The Italian Invasion

December was interesting in that it threw up a few Italian synths, no not home keyboards but real actual synths.

The not so surprising thing was that they all had an Elka connection, after the Synthex it seems they hadn’t given up on dreams of World Synth domination.
The first of these was a GEM S2 whose power supply (an Italian made switch mode) had self destructed leaving a pile of dead semiconductors and another lot of burnt unrecognisable resistors, and a capacitor that had burnt off it’s plastic case making it unidentifiable. All Google searches revealed absolutely no information on this supply so we were somewhat stuck. The client bought a GEM S3 on Ebay to replace it.

Out of the blue an S2 appeared for sale on, an Irish trading site, with obstensibly a similar problem, but from the description not as catastrophic. With the hope that the components on the power supply might be in better condition, we bought it !. A bit of a leap of faith it was, with no comeback if it was a bad choice, it took two weeks before we found time to open it up and check it out.
It was opened up, and there were no burnt bits on the power supply, so we checked out all the resistor values and kept notes, and checked out all the semiconductors, and they all seemed ok, so we put it back into the S2 and switched it on. And it worked for a short time until the display faded and died, but it still played notes and responded to programme changes without problems, interesting !.

The Italian Kurzweil

The S2 turned to be a very pleasant surprise in a variety of ways, and the reference to Kurzweil in the heading is not meant to be facetious, the K2000 and the S2 were both released in 1992, and despite the differences in their geographical origins, they have a remarkable number of things in common, not least of which is almost the exact same chips in it for the processor, engine and DAC’s. It is also as well built as the K2000 and has the bonus of posessing that holy grail of keyboards, Polyphonic Aftertouch. Both the K2000 and the S2 also use Fatar keyboards and they feel identical apart from the aforementioned aftertouch. They also both have largish Graphic displays of the same size, but the S2 has a far better backlight system which won’t die after 2000 hours. It has a CCFL backlight like Laptops used to have (before LED backlights) which is still bright and clear after 20 years.

The S2 and K2000 also both have Digital resonant filters a year before Korg finally cracked it with the Trinity. Altogether a remarkable piece of design, but internet folklore (not wiki) says that the S2 was Elka’s last project before being bought by GEM. If this is true, then the Elka engineers were truly at the state of the art.

Internet folklore also says that the S2’s Piano’s and drums are not very good, and we won’t argue with that but it also says that Brasses and Strings are very good which we would agree with with. In fact select one of the Brass presets and use the Poly Aftertouch and you are transformed into Bladerunner soundtrack territory. It is a beautiful thing. Other Internet folklore regards the S2 as one of the best Digital synths to try to replicate Analog synths and that seems fair, it even has Synthex waveforms in its ROM.

The S2 doesn’t have Kurzweil’s VAST but it concentrates on what most synthesists would be familiar with, three envelopes per voice each with 10 stages, versatile modulation options which most Analog owners would feel comfortable with, and 32 oscillators to assign as you wish (as opposed to the K2000’s 24). The S2 also has a versatile 16 track sequencer with most of the functionality of the K2000’s. The S2 has 2 complete sets of MIDI sockets on the back (as in 6 sockets) which ties in to the S2’s subtitle of Music Processor.

There are downsides to the S2 however, one of which is that despite having a nasty NiCad battery on the motherboard to back up presets, it didn’t have any RAM as standard, you had to save all your work to floppy disk or it would be lost on power down. The S2 Turbo version (which is what we have and indeed most are) had sockets for RAM chips on it but not populated. As the years have past, very little useful information is available on what RAM chips the Turbo would like to see. At least the Turbo expansion supplied a synth full of ROM presets that you could gig with, unlike the original which required a floppy load from a factory disk.

Thanks to an e-mail from an S3 owner, we have been reminded of an important detail of the batteries in the GEM S2/3 synths, the nasty Ni-Cads were only installed in the Turbo versions, the non-Turbo S2/3’s used a Lithium coin-cell to back up the clock and the global settings, so if you have a non-Turbo synth there is no need to worry about the battery until you lose the time or your Master MIDI settings, and even then no harm will have been done.

The onboard battery which runs a real time clock chip (not joking) was a bit leaky, and right next to the battery is the display connector and a chip that drives the display so when the chance, and time is available we will try to sort this one out for good, (and find suitable RAM, 4Mb worth)..

As an aside, the rotary encoder was very iffy as is commonly mentioned with regard to the S2, but a complete strip, clean and rebuild of the encoder sorted it right out. The encoder dial actually floats on 3 ball bearings which sit in hemispherical recesses in the plastic moulding. What attention to detail!..

This is an footnote added in early April 2013, as we have done previously it is here for the sake of continuity.

We obtained SRAM for our S2 after a convoluted search for information on the Internet and spotted a photograph of an S2 with SRAM chips in it. Reading the part number from the photograph made us smile a little as it’s part number was an almost unbelievably long string of digits for a memory chip. It is BS62LV4006PIP55.

We entered this string into Google, not expecting too much but it lit up with dozens of vendors claiming to sell these IC’s, hosting the datasheets and anything else BS etc. related. To cut a long story short we obtained them from a Chinese company (although they were shipped from Singapore) for €50 including shipping.

The goodness that we received was 4 of 512k X 8 SRAM chips which adds up to 2Mb of Ramdisk, which according to the user manual was the maximum that the S2 OS could actually deal with. IC’s of a capacity to actually give 4Mb of static RAM are unobtainable in 32 pin DIL format so the 4Mb limit is a moot point.

We put an 80mAh NiMH battery in there to back up the clock and obviously the new SRAM and this also fixed the weird display problems we were having, the exact reasons for that are a bit of a mystery, but it definitely is a permanent fix, although hindsight is pointing to the presence of a middle Ground layer in the pcb that was a ground point for the display, and the new battery and soldering involved remade the poor ground connection. Our S2 is also running fine on the repaired power supply that started this whole adventure.

This took a lot of work and about 80% of the parts in it had to be replaced for the reason that if Component A has failed, then the component it is connected to has to be a suspect, and as switch mode power supplies will self destruct horribly if there is any fault, any and all suspect components were replaced (and upgraded as well, sometimes it is cheaper to buy better components than to buy the exact originals).

When we find the time, we will set up a resource page for this synth. a large sound library is available online and we have the schematics for this synth courtesy of a fellow S2 owner (thanks Rob!)

GEM S2 Schematics

The sound library available online consists of well over 100 floppy disks which sounds marvelous but getting them from Internet to a disk that you can insert into your S2 is a little more complicated than it used to be.

Very few computers have Floppy Disk drives at all now, and even if you do, Windows by itself cannot write an S2 disk on its own as the S2 disk format is completely alien to a Windows (or Macintosh) machine.

A piece of software called Omniflop is required to replace Microsoft’s floppy disk drive driver which enables Omniflop to at least interrogate a strange diskette and try to recognise it. Our experience is that it is best to format a new disk in the synth of interest and then introduce it to Omniflop which then has a better chance of working out which synth you wish to create a disk for, and if the floppy drive in your synth doesn’t work, then the idea of loading new disks is irrelevent.

The Elka 22

This synth arrived from the other side of the country with a peculiar fault, when a certain key was pressed it would sound, but only once, and the panel switches would stop working. With no service documentation available we thought that we were in for a tough diagnosis. As this was another of those E-bay claim resolution cases, we looked at this one over the Christmas holidays.

The Elka 22 is in essence the Italian Oberheim Matrix 6, it uses the same voice chips (CEM 3396’s) and all 6 voices and their sample and hold’s and the chorus unit and the output stages are all on this board. The other main board is the CPU board, and quite a powerful CPU system it is, no shortcuts or skimping with this one. All the chips on all the boards are in sockets which makes chip swapping much easier when troubleshooting faults, but the downside is chip creep, where over the years and all the heating and cooling cycles the chips gradually lever themselves out of their sockets leading to strange intermittent failures in the instrument.

The synth also has nice Fatar weighted keyboard with mono aftertouch, whereas the Matrix 6 did not.

In essence the Elka 22 is very well designed and built using top quality components throughout, with good quality ribbon cable connectors (like PC floppy and hard drive cables) linking all the boards together..

Getting back to the diagnosis and repairs, we dismantled the synth and examined all the connectors and the keyboard connector was indeed not as tightly plugged in as it could be, as were a few of the others. Many of the IC’s were indeed creeping out of their sockets, and all of them were crunched back tight into their sockets. A lot of the 8 pin IC’s creep sideways as well as upwards, and these have to be completely taken out of their sockets and then reinserted.

After all this was done, and the machine reassembled, it was switched on and all seemed to be good, the whole keyboard worked (repeatedly) and all of the front panel switches also seemed fine. The only part to have failed was the Volume slider whose carbon track was completely worn out. Strangely enough the Roland JX8P and possibly the JX10 used the same sliders, and we have seen these completely worn out. This is unfortunate as the originals are almost completely unobtainable, and no one has cloned them yet. Somewhere in the scrapped parts section we had an old same type track and wiper assembly which went into the Elka 22 to fix the problem.

The synth sounds great, very Analogue in the same good way as any of the great hybrid synths, and is good to go.

The Vox Continental 71

This machine is one of the last of the Jenning’s Vox Continentals, it just has the name on a metal plate, but was completely manufactured in Italy. It still has the reverse colour keyboard but it is plastic rather than the wooden keys of the earlier Connies, and the Germanium transistor dividers of the originals are replaced by an early integrated circuit, the SAJ 110, which took the master frequency for each note in a octave and divided it 7 times to supply all the Octaves and Footages for the entire organ. That being said, it seems to be quite rare, possibly no more than a few hundred were ever produced. and we have had 2 of them!

connie71-2-smlThe above Connie 71 is one we rebuilt a few years ago, the 6 switches in the centre of the console select one of 5 preconfigured presets, while the 6th and right side switch selects the drawbars..

It’s not the worst built keyboard ever, in fact the PCB’s are well laid out and securely fixed to the instrument, and the electronic design is also more than adequate for it’s time, but, and there always is one, that old Italian bugbear of poor metal plating on all the connectors. Tin and nickel that just peel off allowing corrosion of the connectors. The interior of both of these Connies was and is a complete mess of a whitish powder from the salts of tin and nickel and silver from all the connectors that covers the interior.

Internet comments from one of the organ repair sites a few years ago said the Connie 71 was unfixable and it was foolish to try, but it worked once when new, why couldn’t it be made to work again?

The main problem is going to be the connectors, each circuit board has what are called Turret Pins, just simple silver plated thick and shaped pins which allow an Italian Chiri connector to be plugged on top of them.

A nice simple and effective connection system if the metals in the connections remains good, but they don’t.

The solution, lose them all, remove and replace all the Turret Pins on all the boards with new ones, and then cut off the Chiri connectors and hard solder them to the new Pins. This is not the future servicing nightmare that it may seem, as the only PCB likely to give any future problems is the tone generator board (there are obviously 12 of these) which only have 4 connections each, no problem for anyone with a soldering iron.

The other main issue is that of Carbon Composition resistors on the PCB’s, which seemed to be distributed at random and in large quantities across all the boards., I assume that Carbon Film resistors were always the first choice, but getting sufficient quantities of hundreds of different components in quantity in the 1970’s was far more difficult than it is today. We have seen many Minimoog’s with a similar issue. In essence put a second best component in there rather than have a production line not produce any finished boards.

Carbon Composition resistors date back to the dawn of electronics, they are quite literally a bit of ground up Carbon compressesd into a resistor shaped package and then measured to see what resistance resulted, whatever resistance was measured was then marked on the resistance package and sold as what it seemed to be. They are awful, they have poor tolerances, their temperature coefficient is dreadful, and every week their resistance increases until they go well outside of all respectable tolerances, and they are horribly noisy..

This was the next foray on the Vox Connie 71 front, All Carbon Comp. resistors on all boards were replaced with modern 5% Carbon Film resistors, a much better component, Carbon Film resistors have a Ceramic core with an outside film of conductive carbon film on the outside which is cut by Lasers to a high accuracy, and measurements confirmed that all the new resistors were within 1% of their stated values. We bought 0.5 Watt resistors as the normal and perfectly workable 0.25 Watt resistors would look ridiculously small on the generously proportioned PCB’s.


The above picture shows the progress on some of the many circuit boards in the Connie, the board to the upper left is the output mixer and amplifier. All resistors on this board were replaced and all new capacitors will also be fitted when they arrive.

The board on the lower left is the mixer board for the 5 presets, all but 2 resistors were replaced on this, as well as all the Turret pins, ready for direct soldering later.

The board on the right of the picture is one of the 12 tone generators, most of the resistors on these boards were of a good type, with a few random exceoptions on each of the 12 boards. The little board in the middle of the picture is a daughterboard on the tone generator. All the daughterboards on all the tone generators were populated with Carbon Composition resistors and all of them were replaced.

The little bag in the picture is full of the old replaced resistors.


The above picture is of part of the snowstorm that lived inside this machine due to the corrosion of all the connectors in it. Interestingly enough the usual ammonia based foam cleanser seemed to absorb all this very well, but all cloths had to be immediately recycled and handwashing had to be very thorough, the white residue is full of some very toxic metal salts.


Each of the Top Octave Generator boards has a nicely corroded edge connector which plugs into the back of the keyboard block assenbly. These all had to be sanded with very fine grade Emery paper to remove the white Oxides on all the connectors. After all this work, the reassembly of the organ was imminent and went accordingly to plan.

After all this work was done, it was time to rememdy any last remaining issues, and there were a couple of these. The primary one was the complete lack of any F keys and it turned out to be the reddish coloured frequency divider chip. We ordered one from Senso at Vintage Planet, and it arrived a week later.
The only remaining issue was a weird subharmonic being generated on the 8 foot tone generator of the F board, perhaps this was why the original owner replaced the chip with a chip bought from Maplin, which is the black coloured chip that appears in some of the pictures, but we saw this before with the last 71 and it was just a replacement of 2 capacitors in the Sawtooth waveform filters on the tone board.


This one was tricky but the end result was satisfying, and more importantly the customer was happy too.