We got a Korg Polysix in on a Saturday afternoon, the client said that the keyboard didn’t work anymore, and thus we were resigned to a keyboard strip and rebuild with a couple of shims for the most recalcitrant keys that failed to respond to our TLC. Ourselves and the client had communicated during the previous weeks and he had opened it up to see if the Gate LED’s were firing, and indeed they were, but we thought he was describing the voices being triggered by the one working key on the keyboard.
When it arrived here and we got it up on the bench and opened up and switched on, and discovered that while every key on the keyboard worked (which is rare) all the voices produced a note that was a very low C, it was as though no CV’s were getting to the Oscillators, and yet -9.45 Volts was, though we suspected that this wasn’t right. Octave switching and pitch bend as well as Master tune also didn’t work, this was a bit weird.
The programmer section was fine, and still had an enormous NiCad battery in it which looked original but was also absolutely fine, good voltages and no sign of any corrosion or breakdown at all, this synth was a new production model imported at some point directly from Japan, and was in general in astounding condition apart from the current major flaw that it was currently unusable.
Everything looked to be pointing to a problem with the Key Assigner system, so that chip was replaced with a spare, but no difference, so we explored the Digital to Analogue converter section that produces the VCO CV’s but that seemed to be working fine also. After that many hours were spent ‘scoping all the relevent parts of the famous Antilog CV generator to see if some sense could be made of it all. It didn’t really help, there are two multiplexers driven by the CPU surrounded by a maze of Transistors, Resistors and Op-Amps, any one of which could have been the culprit, (and the circuit diagram is practically unreadable even in the best version the web can supply). It was time to make a decision, the two 4051 multiplexers were obviously primary suspects, so they were replaced one by one, IC31 first (no joy) and IC39 (result ! ).
Once the synth was working again, the final touches were dealt with, recalibration and the replacement of all 8 program select switches, which some previous user had danced on with great gusto.
This synth was ready in 24 hours, primarily because the client had driven 200 miles to bring it here, tying it in with a personal commitment, and if it hadn’t been repaired in this timeframe, would have required another long trip at some future date at great expense.
Internet folklore hints that Antilog Generator problems do occur but this is the first time we have seen this for ourselves, but the 4051 Multiplexer IC is notorious for several different ways of failing, and these days is one of our most frequently purchased items.
The Korg M3X ressurected
About a month ago a client dropped in a Korg M3 which wouldn’t power up, as it had a 3 pin IEC connector we assumed that maybe a fuse had been blown, or a rectifier failed. Opening up the unit revealed an enclosed Laptop style power brick that fed the whole synth, rated at 12 Volts at 5 Amps, and quite small to boot.
A quick measurement of the output showed the PSU to be quite dead, and in true Korg tradition, it had a weird sleeved central pin connector used in Laptops, (although no Laptop actually runs on 12 Volts, it’s usually 15 to 20 Volts). A quick e-mail to Korg UK got a surprising result, that PSU is discontinued and no replacement is available. It seemed crazy that a 2 grand synth from 2006 should be left so high and dry, so a long and convoluted hunt for something suitable was begun. Apart from the weird DC connector, the overall dimensions were quite constraining, which ruled out a lot of readily available supplies from a variety of manufacturers and distributors. Don’t even consider using an old Laptop supply in there, even though it might fit !, the power supply board which converts 12 Volts to everything else that the synth needs contains the same types of components as the MS2000, Microkorg and others, and will self destruct if it sees more than 14 Volts, possibly destroying the entire synth.
Because finding something suitable was proving difficult, we cracked open the original supply to see if it was repairable, if a simple resistor or even a diode had failed, it was worth an attempted repair. Checking through everything in the old supply didn’t show up any obvious failures, but it was still dead. The logical failure was the small chip that runs the entire supply, but this was a Surface Mount IC with a Hieroglyph on it’s top which defied all searching as for what it was. It is possibly a standard PWM control IC, but without the information we couldn’t take any chances.
By chance we mentioned the dilemma to a fellow Tech who buys lots of these things to replace dead PSU’s in LED lighting controllers, and he suggested CPC which is a division of Farnell. Hunting through the long list of suitable supplies led to a magic number, so we set up an account and ordered it. It arrived very quickly and we cannot conceive of a better supply for the job than this, it has an IEC connector (which the Korg has internally) and it fits very well, and weighs a lot more than the original which we hope bodes well for longevity. The only extra work required was replacing the DC connector on the power supply board with a more standard 2.1mm DC connector, as this was what the new supply came with. You have options on this, cut the new PSU output cable off and graft the original power supply connector onto it or leave that alone and make the synth suit the supply. We chose the latter as years down the road power supplies with 2.1mm DC connectors will be far more available than weird custom configurations which in fairness is what the M3 has.
The new supply is about 3mm deeper than the original, but a metal saddle secures it in place, and the replacement of the saddle screws with some longer self tappers sorts this one out perfectly.
The important details, the part number we used was PW02712 manufactured by a Chinese company ProPower, and the internet address depends on where you are in the world, for us in Ireland it was cpcireland.farnell.com.
If anyone out there has a dead M3, we hope this was of some help!.
The Oberheim OBXa
This synth arrived in for a battery change, at least that was the original premise, someone in the past had fiited a NiCad to it to replace the original Lithium but hadn’t modified the circuitry to allow battery charging, so it didn’t really work. On the plus side, no harm was done to the circuitry. Putting in a new Lithium battery and restoring the factory presets sent us further down the rabbit hole, the tuning was bad and a lot of the normal functions of the synth behaved somewhat strangely, we hadn’t had a OBXa before but we do know when something is not right.
One of the main issues of dealing with old (and especially classic and valuable synths) is trying to track it’s service history purely based on what you see in front of you, there may be new chips with dates long after the synth’s manufacture date which may provide a clue as to what may have been troublesome in years gone by, but unlike buying a classic car, you don’t have a service history with a synth. A lot of the more expensive synths have been toured worlwide by fairly well known musicians and when they gave trouble desperation kicked in and who knows where in the world immediate problems were sorted out, just surviving from one gig to the next was good enough in those days!
Basic testing made it apparent that the power supply voltages were well out of tolerance, and testing the modulation panel showed similar issues. Correcting both of these issues ended up with the result that the synth would not Autotune at all, all notes were a Fifth out of tune and all 8 lights would flash after any attempt at an Autotune. Endless ‘scoping and analysis of the Autotune state machine didn’t help, as didn’t replacing all of the IC’s in the Autotune system. After a lot of hours with the scope, we tried disabling all but one voice, and trying to make it tune. Logically for a 1981 synth, turning off all the voices but one made a nonsense of the Autotune indication system, no lights blinked but at least the selected voice was vaguely in tune.
Previously the tuning was a fifth out for each voice but also random tuning anomalies in random voices were a regular and repeated occurence, and the source of these were looking more and more likely to be the connector pins on the voice card motherboards. We took out the top tray (Voices 5 to 8) and took the motherboard out and used metal polish to clean every individual pin on all the connectors. These looked very tarnished, and polishing was done until no more black oxide could be seen on a cotton bud. This took some considerable time, and a lot of cotton buds. Putting the motherboard and the voice cards back in afterward didn’t really improve things much, but after trimming voices 5 to 8 so they were in tune with each other, they were more likely to stay together and be more consistent than the bottom tray (Voices 1 to 4), so we cleaned and polished all the connectors in the bottom tray. It really does need to be done, those connectors have been tarnishing for a long time, and poor contact between voice card and motherboard can only lead to randomness, and it’s only when you polish them, that you discover how blackened they are.
A lot of research and checking out of relevent forums yet again failed to deliver anything that was actually useful, so it was back to basics, working out how the synth was meant to work and how it achieved it. The power supply was now fine, the modulation section was thoroughly checked and rechecked and was also fine, and all of the motherboard voice card connectors were polished and pristine, all of the oscillator chips had been taken out of their sockets and cleaned and replaced, so what next ?
Back to basics in essence, we knew what was good, so we had to figure out what was not, one clue was that of all the trimmers on all the voice cards, only the initial frequency trims of the oscillators looked like they had seen some action in the past, almost every other trimmer in the machine still had the latex type compound in the middle of the trimmers, indicating that they had never had to be adjusted before (or at least never were), so we were led back to the original question at the head of this section. What was the service history of this machine? What servicing had this machine seen in the past, and what forms did the problems take?
This particular synth was in remarkable electronic condition as well as great physical and cosmetic condition, the only obvious replacement IC was a CEM3310 envelope chip on one voice, but as this synth was an early model there were a few ECO’s (Engineering Change Orders) which we implemented to improve reliability and update the synth to it’s later brethren.
The OBX-a has an 8 way internal bank of switches which enable each voice to be physically switched off, or more usefully one voice at a time can be calibrated by switching off all the others, so we calibrated all 8 voices one by one until when you played an A, you got an A, and following this and reinstating all 8 voices and running the Autotune sequence, it actually passed. Further testing showed a weird resonance behaviour on voice 3 in 12dB mode which turned out to be a faulty capacitor in the filter coupling stage, which was replaced. The last problem was voice 5 being out of tune and failing Autotune at random, which turned out to be a faulty Initial Frequency trimmer on Osc 1 of that voice which we also replaced. The interesting thing we discovered about this problem was that the Autotune system will not even correct for a Semitone which was surprising for a 16 Bit Autotune system which infers that on an OBX-a the oscillators have to be inherently quite accurately trimmed and in a stable condition, because the Autotune will do nothing to help beyond the very slightest drift.
Roll on next month…