The Moog MemoryMoog

We hear the same thing all the time:

“You work on memorymoogs? I hear they can never be in tune and are really unreliable..”

Okay, maybe not that exact sentence, but 3 or four sentences into a customer seeing a memorymoog, that’s the jist of it…

But in our experience, they’re a joy to own and so far none of the memorymoogs (over 30) that we have restored have come back needing significant return service, another tuning, etc… So why does this synth have such a bad reputation? What makes a person who doesn’t know a lot say “well I hear you need the memorymoog PLUS for them to work. Or the Lamb memorymoog…”.

Well, maybe it’s the poor info all over the net…. But besides that, perhaps maybe the synth tech world (besides Rudi Linhard of Lintronics) wasn’t/isn’t hip on figuring out how to “tame the beast”. Let’s dive a little deeper….


A lot of the history behind the memorymoog is not quite available, and how it came to be and went to market can be vague at best, but I’ll do my best to summarize the facts that I do know about the history, and dispel some common rumors.

The memorymoog was designed by Moog Music after Bob Moog had left the company. The does expressly mean that the Memorymoog was not a product designed or produced by Bob Moog. In fact, it was made by another group of designers headed by Dave Luce, who was the President of Moog Music and one of the leading designers of the Polymoog. Of course the tradition of Bob Moog’s most critical design was implemented into the memorymoog, the ladder filter. Otherwise, the memorymoog is designed around a CEM and CMOS chipset, which is not to be knocked on… read on.

The memorymoog was more or less rushed to the market. This can be noted by the fact that for one, it uses the polymoog’s “faratron” power supply, and that the board interconnects leave so so so much to be desired. The memorymoog was also subject to many field changes in which the schematics were never really updated. In a way you could say it was ahead of it’s time, as today designers rely on releasing broken products just to offer “updates” to the end user via software updates. The original memorymoog (not plus model) initially suffered from very bad tuning stability HOWEVER most of these memorymoogs did receive the auto tune update, and for that reason alone as very comparable, and often more desirable than a plus model. Moog did also end up releasing a sequencer/midi retrofit, which was an optional update for the standard memorymoog, and a stock feature of the plus model. This should dispel the myth that the plus is more reliable than a standard model that contains the auto tune update. In other words, the plus simply means two things: 1. The auto tune update was installed at the factory (though this is true of many non-plus memorymoog’s) and 2. Has the sequencer/midi board from the factory (which was of course an optional upgrade for non-plus memorymoog owners).

As a result of the rushed design, the memorymoog did contain some miscalculations and crash-prone firmware. To fix this, the memorymoog was essentially revamped by Rudi Linhard of Lintronics, who created what he calls the LAMM memorymoog. LAMM stands for Lintronics Advanced MemoryMoog. The LAMM was only installed by Rudi in germany or by Bob Moog’s new company Big Briar. Soon after though, the LAMM could and can only be performed by Rudi in germany. The LAMM fixes some miscalculations in the electronics, updates the firmware, changes 100’s of failure-prone parts, updates the midi capabilities (greatly), adds stereo outputs, and ultimately brings the memorymoog to the highest level of reliability.

The “poor man’s LAMM” is the DBM upgrade, which offers only better midi functionality and a new firmware. It’s only available for the plus model and is worth having. Installation is straightforward, luckily.


The design of the memorymoog is perhaps the only “weak point”, but not in the way you might have heard. Let’s touch on the bad first.

Touching on the board interconnections mentioned earlier, the memorymoog does have one of the most ridiculous board bussing system ever. However, I wouldn’t jump at it to say it’s the worst design possible, but I would happily call it haphazard. Let’s compare it to the Jupiter 8. The Jupiter 8 boards contained, generally, a power supply header with redundant pins, a standard wire connection system, and then 1 or 2 sets of high pin count IDC connectors. These were made very well, and make for a very good design that is easy to work with. In the memorymoog, there are 5+ 16 pin DIP style IDC connectors that then terminate into DIP IC sockets, which is sort of a weak connection at the end of the day, along with some AMP style connectors (not bad). The power supply hard wires into the DMUX board, and is haphazard in my opinion. Removing a board from the machine is a challenge to say the least, and putting a board back in after a rebuild can be very difficult if you weren’t careful enough to mark the connectors and also take some pictures. It’s very easy to put connectors in the wrong place, backwards, etc. Unfortunately, harm can be done if you’re not careful in that aspect. It’s not so forgiving, but also not very punishing.

It’s common to read online that a memorymoog must have the 16 pin DIP sockets and IDC headers changed to gold type. In general, doing so is not canon in my shop, but if a connector is very corroded, we will happily replace them with higher quality parts, as we would with any other synthesizer. We do, however, regularly opt to clean the connectors with alcohol and a pencil eraser. We’ll also tighten the DIP socket if it’s feeling a bit loose. It’s just good measure, though.

Additionally, the digital and common analog boards are on hinges above the DMUX board, and the resistors and IC’s are not laid out in an intuitive sequential manner. This, combined with the many 16 pin ribbon sets strewn about, can make the memorymoog difficult to fix because it’s hard to find what you’re looking for. Further, the silkscreen is usually a bit smudgy, so finding R76 can be hard when you’re not sure if it’s R76 or R78 or even R79 or R70…. The 6 voice boards are also not only hinged, but voices 3 & 4 are upside down and backwards (!!). This makes calibrating a total pain in the ass.

In order to calibrate the memorymoog, you must use a mirror and your mind to learn to read a hexadecimal readout upside down and backwards. This is a horrendous strain on your neck and sanity. The LAMM memorymoog offers a “mirrored” function, which is a god send. Doesn’t make the mirror aspect more appealing, though.

Moving on to facts and good design, the memorymoog does at least offer a very good and well thought out set of schematics, the service manual is complete and informative, including information even on the machines flaws.

The machine contains a computer that is having the crap beat out of it at every instruction. It’s not uncommon for the LED’s to subtly blink along with the LFO, and other oddities caused by the overstrained computer. It does work though and is straightforward to repair when needed. The remainder of the machine is based on CMOS switches (like all other poly synths) and luckily most/all are in sockets, making for quick swaps for troubleshooting.

One interesting design aspect is the intense use of the CEM3360. In most other synths, many functions and controlled by assortments of op amps and 4000 series logic chips. In the memorymoog, it’s mostly CV’s controlling a big array of 3360 dual VCA’s. Back then, this may have been cost effective, but in today’s age means that a non-working pitch bender or modulation wheel is solved by changing a rare and expensive chip. [Update, the chip isnt expensive anymore…]

The voices are based on the 3x CEM3340, which unlike many polysynths using the same oscillator chip, the waveforms are not shaped before the switching section, leaving the oscillators pure raw curtis oscillators. This is far from a bad thing though. The memorymoog sounds huge, and it wouldn’t sound bigger if it used waveshaper circuits. From the Oscillators, they get summed into a crude mixer that distorts in a very flattering way. From there, to the coveted moog ladder filter, and finally into a nice OTA based VCA and then to the output jacks. Throughout the voice board are CD4016’s and other 4000 series chips allowing ample voice modulation capabilities. All of these are controlled by CEM3310 envelope chips. Since I am only a tech, I could not tell you exactly how moog was able to make these envelopes so snappy fast using that chip and such a strained CPU, but it is the best envelope ever implemented in my opinion, and blows away the other 3310 based synths in regards to the envelope. It makes no sense to me.

The build quality from an exterior aspect is very good. The pots are a good quality, the buttons are tactile, and the control layout is a dream. It invites you to fiddle 🙂

Making A Reliable Machine

And finally the most important part…. how does one make these machines so reliable that a geatslut’z head would implode in disbelief? Well, the ingredients of that secret sauce will be kept secret, because after all, this is how we make a living, and we also don’t want owners who think they’re handy with a soldering iron to go in and wreck their memorymoogs. These are at the end of the day very complex machines that require a careful and experienced hand to work on.

The fact of the matter is only 1 legitimate way to have a well working memorymoog, and that is to rebuild from the ground up. A memorymoog is not for every tech… we recommend the following techs for memorymoog work: 
Us (duh), Jareth Lackey, Bell Tone Synth Works, Analogics, Retro Linear, & of course Rudi Linn…