Early 1950s Jenny Ondioline Tube Synth Restoration Blog

[HideawayStudio]

Well it seemed like folk on here were up for it so I've decided to blog my current restoration project - albeit a bit later than planned - a rare early 1950s Jenny Ondioline tube synth...





This poor beast was bought in a bedraggled state at a local auction a few months ago. Rumor has it that they actually had two for sale! As ever, half the battle has been trying to track down as much technical information on the beast as possible.

I have had some good luck finding the original user manual, some schematics, a patent application and some technical information from other owners the vast majority of which is in French which definitely slows things down a bit for me.. should have paid more attention at school!

The Ondioline was invented in 1941 by the Frenchman Georges Jenny and was often viewed as competition with the Jennings Univox, the Gibson Clavioline and the Hammond Solovox which was released a year earlier. It is similar to the Ondes Martenot in that it was designed as a highly expressive instrument but its oscillator is much more similar in design to the aforementioned instruments. In my opinion, out of all of them, the Ondioline is the most interesting and musical possibly bar the wonderfully eccentric "diffuseurs" cabinets on the Ondes Martenot.

It is a curious instrument (like the Citroen DS of early electronic musical instruments!) and I'm very much driven by the fact that some of the vintage demos I've heard of this thing have been genuinely impressive for a solo electronic instrument of this age. The one thing that really stands out is the instrument's expressive qualities and is quite capable of rendering a remarkably realistic impersonation of a violin in the right hands!

By far the most famous, loyal and prolific user of the Ondioline is Jean-Jacques Perrey, now 86, who spent much of his life promoting the instrument.

Here is Jean-Jacques demonstrating his instrument in 1960 on the TV show "I've Got a Secret":

[HideawayStudio]

The instrument was stripped down to its major components some time back but then all was delayed by the recent Polymoog restoration. This wasn't a bad thing as its given me a chance to find as much technical info as I can.

Like the Novachord, the first thing to tackle had to be the PSU. In the case of the Ondioline the PA and PSU are combined and are fortunately of a very simple all tube design.

There were many minor revisions throughout the Ondioline's production life, including a number of packaging options, but perhaps the most major was changing from the larger octal base tubes to the later, more compact, B9A base tubes. In fact on the Oscillator chassis you can clearly see where existing stock of the chassis to receive octave tube bases has been somewhat crudely modified with insulating material to accept the smaller B9A bases!

Ondioline #599 mostly employs the later tubes and was probably built some time between 1950 and 1954. Frustratingly, I've not managed to identify any date codes on the capacitors so far.

The PA consists of a BY3G tube rectifier, an ECC83 double triode based preamp and phase splitter tube and two EL84 output pentodes to make up a class AB push-pull amp of around 15 watts driving a 4 ohm 10" loudspeaker... a really common design from the era used in countless grandiose warm sounding walnut encased 1950s radiograms, punchy 1960s jukeboxes and small but often surprisingly loud guitar amps.

The PA chassis was thoroughly cleaned. Safety, or rather the lack of it, was quickly identified as an issue. To my amazement, the mains input, speaker output and signal input sockets were all on 1" pitch banana connections!!!

Even more alarming was the circular connector that made up the umbilical carrying the mains switch wiring, heater and 280V B+ rails to the oscillator chassis was female so the live pins were exposed when the connector was not mated!!

If that wasn't bad enough its since become obvious that if the instrument was run without the oscillator signal connector plugged in the ground is lost to the metal oscillator chassis with nothing stopping the whole thing sitting at 280 volts DC!! This has clearly been done to prevent the volts drop of the 6.3V AC heater load (around 2 amps) in the return path coupling excessive hum on the signal returns if locally tied over to the oscillator chassis ground reference but since the connection is easily removable in isolation its a nasty accident waiting to happen.

I removed the mains input connector and decided to fit an IEC inlet and fuse holder at the same time. I retained the speaker output and signal input connectors but decided to dress the wiring in such a way that it was obvious which cable went to each connector.



Quickly it was determined the smoothing caps dotted around the chassis were no good and indeed a burnt lead fell out of one of them so they were replaced with new equivalents under the chassis whilst retaining the disconnected originals above the chassis for authenticity.

The loudspeaker was removed and found to be in tact but quite rusty so it was given a lick of paint and refitted.

I decided to reverse the male and female connectors on the umbilical to the oscillator chassis and so I mounted the female connector on the PA chassis and rewired the cable to match. This resulted in a much more safe connector configuration. I also fitted a DIN socket and high voltage blocking capacitor so the instrument can be recorded directly from the line output.

Sadly the PA chassis was quite badly tarnished and rusty in several places. In order to have completely finished the chassis I would have had to strip it down to an empty shell (which is what I did with the Novachord PSU) but since it was complete and clearly not far from being operational I decided to clean it all up as much as I could.



After recapping the PSU I rewired the loudspeaker cable and connector and slowly fired the PA up for the first time in probably very many years on a Variac.

To my delight the PA came up and to my surprise the speaker was clearly in surprisingly good condition (depressingly, rather too often loudspeakers of this age can produce nasty scraping noises due to the moving coil and spider going out of alignment or delaminating with extreme age and scuffing the magnetic pole piece its positioned over) . Together, with the speaker back in the cabinet, they sound great!.. a nice warm full bodied jukebox like tone and surprisingly loud with it.





The PA ran for several hours so I decided to move on to the oscillator chassis.

The oscillator chassis was really very dirty when it was removed from the cabinet. 60 years of dust and grime had to be removed before any further work could be undertaken.

Quickly it became obvious that the plastic overlay on the slide switch panel had shrunk with age and was binding against the edges of the metal levers. The metal plate and overlay were removed and cleaned. The overlay was filed to match the profile of the plate below it. On reassembly the switches moved much more freely.



The oscillator chassis is now awaiting a full recap after testing several metal film and electrolytics only to find most of them were not in good condition.

[HideawayStudio]

I have spent today recapping Ondioline 599's oscillator chassis. The decision to replace pretty much all of the caps bar a handful of mustard types was made after making several spot checks only to find many of the original 100, 220 and 500nF caps were reading as much as 100% out of spec!

I wasn't in the mood for taking chances having again narrowly avoided a nasty incident with a 60 year old German built PSU decoupling electrolytic in a piece of quality test gear which was so leaky at 280 volts it burnt out the series resistor feeding it! The point is that this cap tested pretty much fine on the meter and is another example of why testing old electrolytic caps in tube gear really needs to be carried out at voltages of at least 50% of the caps voltage rating using a series resistor to measure the leakage current.

Armed with several bags of high voltage metal film caps the exercise went very smoothly with only really the usual issues of coming up with ways to mount discrete electrolytics in place of metal can multicaps. My preferred method of late has been to cut the legs off of the old caps where they are mounted above chassis and then binding two caps together side by side, sometimes in heatshrink, by their negative leads and solder the common lead to a solder tag bolted to the underside of the chassis. This leaves the original cans in place above the chassis thus retaining an authentic look.

Out with the old caps:


In with the new.. turned out very neatly..


The plan is to attempt to run the chassis up on the bench using a current limited bench supplies for the 6.3V AC heater and 280V B+ rail.

I have a superb old Hewlett Packard HP6207B current limited high voltage supply which has proven its weight in gold. It only goes up to 160 volts but that is ample to reform HV caps and at least determine whether there is any life at all in a tube design. In fact, it often surprises me how low you can wind the B+ rail on much tube gear and still have it work. As long as the heater rails are 100% many pieces of tube kit will quite happily do something useful even at 90 volts.

[Mooger5]

Great work as usual. Thanks for sharing. Can´t tell if the expressiveness is due to the all-tube electronics, the player that was a virtuoso, or the "architechture" of the sound engine. What are the controls A, B. C etc? Harmonics like in the RMI? And in the photos they look continuous but in the clip they seem to be either full on or off.

[HideawayStudio]

The last 24 hours have been intense!!

I managed to run up the heaters on the oscillator chassis on a bench PSU and then hooked up the bench HV supply and slowly wound the B+ voltage up to 170 volts. All of the tube heaters ran and the current drawn from the B+ rail was less than 15mA...

The Recapped Oscillator Chassis Safely Powered Up on the Bench..


So I turned my attention to the keyboard assembly as there a few too many connections to attempt to convince the oscillator chassis it can run without a keyboard attached.

Very much like the five Jennings Univox I've recently overhauled I was expecting a battle with the keyboard assembly and I was proven right to be apprehensive!

One of the things that sets the Ondioline apart from the other Clavioline type instruments is its "Progressive Attack" Controller which is a little like primitive monophonic aftertouch. Unfortunately the rumors are true that there is a very small amount of asbestos within this mechanism which acts as a dielectric between two moving plates to form, in essence, a contactless variable impedance.

With this in light I took the whole keyboard unit outside and doused it in copious amounts of water. The cabinet was then opened and the entire offending mechanism removed and put in a bag to deal with appropriately later. The rest of the keyboard cabinet and assembly was then very carefully cleaned up with damp clothes.

Now in a position to see what was going on I quickly realized that although everything was complete the effects of 60 years of damp and heat had left its toll.

Horribly Distorted 60 year old Plastic and Shorted Plate Capacitors!...


The Plexiglas type material the octave range switch, tuning capacitors and octave trim pots had badly shrunk and sagged to the point everything was badly geometrically distorted.

Seized Solid!... The Tuning Trim Pots


The shaft on the range wafer switch itself was seized solid to the point even after lots of oil the shaft sheered off whilst trying to free it up. Similarly the wipers in all of the tuning pots had corroded solidly onto the carbon tracks and worse still all four plate capacitors for the range trims were testing short circuit!

I went away to have a think and decided since it was clear the entire range switch subassembly will need to be rebuilt I decided to have a go hard wiring a single range.

The Above Nastiness Disconnected and Bypassed...


After consulting the schematics it turned out that with most of the wiring disconnected to the range switch hard wiring a single range was simply a matter of bypassing the trim pots and fitting a single capacitor across the main tuning pot. To my relief the tuning pot itself tested ok and this is fortunate as its a very unusual dual ganged pot made up of a 20K and a 5K pot on the same shaft. Equally, the rather nifty lateral variable capacitor mechanism for the manual vibrato also tested good.

On closer inspection the rubber insulation on the shielded audio output cabling to the swell lever mechanism was badly perished so I removed it all and replaced with with modern silicone rubber wire and fitted a new high voltage blocking capacitor mounted inside a metal tube next to it. Rather surprisingly the swell potentiometer itself tested very smoothly after a good dose of DeoxIT.

The Rewired Swell Mechanism and Output Capacitor..


Looking at the schematics it became clear that the Ondioline wasn't going to function properly without its Progressive Attack Controller as this forms part of a feedback path so I made a bit of a guess to its capacitance and fitted a simple fixed capacitor in its place for now. The hope is that eventually I can find a suitable substance to act as a dielectric and the controller can be rebuilt and reinstalled.

Thankfully the divider resistors in the keyboard mechanism are high quality wirewound types which are very stable even with extreme age. Equally fortunately the key contacts were testing good.

So with everything wired up and seeming to test ok on the meter I temporarily reassembled the keyboard assembly and after replacing several rusty screws I refitted it to the top of the main cabinet and reconnected the wiring to the oscillator chassis.

The Oscillator Chassis Back in Situ and Hooked Up..


With the Variac plugged into the bench isolation supply I plugged in and slowly wound up the power.

To my delight... she powered up!!

To my very great surprise my guesstimated tuning capacitor was close enough too... it actually tuned to A440!

All of the tabs work - can't really fault the recapped oscillator and PA chassis...

Time melted away at this point but I did manage to put a quick test recording together...



It certainly is a very characterful little instrument and the manual vibrato, created by wobbling the keys from side to side, is very effective and can be used as well as the automatic vibrato.

Still a Lot to Do but Ondioline #599 is At Least Now Playable!..


So in answer to the question regarding the tabs... they are basically slide switches. They appear to have two positions - I say this because I'm not sure if the central position is a valid position.

On feature I am particularly impressed with is the waveshape appears to be continuously variable and the knobs on the left and right of the tabs appear to dial in variable amounts of harmonics consisting of the octave above and below the fundamental which produces a whole array of useful tones. There are tabs for basic waveshape, a number of resonators, various EQ settings, vibrato, vibrato depth, an repeat mode, a percussive mode and a sub oscillator.

[Mooger5]

Probably the asbestos were there to prevent fire hazards. According to http://www.clippercontrols.com/pages/Di ... alues.html the dielectric constant is between 3 and 4.8 but there should be lots of air in between the fibers so the actual dielectric constant must be lower. I'm guessing some teflon tape would be a safe substitute, in case of arcing between the plates. Not the plumber's type. The one used in gaz conduits is much thicker.

[HideawayStudio]

Having removed the small amount of asbestos from the Ondioline's infamous "Attaque Progressive" the subassembly was stripped down to its component parts.

The mechanism is basically made up of a grounded metal enclosure inside of which two small fixed blocks of wood covered in tinfoil are screwed to it. The input signal is applied to one via a solder lug and the output signal is taken from the second block in the same way. Originally these blocks had a layer of asbestos wadding glued to the top of them to act as a dielectric material. A metal paddle shaped plate supported on a wooden insulator was then suspended above both blocks. The sensor paddle is actuated by a common contact bar (or "Barre de Contact Central de Masse") which moves upon any key press. The further any note is pressed, the closer the metal paddle is positioned to the fixed blocks. The result is a device that relies on the effects of Mutual Capacitance to offer more signal to the vari-mu preamp in the final formant filter stages the closer the plates are brought towards each other due to the increased reactive coupling from input to output plates (ie. its capacitance increases in much the same manner as when the set screw brings together the plates on a basic trimmer cap in a radio or oscillator).

The operation of this sensor is frequently misunderstood and often incorrectly described as a resistive sensor. This may be possibly due to the original technical descriptions being in French and it should be noted that any reference to "resistance" is in fact more correctly a reference to "impedance" ie. the sensor can only physically pass signals in the AC domain.

The plan is to partially rebuild the subassembly and try out a number of different substitute materials in place of the original asbestos wadding. Before this can be undertaken the wooden blocks need to be clad in new tinfoil and suitable insulators cut to replace the cardboard that was originally separating the blocks from the outer metal casing.

The signal into the controller is fed off half of an ECC82 double triode tube acting as a common cathode preamplifier with a blocking capacitor and grounded leak off resistor off of the anode load resistor thus no high voltage DC should be present on the sensor.

[HideawayStudio]

My experiments with using acetate film (overhead projector film) have been pretty promising so I cut and glued down two small sheets over the top of the refoiled wooden blocks and reterminated the input and outputs with new silicone shielded cables.



The progressive attack controller is driven directly from a common cathode triode stage with open loop gain so I tested the unit with an old Rohde & Schwarz tube signal generator that produces up to 30 Volts output.

The rebuilt Progressive Attack controller appears to work nicely and when used to modulate the output of the signal generator almost had a Theremin quality in its own right. With the exception of slight crackling when the sensor hits its rest position the sensor is completely noise free in operation and with no moving contacts should last indefinitely which was very much the whole point of the original design.

One thing I did learn the hard way is that the wooden blocks carrying the foil must not only be positioned on insulating material but the screws that secure them must also be insulated otherwise the resistance of the wood itself is enough to make the sensor leaky and ineffective. I found the plastic top hat insulators for TO220 transistors perfect for the job.

Rebuilt "Attaque Progressive" Being Tested on the Bench:

[HideawayStudio]

I spent this afternoon wiring in and remounting the Progressive Attack sensor in the keyboard cabinet and after a little adjustment... it works!!

This contactless sensor permits the player to create everything from staccato sounds to tremolo effects and swells or anything in between simply by how you hold down each note. This is pretty impressive for an electronic instrument that was released over 70 years ago!

Manual vibrato is made by moving the keys laterally so it takes quite a lot of concentration to get it right...

Ondioline Brass: The "Attaque Progressive" Gives Some Life to Otherwise Static Formants...

[musomark]

It's sounding great! Thanks for sharing.

[Mesmerised]

Thanks for sharing this adventurous restoration! I had seen (and heard!) an Ondioline in April at Nantes Synth Fest and had been very impressed by its sound! (it's really loud, you could always hear it above the din in the hall...)
Also the expression is very impressive and all the possibilities to change the sound with the levers.
I had been curious about the insides and thanks to you I could now get some insight.

[HideawayStudio]

As you can see from the picture above - the Ondioline was available in a number of packaging options. The rarest arrangement appears to be a variant where the control tabs were positioned in a shallow but wide cabinet to one side of the keyboard. Ondioline #599 has pretty much the same cabinet style as the one shown in the Ondioline Beginners Guide.



And here is list of timbre patches from the same guide:

Liste Des Timbres:


The Rebuilt Expression Controller Back in Situ:


The next major task is to completely rebuild the subassembly running along the top of the picture which is the octave range control. To date I've had to hard wire the instrument in one of the lower registers but there should be four octave ranges in total.

The original plastic plate supporting the trim resistors, trim capacitors, main tuning control and range wafer switch has become so badly distorted with age its distorted the trim caps to the point they have short circuited and all of the switches and trim pots on it have seized solid with corrosion.

The current plan is to machine a piece of angled aluminium to remount the existing tuning control and new trim pots along with a modular three pole four way wafer switch I have just assembled. The bulk of the tuning will be reined in with metal film capacitors and fine trimmed either by select on test or, if I can source something suitable, some small high voltage trim caps.

From what I can determine to date with the one range I have hardwired, the trim pots appear to affect the span more than the tuning which is somewhat different to the way the Jennings Univox is calibrated.

Very fortunately, the custom wirewound resistors (each having a sequential number printed on them) are of very high quality and are extremely stable.

I have a wonderful vintage decade resistance box in a dovetailed hardwood box which is at least 70 years old made up of very similar resistors which covers 0 to 9999 Ohms and when tested with a modern digital multimeter tallies to all four digits no matter what resistance is dialed up on the box!

As I've said many times, it amazes me how stable, reliable and robust some vintage components and materials are.

[sizzlemeister]

Absolutely fascinating thread! I'm all into these kinds of rebuilds, and while I can rebuild a tube amp, I don't know if I could do a tube synth.

The clips and pictures are marvelous! The sound one of these ancient synths puts out must be hypnotic in person.

Keep it coming!

[HideawayStudio]

Today has been another intense day!...

Finally all the parts had arrived to built the new Octave Range Control and calibration subassembly including ceramic trim caps, metal film capacitors, aluminium bracket, wafer switch, trim pots and tag strip.

Sadly the original subassembly was just too far gone. Bar the main tuning control nothing else was salvageable due to severe corrosion and plastic deformation.

One hurdle to overcome was that the original compression-plate trim capacitors were destroyed by the deformation in the plastic and it proved quite difficult to source trim caps with a sufficient high voltage, capacitance and range. For a while I toyed with using radio tuning capacitors but then got lucky when I discovered a high voltage compression trim cap used in the LA-2A tube Leveling Amplifier was available as a spare part.

It just so happened that the contacts on the tag strip I ordered were exactly the right pitch to permit 4 trim caps to solder directly between them.



Better still, the tag strip fitted perfectly on the aluminium bracket which was drilled to accept the original tuning knob, the three trim pots, the tag strip and the octave range switch.



The hardware was mounted on the bracket and then trim pots wired to the first wafer on the Octave Range Control Switch - each acting as a spanning control for each octave setting. Each trim cap, which act as tuners for each range, was wired to the second wafer. The third wafer is reserved for dialing out excessive manual vibrato in the upper octave ranges.



To the right of each trim cap two tags were paralleled with them to permit easy access whilst selecting suitable capacitors to achieve the correct octave ranges whilst the instrument was running. Although there are not massive voltages on these caps I did find myself receiving a few tingles from time to time! These fixed capacitors act as the bulk of the capacitance in each R/C network with the trim cap offering fine tuning.



To my delight the proposed solution works very well and with a little extra tweaking tomorrow it should be possible to switch between all four octave ranges without having to retune.

One the spanning and relative tuning is established the keyboard console front panel, top cover, trim plate and knobs can be refitted.

Although the tuning is not yet perfect it has been very nice to hear the other octave ranges in particular the lowest range which is great for producing bassoon and horn like timbres.

[HideawayStudio]

After doing a little more fine tuning with the bulk capacitance on each of the four octave ranges this morning I have reassembled the keyboard cabinet ready for a quick spin.

The Octave Range Switch is now working well and after calibration now permits the player to switch between each range without having to retune:



I am really quite taken by this instrument. It is remarkably expressive under the fingers and has a surprisingly wide sonic palette for a 70 year old design. It must have caused quite a sensation in the early 1940s!!

The M tab adds a whole dimension with the extra divider circuitry which permits varying degrees of harmonics to be dialed in on the controls to the left and right of the tabs.

The Progressive Attack aftertouch sensor works remarkably well on staccato and brass timbres: