[sdiy] Memorymoog Overhaul Suggestions

[Michael E. Caloroso]

James R. Coplin wrote:
> I'm having to open up my Memorymoog to install a Kenton MIDI kit I have and
> replace some out LED and burnish a dodgy switch. I figured while I was in
> there I might as well do a complete overhaul. Here's what I'm considering,
> are there any other suggestions? I'm pretty resolved to be buried with this
> unit so I'm pretty open to making as good as possible.
> 
> MIDI kit install

I had a Memorymoog in for repair with a defective Kenton MIDI kit.

Turned out it wasn't the Kenton.  The case had shorted out against the 
vertical mounted resistors on the Kenton board when it was closed.  When 
the case was open it was fine.  Fish paper fixed it.

> Convert all LEDs to blue (Any recommendations on whose I should use?)

Frivolous - I have yet to have an LED blow on mine.

> Recap the power supply

Good idea.

> New ribbon cables throughout

YES!  This and the trimmers are the Memorymoog's Achille's heel.

> 
> Replace all those crappy trimmer pots on the voice cards (Again, whose would
> be best?)

Rotate the trimpot 180 degrees.  This is a temporary solution until you 
need new trimpots.  This will require a complete calibration of the 
voice cards though.

I'm not a fan of ten turn pots, no matter what brand you get they are 
too sensitive to mechanical shock for such a small area of resistive 
element that the wiper is on.

I have yet to experiment with this, but my idea was to mount standard 
single turn trimpots on a subboard underneath the keyboard.  That way I 
can calibrate the MM without having to translate &$*%^% upside down 
calibration figures on the alphanumeric display.  To use single turn 
trimpots you would also need selected bias resistors to get the trimpot 
in the right area, IE 1K trimpot plus two bias resistors on each leg of 
the trimpot should equal the original 10K multiturn trimpot.

Again, have not tried this yet.

> Refinish case. It's perfect except for where I've worn through the finish
> under the number pad! 
>
 > Change out all the OpAms if possible (Suggestions? The chips are all
 > on sockets which is nice)
 >
> This thing has so many opamps in both the cv and signal path that I'm really
> hoping to get improved stability and sound by upgrading them. I had this
> done to my 2600 and I was floored by how much better it sounds. If only it
> could be made all discrete! So, what else should I be considering?

The MM uses better opamps in the audio than was available when the 2600 
was designed in 1970.  I doubt you'll gain much upgrading opamps, better 
to replace the crap IC sockets where critical CV is processed.

I did a major hotrod on my Memorymoog when it was too sick for live 
gigging.  I identified all the connectors that carry CV and replaced the 
tin plated connectors with gold plated ones and likewise with the ribbon 
connector sockets.  None of the plugs were replaced, only the sockets 
and pins on the circuit boards.  I got a huge improvement in tuning 
stability and in reliability, and my MM has been kicking like a top 
since 1996.

I detailed all my hotrodding on AH, here's the (long) post for the archives:

***************************

To: 'Analogue Heaven' analogue at hyperreal.org,
From: "Caloroso, Michael E" analoguediehard at att.net,
cc:
Subject: [AH] DIY: Analog synth (Memorymoog) tuning/connector fix (long)
Date: Mon, 19 Jul 1999 14:42:46 -0400

I've had numerous inquires on the details of this, so I've decided to 
lump all of it in one post for the archives. The focus is on the 
Memorymoog, although this info applies to any analog synth.

----------------------------------

The Memorymoog has a history of problems with tuning. If your unit has 
to be calibrated more than once a year, or the calibration values get 
noticeably worse each year, or the tuning changes from day to day, or 
the monophonic mode is out of tune with the polyphonic mode, or the 
tuning changes with any movement or impact (just hold a note and hit the 
case with your fist), that is the result of oxidation on the connectors 
inside the Memorymoog. There are almost 100 connectors inside these 
beasts, which aid in servicing but are a frustrating source of 
malfunction. In my experience, I've found that the most reliable 
equipment either have the fewest connectors or the connectors are high 
quality.

The connector contacts in the MM are tin-plated. Over time, oxidation 
forms on the contact surfaces in the form of an impedance, which 1) 
varies with humidity and temperature, and 2) is non-uniform over the 
contact surface, where simple movement changes the impedance. The 
consequence is that you lose voltage across these oxidized contacts.

The polyphonic control voltages that drives the voices cards travels 
through three boards (six connectors), while the monophonic control 
voltage goes through four (eight connectors). If *ONE OR MORE* of these 
connector contacts has oxidation, the control voltages are degraded. 
Note the emphasis; at worst there will be six points where the 
polyphonic CVs are losing voltages across the connectors. This is the 
primary cause of the MM's tuning problems. To add to the problem, these 
critical voltages are carried over ribbon cables via DIP16 plugs, and 
the mating sockets in the stock MM are poor quality for analog signals. 
You can verify this with a DMM; probe the signal at the DUMX board and 
at the voice card where it terminates, if you measure any voltage drop 
then you're seeing contact loss.

Power busses are equally vulnerable, if the voltages aren't stable no 
amount of calibration will help. TTL logic signals (d*g*t*l) with their 
built-in interim zone between high & low are insensitive to these losses 
and generally don't have a problem.

As a temporary fix, the contacts can be cleaned by periodically 
removing/re-inserting the plugs as to "scrub" the oxidation off, or they 
can be shined up using an eraser.

Ultimately the contact surfaces will corrode beyond cleaning. I am the 
original owner of my MM, and it took less than ten years for this to 
occur. I was now faced with having to replace connectors.

During that time, I had acquired an ARP Pro-Soloist which predates the 
MM by ten years. It had laid dormant, yet was perfectly in tune and 100% 
functional when I powered it up. This is an instrument that is over 
twenty years old yet at half its age was in far better health than my 
MM. My interest piqued. Closer inspection revealed that the connectors 
used in the ARP were a combination of tin plating and gold plating, IE 
the pins were gold plated while the contact inserts in the mating plug 
were tin plated.

Before starting ARP, Alan Pearlman designed and built amplifier 
assemblies for the space program and the military. I also have many 
years of experience in military electronics. The military demands 
reliable equipment, and the standards mandate that all connectors 
contacts have gold-plating; tin plating is not acceptable. Al's 
experience with military electronics carried over to his synthesizers 
and is a big reason why his instruments were more reliable.

Connectors with tin-plated contacts are rated for ten years, while 
gold-plated contacts are rated for TENS of years. Gold also does not 
oxidize.

My first impulse to fix my ailing MM was to replace *every* tin-plated 
connector contact and pin with gold-plated equivalents, but there was a 
problem. The gold-plated contacts for the AMP MTA plugs used in the MM 
couldn't be found in any catalog and were available only from AMP in 
lots of 10K, and AMP only produced a stock when the demand justified the 
expense. You also need a special insertion tool for the MTA plugs.

However the ARP demonstrated that you can get years of reliability at 
reasonable cost by combining gold/tin platings on connector systems. So 
I decided to focus my efforts on replacing only the connector pins and 
DIP sockets on the PC boards and leave the plugs as is. This was more 
cost-effective in parts and in labor, and the parts are easier to 
locate. I also minimized my work by targeting only those connectors 
which carried any analog signals, since logic signals weren't a concern. 
You can find this info in the master connector chart on the MM's 
schematics, which you'll need for this work. If you need schematics you 
can get them from the world famous Mark Glinsky's manual manor at 
http://www.magicnet.net/~mglnsky/msg4.html.t years of reliability at 
reasonable cost by combining gold/tin platings on connector systems. So 
I decided to focus my efforts on replacing only the connector pins and 
DIP sockets on the PC boards and leave the plugs as is. This was more 
cost-effective in parts and in labor, and the parts are easier to 
locate. I also minimized my work by targeting only those connectors 
which carried any analog signals, since logic signals weren't a concern. 
You can find this info in the master connector chart on the MM's 
schematics, which you'll need for this work. If you need schematics you 
can get them from the world famous Mark Glinsky's manual manor at 
http://www.magicnet.net/~mglnsky/msg4.html. They're a far better image 
than any online schematic.
They're a far better image than any online schematic.

I also nailed the CEM3340 VCO chip sockets on the voice cards, just to 
be safe, because the sockets carry critical control voltages.

----------------------------------

For the MM, you'll need to find a source for header strips with 
gold-plated pins on 0.100 centers and 16-pin DIP sockets with 
gold-plated contacts. These are not hard to find in catalogs or online. 
I got lucky and sourced gold-plated parts at a surplus shop. I recommend 
Augat DIP sockets if you can get them. You can buy header strips with 40 
or so pins which can be cut into the quantity you need. Side cutters 
work well, although occasionally you'll lose part of a pin. Buy more 
than you need. All together, you'll need 42 DIP16 sockets, more than 
enough header strips for 305 pins, and header strips with right angle 
pins for 32 pins (you might get away with the straight pins but I don't 
know how it will clear when put back together).

If you can find them (and let us know where 'cause I would really really 
like them for mine), you'll want new headers for the power supply board. 
These are the larger pins and are 0.156 centers. You'll need a 15-pin 
strip and three 3-pin strips. I have yet to find a catalog (Mouser, 
Digikey, etc) that carries these.

The complete list of connectors to replace is:

Voice Cards A-F (x6): S11, S12, P13, P14, P15, P16, P17, sockets for 
CEM3340s (x3)
Common Analog: S21, S22, S23, P25, P26, P27, P29
Contour/Glide: S32, S33, P31, P34, P35, P36A-F (x6)
Digital: P40, P48a, P48b
DMUX (big one!): S53, S54, S55, S56, S57, S58, S59, P511, P512, P514, 
P515A-F (x6), P516, P517, P518, P519, P520
RSC: P63, P64
LSC: P76, P77, P78
Power Supply: P131, P132, P133, P134
Keyboard: 16 pin right angle

Note that the LAMM update from Lintronics does not include this work.

If you have an older MM, you may need the reliability updates that Moog 
Music has issued, the complete list of service bulletins are online at 
http://www.ne.jp/asahi/interlogic/oku/memorymoog/mod.html,es that Moog 
Music has issued, the complete list of service bulletins are online at 
http://www.ne.jp/asahi/interlogic/oku/memorymoog/mod.html,

and Cary Roberts has clearer images of 840/841/842 at 
http://www.memorymoog.com/docs/

----------------------------------

And now, words for the DIYer:

This is not a job for the novice. You should have excellent 
soldering/desoldering skills and experience with double-sided PC boards. 
For an online primer see 
http://aupe.phys.andrews.edu/diy_archive/references/soldering.htmling/desoldering 
skills and experience with double-sided PC boards. For an online primer 
see http://aupe.phys.andrews.edu/diy_archive/references/soldering.html 
and http://www.synthtech.com/tutor/tutor1.html
and http://www.synthtech.com/tutor/tutor1.htmlhtml


The task of removing DIP sockets and header pins is tedious and can 
damage the PC board if you're not careful. A good soldering iron is 
definitely recommended. You'll need desoldering tools (wick or 
solder-sucker) to remove the parts. In most cases you'll have to 
physically destroy the DIP sockets to get them out. If you've never 
removed parts like these before, start with the voice cards since they 
are single-sided boards.

Before you install the new parts, apply flux to the pads so that you get 
a clean solder joint, and clean up the excess flux with flux cleaner. I 
use a brush to apply flux cleaner and a wadded paper towel to soak up 
the residue, using a towel fresh surface each time. Good lighting makes 
a difference here. If you don't clean it up, leftover flux will cause 
problems down the road like low-impedance shorts across PC board traces.

Wear a ground strap or do your work on a static-safe mat. You're going 
to be working with irreplaceable static sensitive components. The CMOS 
and CEM chips are safer in their sockets on the boards.

I used a Panavise with a PC Board clamp so that I can work with both 
hands free, which helped a lot. Some of the MM's boards are too big to 
fit this clamp.

I suggest that you work with one board at a time and confirm 100% 
functionality after rework before you move to the next board.

I also suggest that you label plugs with a permanent marker as you 
remove them, IE S11E or P16E for voice card "E". There are too many 
places to plug the wrong connector and too many connectors to get mixed up.

Remember that when you install a new part that you have to have proper 
solder fillets on the pads of *both* sides of the double-sided PC board. 
If you're good enough with the iron and apply the solder to the pin and 
not the tip of the iron, the solder will flow well enough to make a 
fillet on the other side by itself. Note that only the voice cards are 
single-sided.

Some connectors have a "key" so that you don't orient the plug the wrong 
way. The mating header simply has a pin (and PC board thru-hole) omitted 
where the "key" is. In you new part you can pull out a pin by heating it 
and pulling it out of the header strip as the plastic softens.

Do not handle any of the gold plated parts with tools. The plating on 
the pins is extremely thin (10 or 30 microns) and can be destroyed if 
you scrape it with metal tools.

----------------------------------

Here's another MM foible: failing range/scale trimpots on the voice 
cards cause tuning problems. On the resistive elements of the trimpots, 
the precise area where the wiper contacts corrodes over time and will no 
longer work reliably.

You can unsolder each trimpot, rotate them 180 degrees on the PC board, 
and resolder them. By rotating them you will have to set the resistance 
to its previous value, but it will be no longer around the corrosive 
area. You will have to perform major oscillator re-calibration when you 
do this, because they will be WAY out (hint: don't use the C-7 routine. 
Calibrate the oscillators by ear with a reference).

Of course you can only do this once. After that it's time to spring for 
36 new trimpots.

----------------------------------

Altogether this can take over 40 hours of work, but the reliability 
improvement is worth it. In the two years since I have done this to my 
MM, *none* of the tuning problems have surfaced. I recently checked the 
calibration, and the oscillator range is slightly off but the scaling is 
*rock solid*.

***************************

MC