The development and release of the Moog MemoryMoog (and MemoryMoog Plus) was the last gasp of the Moog company in the 80s. Around 1980, the two younger American synth companies, Sequential Circuits and Oberheim, were thriving, putting out one new synth after another. By the time the Memorymoog came out, SCI and Oberheim had already released multiple true polysynths. The Japanese companies were cranking out one new polysynth after another. The two remaining from the old guard of major American synth companies, ARP and Moog, were acutely aware of the serious market pressure to put out polysynths of their own. Each had already barfed out a big, cumbersome, paraphonic psuedo-poly (the ARP Quadra and the Polymoog, respectively) but it was REALLY time for them to get their act together, hire some programmers and design a true polysynth with digital voice assignment and control.
They each poured all of their resources into designing what would be their first polysynth, and their last synth. Each of them collapsed under the financial strain of the effort.
ARP was only able to produce a handful of units of its monster polysynth, the Chroma, before bankruptcy forced the company to liquidate in 1981. They sold the rights to the Chroma to Fender, who rebranded and sold it as the Rhodes Chroma. Moog released the Memorymoog in 1982, and the even more hastily developed Memorymoog Plus expanded version soon afterwards, but soon after transitioned away from producing their own synth designs and started using their factory to do contracted work for other companies.
Working on the MemoryMoog
When you look inside the MemoryMoog, it definitely looks like something produced in a time of desperation. It is pretty much a huge mess inside. There are many different types of connectors and cables flopping around in all directions. It really contrasts starkly to the neatly organized and service-friendly Oberheims and Sequentials of a similar scale.
We have worked on quite a few MemoryMoogs and MemoryMoog Pluses at this point and learned a lot about what is essential to making them reliable. It is not the kind of synth that is really worth working on unless you are going to do a lot of work. I think if someone called us up and said “I just bought a MemoryMoog that has never been serviced but everything works except this 1 thing, could you fix it for me?” we would have to tell them that we just CAN’T work on a MemoryMoog that way. An unserviced MemoryMoog is just going to continue to have one issue after another until it is fully overhauled, but after a full overhaul it can be pretty reliable.
The last MemoryMoog we worked on was a Plus and required probably the most extensive overhaul we’ve ever done on one of these.
It’s almost not even worth testing some synths until the power supply has been overhauled, and the MemoryMoog is one of them. That’s always what we do first: new electrolytic capacitors, rectifiers (discrete diodes in this case which are even more prone to failure), regulators, and power transistors, and fresh heat sink compound on the heat sink.
This one also had visible leakage from almost every single other electrolytic so it got a complete re-cap of all boards. We also mandatorily replace the Mullard “tropical fish” film capacitors here and anywhere else that we encounter them, as they are actually very unreliable at this point, often visibly cracked and functionally unstable. Finally, we replace every trimmer that is used for CV on the Common Analog and all voice boards. These are always jumpy and can cause not just difficulty in calibration but much more serious-seeming issues as well, such as extreme instability of oscillators and filters, loud crackling noises, and even voices dropping in and out. In addition to replacing all of this stuff, we normally have to clean all of the switches and key contacts too. The switches on the Memorymoog are a weird combination of metal tabs and custom injection-molded plastic parts, but luckily the contacts are pretty easy to deoxidize. We replace the 4053 multiplexers in the panel preemptively because they have a high chance of being bad and we want to reduce the likelihood of having to take the panel boards out again. Finally, both sides of every single connector everywhere are deoxidized overnight and then copiously flushed to make sure no sticky DeOxit is left behind.
Only then are we ready to really start testing and troubleshooting. One of the most important things that we gain by working on a certain model of synth many times is the understanding of what service tasks should be mandatory in every restoration based on its specific weak points.
When we finally put this synth back together and tested it, we counted a large number of issues in the voices. We’ve observed that the most common causes of issues beyond what is addressed by the steps that I mentioned above, in terms of component failures, are bad CV buffer op amps (mostly TL082 and 4558) and bad CMOS analog switching ICs (405x series, 4016). I would replace them all in every MemoryMoog if I could. Typically a few more will fail over the course of testing and working on the synth, just from the synth being used and power cycled many times. Most of the MemoryMoogs we get in have sat in disrepair and unused for years so once they’re pushed into service a lot of new issues will continually pop up for a while. Our goal is to stress the synth until the point that we get through all of that, so that all of the issues that might be coming down the pipe anytime soon happen while it’s with us, and we fix them, before it is sent home.
It only took us working on a couple of polysynths to realize that a lot of troubleshooting of voice issues can be done mostly based on looking at the schematic and deductive reasoning. Because of the particularly linear architecture of the voices in the MemoryMoog (vs an Oberheim or Sequential Prophet 5/10 with more complex modulation routing) this is especially true with this synth. So in many cases we can just observe a symptom, look at the schematic and know what IC to replace based on just what isn’t working right. Beyond that, having many other voices to compare a point on a circuit to also makes it a lot easier when you do start having to examine the circuits more closely, and the socketed ICs make it easy to swap chips between voices and see if the issue moves with the chip. While as a troubleshooting method it may seem more “correct” to trace signals through with an oscilloscope, and this is always an option, it’s worth recognizing certain cases when this approach might not the most efficient, or necessary, and we might as well just swap chips since we can!
We were cruising along pretty well with this particular MemoryMoog Plus and then all of a sudden it stopped booting. Digging through the digital side of the MemoryMoog (Plus especially) is a whole other matter and is truly miserable. The fact that it was forged in an environment of frantic disorganization is perceivable both in the construction and layout and the design itself. One externally visible symptom of this is that the LEDs all flicker constantly in the Plus because the CPU has too many tasks to do in each cycle before it comes back around to “update” them and “tell” them when they should stay on. Many synths divide their tasks between a master and slave CPU so that no CPU has to work so hard. Presumably the developers of the MemoryMoog didn’t realize they would have this problem until they were in too deep to be able to restructure the entire digital control side to use two CPUs.
Many of the chips were receiving NO addresses and I wasted a lot of time going down a lot of dead ends, with things complicated by the fact that I did not then have the schematics for the “Plus” board (where the CPU resides in the Plus version) so I was flying blind. I tried replacing the CPU, address decoders, validating the EPROMS, and a bunch of other things before I realized the problem was something really stupid– the giant IDC connector between the main CPU board and “Plus” board was starting to come uncrimped. Sometimes trying to troubleshoot thoughtfully doesn’t help much when the reason for the issue is something so stupid.
Soon after one of our clients gave us this amazingly huge binder of official Moog service manuals though, so next time I have to debug a freaking-out Plus board I’ll have the schematic at least… can’t wait.
The last step is normally calibration. The tuning of the Memorymoog is a computer-assisted but awkward process due to the synth’s size and shape; the intonation status of each oscillator as you’re tuning it is given in hex code on the little bubble display readout under the main display, but the synth lid has to be open while you’re tuning it, so you have to put a mirror in front of the top/front of the synth and view them in that, upside down and mirrored. While calibrating, it’s helpful to keep the voice board you’re tuning as warm as possible, or when you close the synth up, the change in temperature conditions on the board can easily shift the tuning. We drape a towel over the boards to at least trap the heat they themselves are making, but this still doesn’t really compare to the heat generated inside the synth when it’s closed.
In this one though, the client requested that I replace the fan, and I did this before calibration. The original fan is really heavy and loud, and I’m no fan expert, but there’s also no way it could even work well as a fan… since the back is covered by the motor, it can mostly only swirl air around in a circle around the outside, not pull it through. The new fan is a desktop PC fan which is light, quiet, and makes the synth actually feel cooler when it’s running, which will also improve its tuning stability.
It may seem like I’ve been throwing a lot of shade on the MemoryMoog. I think any synth tech will agree that it’s definitely not good from a construction/engineering perspective, but I can’t deny that it sounds HUGE. If that’s your thing, you probably can’t top it for hugeness of sound without getting an actual pipe organ!