Norlin Moogs and their issues

During the latter part of May last month and the beginning of June we had a Moog Rogue in for a rebuild, which ostensibly just had the old tar from the front panel liner syndrome. The Rogue, as does the Opus 3 has a panel wide Bitumen based foam which is meant to keep dust out of all the sliders and switches, and was meant to mean that the switch and slider bodies were hidden from view with a nice neutral black surface under the panel.

We have done a couple of Opus 3’s before and while it was a bit messy to remove the now sticky foam liner from them, nothing had prepared us for the state of this Rogue.


The above image may give a clue but not a real idea of the extent of the difficulty in removing this stuff from the board, the tar, and it certainly was now tar, was everywhere, on all the cables, all over components on the board, deep in the sliders (but not the rotary pots as it turned out). Even the inside of the front panel required scraping with screwdrivers and a lot of Isopropanol to get it clean, the Tar literally ate into the metal of not only the front panel but also the metal parts of the switches and slider bodies, the tar also removed the coloured stripes from a lot of resistors and other components on the main pcb.


The above picture is from part way through the process, most of the major tarry bits had been removed and obviously the fader stripping and cleaning had begun, tar is still all over the wiring and all of the switch and rotary pot bodies, all the switch bodies had to be scraped to even try and make them decent, and in the end all the switches had to eventually be removed from the pcb to be thoroughly scraped and cleaned in order to make them function correctly. Tiny fragments of tar got everywhere, all over tools, hands, clothes, everything.


The above photo is of the board getting close to being complete, the wiring is all clean, the pcb is pretty clean, and only one slider is due for its rebuild, but the fun was only just beginning.

The keyboard wasn’t working well, the majority of the keys in the centre of the frame were not working properly, so we thought a good cleaning would sort this out, if only it was that simple. Someone in the past had spilt tea down into the keyboard and it had wreaked some havoc with the contacts, to the extent that for one key the contact bar had been eaten away and was just gone. A piece of Pratt-Reed J-wire was inserted into the contact retaining plastics with some difficulty, and hard wired to the common bus bar. The keyboard mechanism in the Rogue is a Panasonic one with a contact system of twin leaf switches per note, one for pitch and one for gate. Very thin and very delicate and difficult to clean and realign if necessary, which it was. This keyboard mechanism is shared by the Moog Prodigy, which we will discuss in the next article.

When we got the synth and had to take it apart, we marked the connectors in the usual way, permanent marker applied to both sides of a pluggable connector, different connectors are marked in different places so that mistakes should be impossble. Somehow or other we had tremendous problems with the CV voltages from the keyboard which would plunge to -10V on key release, and weird behaviour from the envelopes and VCA, and it was… yep the keyboard connector was in upside down, we did mark it and put it back together as marked, but it was obviously wrong. The reason that this took up far more time than is decent is because the scaling and tuning of the keyboard were still fine, we got 1V/Oct across the keyboard without problem. We replaced every chip in the envelope and VCA section trying to hunt this one down and it was only in desperation that we went back to the the details of the Service Manual and checked wiring colour codes against pin numbers on the connector. Obviously the penny now dropped and we rectified this. The irony is that the synth’s tuning didn’t need to be recalibrated at all, we had proper envelopes and still perfect tuning, no wonder that this was a tough one to find. At least the two envelope chips (CD4007UBE) are now in good quality sockets and they are well known failures in old Moogs, and some Korgs, so replacement in the future will be just a chip swap, if you can get them. They are not like CEM’s or SSM’s but they have been out of production for decades as their day has gone. They are not particularly expensive, just proving more elusive by the year.
Obviously we could not leave the toggle switches naked and exposed to view, so we made black cloth tape covers for them all that supplied all of the visual benefits of the original foam.

The Prodigy-Smack my synth up

The title line isn’t entirely a joke, the woodwork around this synth was entirely broken up, it apparently had been knocked off a keyboard stand and fallen heavily, but this wasn’t its first fall, there were signs of woodwork repairs internally from many years ago, and whatever glue was used, it was white and very brittle, as from the synths second dive, most of this had broken up. The synth was originally sent in to be tuned, and to have the woodwork sorted but even this one had a lot of hidden gremlins. Oscillator 1 was supersonic most of the time, creating a high pitched whistle which defied keyboard control, or oscillator octave switching, or anything else. It was indeed the obvious CA3046 which had failed, it’s a well known phenomenon with these mid-time Moogs. The keyboard once again became the main focus of this repair, it was awful, again a lot of the keys in the middle of the frame behaved intermittently and badly, a full day was spent on this mechanism trying to make the gating and CV contacts behave correctly, and even then the next day there were keyboard issues.

Skipping those for a while we had to investigate the very poor tuning reliability, all attempts at calibration were fine at the time of doing but futile minutes later. The power supply was solid, no ripple or dropouts so that wasn’t it, but testing led to suspecting the octave select switches. It turned out that the contacts in ALL the slide switches were as black as the ace of spades, and whereas some were accessible to solvent cleaners directly, the Octave select switches had to be removed from the PCB to be cleaned properly, Isopropanol and Q-Tips were good enough to clean these, as seems to be the case for most open slide switches on any synth.

So far so good, the next gremlin was an intermittent massive modulation problem which would badly affect the power supply, and modulate everything, it seemed like a dying IC and indeed it was, the clue to which one was the fact that the synth began to drone endlessly if the sustain was turned up, yes it was one of the 4007’s in the filter ADR, and a day later the amplifier ADR chip also failed. We were not massively surprised, these IC’s are very delicate and very old, and it is better that they die here than 200 miles away and back with the client. The reason for this IC suicide we think is due to the large number of power cycles that occur when a synth is being repaired, during a difficult repair a synth may be powered up and down a hundred times or more, which is probably far more than most users would do in a year, and old and weakened IC’s will give up under the onslaught.

I apologise for not having any Prodigy pics for this one, it was very time consuming, and no pictures were taken apparently. We got the keyboard working well enough but didn’t trust it for the long term, so we fitted CV and Gate Jack sockets on the back, but beware, there is a technical mistake on all the documentation for the official updates from that era, if you connect to pin 2 of U12 as the keyboard in/out as it states, when you feed an external CV in, any change in the filter keyboard tracking switch reduces the Voltage and detunes the synth and by a considerable amount, due to the 4k7 resistor in the feedback loop of the Op-Amp. Use pin 1 for the keyboard switching Jack and all will be good.

As a small aside, the power LED had failed, so we replaced it, but we tried an experiment where we used a Tricolour LED with the red side wired for power as per normal, but used the green side of the led to receive drive from the LFO comparitor so the power LED flashes at the LFO rate, the LED is on the PB/Mod panel so this wasn’t entirely out of place, and no extra holes had to be drilled. The client hasn’t complained so perhaps he likes it too, all synths should have blinky lights for LFO rates!

Jupiter 8, When Good Synth’s go Bad

Out of the blue we got an e-mail stating that a client had a JP8 which had died a number of years ago, and which he would like repaired, fair enough, no one wants to own a dead JP8. This one was fitted with a Groove MIDI interface from 1987 and he suspected this might be the issue. We have a lot of experience with JP8’s, having owned one in the past and having serviced a surprising number from this small island of ours, many people thinking they have the only one in the country and being shocked to hear that we have worked on a dozen at least.
They have a reputation for being a very reliable synth, and this is well deserved, they are very well engineered and a quantum leap from it’s predecessor, the Jupiter 4, the engineering leap between the two still staggers us.

But they are getting old, and random IC failures plague this beauty as much as it’s contemporaries, and in a similar manner to the OBX(a)’s and the Prophet 5’s, old chips will die.

Due to circumstances that occured several years ago, we have a spare CPU board, Interface board and one Module Controller board in our spares section, all of which were fully functional last time we checked them in a JP8, so we replaced the CPU board in this Jupiter with ours, and it booted fine, Autotuned and basically worked, at this time we diagnosed and fixed op-amp failures on two voice cards, one to do with the VCO modulation section on Voices 1 and 2, and another which concerned the filter frequency control on Voice 3. At this point the 8 voices all worked and sounded similar enough to each other, so attention was now drawn to the front panel, and the scratchy and intermittent sliders, and broken pots and switches. Full recalibration of a JP8 requires a fully functional front panel, it doesn’t have the test programs of the later Juno 60, and as there are about 220 trimmers in a JP8, this is no small job. Apart from being scratchy, many of the slider stems were bent and some were distorted due to being bent many times, you can’t take a pliers to these in situ and hope for a good result, so the entire front panel set of pots and sliders were taken out of their pcb’s, dismantled, fixed, cleaned and rebuilt, and new lubrication applied, as the old stuff had turned to a sticky jelly, and wasn’t smoothing anything.

And then the unlikely happened, the JP8 refused to boot in a similar way to the original problem with our CPU board in it, all scoping seemed to point to the Z80 CPU having failed, as it had with the synth’s own CPU board. We reckoned that the Sharp Z80 CPU in the original board was the main problem in the synths own CPU board so we took it out and put in a Mostek Z80 from our Sequential Tom drum machine, we will replace this with a Zilog Z80 in time but it did work, the only issue was Autotune failed miserably every time, the tuning comparitor LM311 on the synths original CPU board had also failed. Replacing this gave us full Autotune success and almost full JP8 functionality, We only qualify this because only full calibration can reveal any nasties that remain in this synth, and we do expect a couple more, dying NEC4558’s have been the main other problem so far.

The Groove MIDI interface has been unplugged throughout all of this, so it wasn’t a failure mechanism, but obviously we have to get the synth as perfect as possible before reintroducing the Groove, just in case there was an issue with this. I do have some photographs of this process, but actually nothing that looks good on a website

We are into mid July at this point in time, but for continuity we’ll continue this story here, this Jupe is a gift that keeps on giving, problems that is. Suddenly we lost Osc 1 on voices 5-8, this was due to a blown 4558 in the VCO frequency summing section on the Module Controller card (Mod Con or Con in Rolandspeak). The Op-Amp went to full voltage rail output sending all the Osc 1’s supersonic so it seemed like they were dead but were just oscillating at 50 kHz. Fixing this gave us full Autotune goodness again. A day was spent calibrating the upper 4 voices, it all looked good, most of the trimmer settings were still fine after 30 odd years but the odd ones were well awry, all looked good for finishing this the next day, but then all of a sudden Voice 3 Osc 1 had failed Autotune, and scoping it showed the Oscillator itself to have gone supersonic, we’ve seen this before many times in Moog synths (the Micromoog, Prodigy, Rogue) and it almost inevitably means the 3046 matched transistor chip has failed, theoretically no problem but for the fact that in a JP8 the two Tempco’s are tiny and irreplaceable and glued to the 3046, and due to the short leads of the Tempco’s, putting a nice gold plated DIL socket is not a viable option, so with some very careful surgery to safeguard the Tempco’s, cut out the 3046, replace it and find a good way of thermally bonding the Tempco’s back to the new chip this can be made good. We have worked on a lot of JP8’s and this is the first time we’ve seen an Oscillator failure like this, but in the grand scheme of things these failures have happened elsewhere and we are aware of the failure mechanism.

Fixing the Oscillator has proven to be more problematic than we thought, the obvious suspect (the CA3046) was not the culprit, it usually is on a Moog. It was quietly oscillating at 65kHz and doesn’t seem to respond to voltage control. As the VCO is a closed loop system with many components. The Oscillator timing cap C6A was taken out and tested, IC3A was replaced, TR1A was replaced but with no change, the NF510 FET was taken out and checked and seems fine, we will probably swap it with Voice 4 Osc 1’s FET to confirm it is good, if it is the faulty item then Voice 3 Osc 1 will tune and Voice 4 Osc 1 will not , but I doubt that will happen. All that is left are IC1 (which seems ok, voltage summing seems fine but it is a C4558) and IC4 which is the comparitor. Removing IC’s from this Voice board is very difficult indeed, it just doesn’t want to let them go and as TL080’s are not readily available, cutting them out is not an option, time and patience are required. If IC4 is at fault then we will take one from the Envelope buffer as that can be replaced with a TL081 without any consequences, as the 80’s higher slew rate in not needed in an EG, I suspect the only reason the Envelope circuits have TL080’s is because it kept inventory simpler at Roland, TL082’s are used extensively elsewhere.

The answer did turn out to be IC1, the C4558, the voltage summing half of the chip was fine but the Servo/ Integrator was obviously faulty (with hindsight) as the bad readings on this op-amp could have been the result of any problem in the Oscillator system and logically this one seemed the least likely, the only reason we changed this before IC4 (the TL080 comparitor) was because our run of luck with this synth seemed to have C4558’s dropping like flies, and indeed this was another one. As of this time there haven’t been any further casualties and the synth seems perfect after calibration. By the way, the Oscillator tuning is easy and very reliable, DO NOT mess with the Filter calibration, the service manual is very cryptic, relying on pictures of an Oscilloscope readout which is difficult to achieve even with a good scope and which in every JP8 we’ve seen are just plain wrong, the resonance settings on every Jupe we’ve seen are far stronger than the manual indicates, and while the JP8’s filter doesn’t self oscillate as standard, it could but I suspect the filter sound in response to programming would change the nature of the synth too much. This JP8 was an earlier one which didn’t have the two little switches on the Module Controller board to short out Noise and to make all the filters resonate (used by Roland later for filter calibration to simplify it in production) so the Resonance boost update was added to both Module Controller boards (shorting out the noise is simpler to do during calibration with a jumper wire). We fitted jumper links like you would see on the back of computer hard-drives, there are spare pads designed into the pcb for such weirdnesses along with space for a spare TL082 which seems to have never been needed.

After checking everything out again many times, stripping and rebuilding the output board (volume balance and Arpeggio speed) and testing again, it was time to reinstate the Groove Electronics MIDI Interface. This went well, we half expected the worst but it didn’t happen, the JP8 booted normally, tuned very quickly and behaved flawlessly including accepting MIDI in and producing MIDI out. The Groove was unusual (and great) for the time in that it sent keyboard information and program change information to MIDI out, something that has become commonplace only in recent times, but for 1987 this was forward thinking. Probably this was just a consequence of integrating the MIDI interface with the original Operating System on a very intimate level, a MIDI interface replaced the DCB interface, and the new firmware swapped DCB stuff for MIDI compatibility, the Europa upgrade for the Jupiter 6 seems the closest analogy, and Lazlo’s upgrades for the JP4, the Juno 60 and hopefully soon for the JP8 all seem to embrace this ethos, as have many other modern updates of recent times. The Kenton approach of older times of one interface fits all doesn’t seem efficient or cost effective these days, and this is probably why most have been discontinued. We have a few old Kenton interface boards here and they were hand crafted for each individual model of synth from a common template, it must have been very time consuming and expensive for Kenton to do the individual mod’s for each type of synth, not to mention writing the Firmware for them all. Kiwitechnics have taken a no holds barred approach to upgrades, some require just processor changes, some entire pcb changes but there is no doubting the quality of the supplied parts or boards even if they are not cheap, but they are good!


The above picture gives some idea of the number of Trimmers on each voice card, the pic shows voices 1 to 4, voices 5 to 8 are in a similar position underneath.


The above picture is of the the upper Module controller board, the jumper link for the Resonance boost is quite discrete but it is the small green jumper to the right of the Env1S trimmer, all wiring was done on the underside of the pcb for neatness, and didn’t affect signal flow as this was what Roland did on the later ModCon boards.

This synth is finished and back with it’s owner, and all seems to be well.