal
> by fading it out).

I'm not sure if I follow you completely: Of course the signal will pass up
to the envelope (or more correctly: to the VCA). There will always be
signal in front of the VCA, regardless if a key is pressed or not! (Ok, it
depends on several other factors as well; but with a basic patch you will
get signal.)

At this point I would start by verifying if the ADSR is working properly
and is triggered/gated properly. This is easiest performed with either the
VCO or the VCF (due to the still persisting CV problem I'd suggest the VCF
route):

Set all four ADSR sliders to about 50%.
Send noise into VCF.
Set VCF Freq and resonance to roughly 50%.
Set ADSR to VCF to 100%% (and of course the switch to ADSR.
Set VCA Gain to 100%.
Set VCA ADSR/AR slider to zero.

Observe results. You should get a constant smooth noise. Now press a key.
If the ADSR works the noise should get brighter, then slightly darker (and
should stay this way as long as you hold the key), and then get darker
again after you release the key.

Hi again! I finally had a chance to go through this. Firstly, the synth did need need a full calibration. I've got the calibration values ultra close (considering the trimmers a still original ones). Here are my responses to your reply:

>> Verifying if the ADSR:
I can confirm that it behaved as you described. I also tested the attack & gain in the AR and can confirm that they work as well.

>AR:
To clarify, what I'm saying is that if you set the AR gain is increased above 0 you will hear the signal regardless of whether you play a note or not. It also makes no difference whether you have the gain selector switch selected one way or the other (to AR or ADSR). As long as AR gain is up in any way, you'll hear the signal. (Perhaps I'm crazy and it's *supposed* to do this. I thought that that it was only supposed to pass a signal when a note is played.)

> Semitone change on release & CV Memory:
I actually gave all three boards a thorough cleaning so no dust or grit is present. After the calibration, the note still swings by a semi tone on release. I'm heading back to Board A now to troubleshoot it further.

Curious. Even with Board A disconnected, Board B & C still generate a low tone (again with AR's gain up).

Hello,

it's been awhile and I was away for two weeks; so let's see if we can get
any further now. For now I would suggest working on one problem at a time.

Let's start with the VCA leakage.

You wrote:

>>AR:
> To clarify, what I'm saying is that if you set the AR gain is increased above
> 0 you will hear the signal regardless of whether you play a note or not. It
> also makes no difference whether you have the gain selector switch selected
> one way or the other (to AR or ADSR). As long as AR gain is up in any way,
> you'll hear the signal. (Perhaps I'm crazy and it's *supposed* to do this. I
> thought that that it was only supposed to pass a signal when a note is
> played.)

Before we continue here let's make sure we're talking about the same thing:

The slider in question is the one in the lower row (left from the ADSR
sliders)? Or is it by chance the one in the upper row (left from the AR
sliders)?

Hi Malte! Sounds like it was perfect timing! 

I'm referring to the upper row: AR gain. This is the one that allows the leakage. Lower row - ADSR gain - does not allow leakage.

At 12:59 Uhr +0000 24.07.2011, Boddington wrote:
> I'm referring to the upper row: AR gain.

That's not what it is called and labelled. As we can see it is called "VCA
Gain" and is supposed to do exactly what you describe. So there's no actual
problem here; for "normal" playing this slider needs to be in the bottom
position. You only raise it for "drone" sounds that are supposed to go on
forever.

I believe mainly ARP synths have this feature (I've seen it on the 2600,
Odyssey and Axxe); and it seems to be one of the most typical sources of
confusion. I recently sold a MkII and of course the buyer also called to
ask if there is anything wrong with it because the sound wouldn't go away
completely. There are even more prominent users having encountered this: I
recall a story that an ARP representative delivered a 2600 to Joe Zawinul
(of Weather Report fame); and a few days later he got a call from Joe "this
is a totally great sounding instrument, but how do I make it stop?". (I
think the story is in the "Vintage Synthesizers" book by Mark Vail; I just
can't find my copy at the moment.)

So let's conclude this with a quote from the Odyssey Owner's Manual:

"The VCA GAIN control has nothing to do with the keyboard. The VCA Gain
control allows a certain amount of signal to pass through the VCA at all
times."


I'll think about the CV problem later.

Thank you for clearing that up. I suspected that this was the case as well but I never found that clarification. As you pointed out, the documentation suggests it but I wasn't 100% sure. I've only had one other Oddyssey here before and it was in such bad condition that I didn't proceed with any repairs to it. Sorry to waste your time on that.

I've started to look at that semi-tone issue. On Board A, pin 8 - cv out - I can see that the voltage shifts on release. So far I've traced it back as far as Q8, the 3958. I can drain the voltage from there. I ca't say that I'v found anything faulty yet. I'm looking into what inputs and outputs one should expect to find.

At 7:30 Uhr -0700 24.07.2011, JP Szarak wrote:
> can anyone direct me in how i should go about fixing a non functioning
>white /
> pink noise generator on a mk ii odyssey?

The noise generator resides on board A-II; I've found that often the first
opamp has failed. This is Z5 which can be replaced by a LM301.

Since both white and pink noise don't work it's either this one or even
before that. I presonally would simply switch Z5 because I've seen several
Odysseys where Z5 went bad.

> I've started to look at that semi-tone issue. On Board A, pin 8 - cv out - I
> can see that the voltage shifts on release. So far I've traced it back as far
> as Q8, the 3958. I can drain the voltage from there. I ca't say that I'v
>found
> anything faulty yet. I'm looking into what inputs and outputs one should
> expect to find.

I would probably replace A1 (LM1458) first. I'm under the impression that
the quality of opamps has improved significantly. Some Odysseys I worked on
immediately got more stable after replacing the opamps in hte CV circuit.
One could even substitute it with one of the higher quality opamps; however
there might not be any significant improvements.

thank you.  i'll try to repair it.

jp



________________________________

From: Malte Rogacki <gacki@...>
To: vintagesynthrepair@yahoogroups.com
Sent: Sunday, July 24, 2011 11:28:42 AM
Subject: Re: [vintagesynthrepair] Arp Odyssey: noise generator


  
At 7:30 Uhr -0700 24.07.2011, JP Szarak wrote:
> can anyone direct me in how i should go about fixing a non functioning
>white /
> pink noise generator on a mk ii odyssey?

The noise generator resides on board A-II; I've found that often the first
opamp has failed. This is Z5 which can be replaced by a LM301.

Since both white and pink noise don't work it's either this one or even
before that. I presonally would simply switch Z5 because I've seen several
Odysseys where Z5 went bad.

Just replaced A1B. No difference with regard to the semitone issue I have. I'll move on to A1A and replace that too. No harm in trying.

At 16:32 Uhr +0000 24.07.2011, Boddington wrote:
> Just replaced A1B. No difference with regard to the semitone issue I have.
> I'll move on to A1A and replace that too. No harm in trying.

Ahem.

A1A and A1B are the two "halfs" of A1. If you replaced A1B you
automatically replaced A1A as well.

Ha! Right of course. I'm doing 5 things at once here. I meant to say A2. I'll look there as well.

So let's try a different approach here.

How exactly does the whole CV memory circuit work? The CV is stored by
charging C8. As long as a key is held nothing happens as far as the memory
function goes. If the key is released Q7 is turned off so C8 cannot
discharge; now we're in the memory "function". Q8 and A1B are the buffer.

It would be interesting to learn more about the exact behaviour of the
problem. Does the semitone change happen immediately and at once or
gradually? Does the note then stay at the same pitch afterwards or does it
continue moving upward??

Depending on that information I could make a better guess.

Sounds good!

> Does the semitone change happen immediately and at once or gradually?

The semitone change always occurs immediately on release. It is not never gradual. I tried increasing the portamento slider to hear if it could be change its speed but it can't.

> Does the note then stay at the same pitch afterwards or does it
> continue moving upward?? 

The note stays the same pitch when it's done. The semitone shift is always an upward change. 90% of the time it is the equivalent of a semitone. 10% of the time it's bit more (up to a full note or note and a half). It never continues to move upward once that one upward swing has occurred.

Ok, this is making things a bit clearer. I'm thinking about two scenarios here:

1. Q7 turns off a little too late. This allows C8 to discharge for a very
small time. Different possibilities:

1.1. Q7 itself is faulty. You said you already replaced it so this isn't
very likely but cannot be ruled out.
1.2. The various keyboard contacts break contact in the wrong order.
Usually the gate buss should break contact first, then the CV buss. Is the
behaviour uniform for all keys? Is the behaviour identical when releasing
the key slowly or quickly?
1.3. Other things in the path from the gate buss to Q7.


2. The circuit finds some kind of "balance" after Q7 turns off.

2.1. Are +/-15V exactly identical? Is there possibly ripple on one of the
rails?
2.2. Perhaps something's wrong with Q8.


At this point I would probably first check the power supply voltages and
the try to see (possibly with a two channel scope) if the busses are
switching in the right order and if there is some delayed switching of Q7
(compared to the actual gate buss).

>> 1. Q7 turns off a little too late. This allows C8 to discharge for a very small time. Different possibilities:

>> 1.1. Q7 itself is faulty.

Indeed. This is a new component now. At least that sweep is gone. It does appear to work but I'll need to investigate that possible delay you describe.

>> 1.2. The various keyboard contacts break contact in the wrong order. Usually the gate buss should break contact first, then the CV buss. Is the behaviour uniform for all keys? Is the behaviour identical when releasing the key slowly or quickly?

The behaviour does occur on each key but I have to say that the amount of swing sounds inconsistent. It definitely happens on each key but sometimes it's more dramatic. I'll apply some external CV / gate and see if it produces the same problem. That should help me troubleshoot the kbd assembly.

>> 1.3. Other things in the path from the gate buss to Q7.

On key press CR2 and R21 show that the gate is active. The opposite side of C7 correctly shows -15V.

2. The circuit finds some kind of "balance" after Q7 turns off.

>> 2.1. Are +/-15V exactly identical? Is there possibly ripple on one of the rails?

They are, yes. +/-15V. All new caps on the PSU board to be 100% sure. I checked for any ripples but I'm finding any so far.

>> 2.2. Perhaps something's wrong with Q8.

I'm beginning to think that the 3958 is faulty too. I'm not clear on what measurements I should find from each pin though. I also don't have a direct replacement so I'd have to concoct a version with two transistors.

>> At this point I would probably first check the power supply voltages and the try to see (possibly with a two channel scope) if the busses are switching in the right order and if there is some delayed switching of Q7 (compared to the actual gate buss).

I'll have to see if I can set up a dual arrangement. I'm working 
on this at another location and will have to come back to this one.

Joy! It was the other 3958! I did a few tests between the two transistors. Sure enough, Q6 was the one shifting the voltage.

--- In vintagesynthrepair@yahoogroups.com, "Boddington" <f115@...> wrote:
>
> Joy! It was the other 3958! I did a few tests between the two transistors. Sure enough, Q6 was the one shifting the voltage.
>

Sorry to hijack this thread!

Is there any easy to find equivalents for:

Q7 - 2N6076 ->
C8 - Polycarb Memorycapacitor ->
Q8 - IMF3958 ->

Thanks in advance!

/Martin

No problem Martin. C8 is still available:

http://www.newark.com/jsp/search/productdetail.jsp?sku=66K4992

I haven't had to look for a 6076 but for the 3958 I swapped it out for two MPF102s:

http://www.newark.com/fairchild-semiconductor/mpf102-d27z/n-channel-jfet-25v-to-92/dp/78K6093

You have to match the pin positions correctly but it will work. As you may have read, this same swap worked fine for Q6 on Board A as part of the same memory circuit as well.

If you need datasheets on these to get the pins right, just ask.

> 
> Is there any easy to find equivalents for:
> 
> Q7 - 2N6076 ->
> C8 - Polycarb Memorycapacitor ->
> Q8 - IMF3958 ->
> 
> Thanks in advance!
> 
> /Martin
>