Elka Synthex is back!

milkshake · Post by **milkshake** » Fri Apr 03, 2015 2:07 am

http://bringelkaback.com/index.html

skyshaver · Post by **skyshaver** » Fri Apr 03, 2015 4:03 am

Very pretty website at least...

brokensolderingiron · Post by **brokensolderingiron** » Fri Apr 03, 2015 4:09 am

Generalmusic will offer unbelievable Italian sound together
with superior technical quality - made in Finland.

Finland ceramics , textiles, rubber boots, Hakkapelita winter tires and classic saunas is all very good, but a redo of the Elk is sort of lame really, from where are they supposed to get the CEM3320 etc? Seams that someone have just bought the old trademark names and no old inventory. Do something new instead for god sake!

Post by **Paranormal Patroler** » Fri Apr 03, 2015 7:04 am

Is GEM still active?

dopefiend · Post by **dopefiend** » Fri Apr 03, 2015 8:41 am

Hmmmmmmm......a huge synth with no touch sensitivity. No thanks. I'll keep my Rhodes Chroma....or buy back my CS80....

Unless they upgrade it with:

A.- Less bulk.

B.- Touch sensitivity. I mean, c'mon....it is in Italy where Fatar is made!

thefutureeaters · Post by **thefutureeaters** » Fri Apr 03, 2015 11:33 am

I'm in. Thanks for the head's up.

sneak-thief · Post by **sneak-thief** » Fri Apr 03, 2015 12:35 pm

Spent a few weeks last year doing a minor Elka Synthex overhaul.

As a DCO synth, it's actually quite lively due to the complex modulation options and 2-voice layering. Nicest DCO synth I've ever heard. Listen to Jean-Michel Jarre - Rendezvous to get an idea of what it sounds like.

I'm not sure how important the CEM3320 is to the sound of it. I have a feeling you could get pretty close using a similarly configured 2164 dual state-variable filter:

- http://www.electricdruid.net/index.php? ... fo.cem3320

Here's the one I worked on:

OurDarkness · Post by **OurDarkness** » Fri Apr 03, 2015 3:33 pm

Awesome news. Bring it on!

skkatter · Post by **skkatter** » Sat Apr 04, 2015 4:14 am

The Soundonsound review and its service manual say it uses digital oscillators rather than DCOs. You got any higher resolution photos of the oscillator section?

sneak-thief · Post by **sneak-thief** » Sat Apr 04, 2015 6:53 am

skkatter: you're correct - it's not a traditional DCO in the sense that there isn't a charge pump capacitor, but calling it a flat-out digital oscillator glosses over some curious aspects such as it's lack of traditional DACs.

Personally I'm not sure what to call it. Many call it a DCO synth so at the time I was just going with that for convenience. But now that you bring it up, here's a quick, rough overview:

1. Each of the four oscillator PCBs generates two pairs of OscA/OscB voices.

In the pic below, I isolated one of the OscA/OscB voices.

On the bottom of the first page of the schematic I posted, you'll see a digital signal from the main digital bus come in (D2, D3, D4, D5, D6, D7).

2. At this point, the signals designated as Saw1, Saw2 are still TTL-level digital pulses. I didn't scope the Saw1/2 digital signals so I don't know if they change when you select triangle instead of saw.

As we move upward in the schematic, we see a combination of digital control signals go through what looks like an improvised 8-bit ladder-type DAC using two 74LS193 up/down binary counters. Out comes VCO2, aka OscB.

A little bit further up is the octave-switching logic for OscB which controls the aforementioned 74LS193's.

3. The top half of this part of the schematic for OscA is a duplication of OscB with the addition of oscillator sync input. Supposedly the sync is on OscB, so either the names here are swapped or this sync somehow feeds back onto OscB below.

4. On to the second schematic: now we have VCO1 & VCO2 going through a series of 4066 switches which look suspiciously like a 4-bit DAC ladder.

What are they doing? Digitally-controlled waveshaping? Not quite sure as the schematics are pretty opaque when it comes to designating actual functions of each section. Those 4066's are also switched via the main digital bus (D2, D3, D4, D5, D6, D7).

5. Next the VCOs go through the CEM3320 multi-mode filter then an LM13600 VCA.

skkatter · Post by **skkatter** » Sat Apr 04, 2015 7:58 am

Thanks for the overview, it's an interesting piece of engineering.

Somebody on a different forum pointed out that you can't do cross modulation with DCOs, but you can do cross modulation with the Synthex, which is another plot hole I suppose. :)

sneak-thief · Post by **sneak-thief** » Sat Apr 04, 2015 8:22 am

The Synthex doesn't really do proper cross-modulation. Same goes for sync & ring-mod which are digital approximations. My guess is MR1 and MR2 deliver the logic signals for faked cross-mod and R & EOR might be for ring-mod. Not sure what selects tri vs saw.

I'd be curious to see a scope of just one of the Synthex's saws and triangles. Adding various square footages could also create a staircase approximation of a saw/tri (like the Poly 800 or various divide-down synths).

FWIW, the LFOs for the VCF, VCA and PWM are analog via a LM13600 but the VCO mod is a digital signal (MOD1 & MOD2)

chamomileshark · Post by **chamomileshark** » Sat Apr 04, 2015 12:02 pm

Funny thing to bring back.

I remember these came out at some high figure just as the DX7s came out. I think even before that it would have been regarded as an also ran - it was I think nearly £3,999 when it came out - not much cheaper than a Prophet 5 or one of the Oberheims?

It then turned up in E&MM magazine (electronics & music maker) for a special price of £999.

I know JMJ rates them and I'm sure they are nice but I don't think they are that special are they?

ndkent · Post by **ndkent** » Sat Apr 04, 2015 12:45 pm

The Forrest A-Z Synthesisers book says £3199 and mentions the £999 blow-out. At the time people were not looking for knobby polysynths from a brand not known for pretty advanced synths with no advertising budget.

I never saw one in the USA though I did and tried one out in Europe and was impressed. I think the majority of people into analog polys would be impressed if they tried one (I'm not talking about someone not interested in the category in general).

Oh, and true it doesn't have velocity, not that the P5, MemoryMoog, Jupiter 4-6-8 or OBX-Xa-8 did

hageir · Post by **hageir** » Sat Apr 04, 2015 6:30 pm

OMG OMG OMG

brokensolderingiron · Post by **brokensolderingiron** » Sat Apr 04, 2015 7:39 pm

sneak-thief wrote:Spent a few weeks last year doing a minor Elka Synthex overhaul.

As a DCO synth, - it's not a traditional DCO in the sense that there isn't a charge pump capacitor, but calling it a flat-out digital oscillator glosses over some curious aspects such as it's lack of traditional DACs.

Personally I'm not sure what to call it. Many call it a DCO synth so at the time I was just going with that for convenience. But now that you bring it up, here's a quick, rough overview:

1. Each of the four oscillator PCBs generates two pairs of OscA/OscB voices.

In the pic below, I isolated one of the OscA/OscB voices.

On the bottom of the first page of the schematic I posted, you'll see a digital signal from the main digital bus come in (D2, D3, D4, D5, D6, D7).

2. At this point, the signals designated as Saw1, Saw2 are still TTL-level digital pulses. I didn't scope the Saw1/2 digital signals so I don't know if they change when you select triangle instead of saw.

As we move upward in the schematic, we see a combination of digital control signals go through what looks like an improvised 8-bit ladder-type DAC using two 74LS193 up/down binary counters. Out comes VCO2, aka OscB.

A little bit further up is the octave-switching logic for OscB which controls the aforementioned 74LS193's.

3. The top half of this part of the schematic for OscA is a duplication of OscB with the addition of oscillator sync input. Supposedly the sync is on OscB, so either the names here are swapped or this sync somehow feeds back onto OscB below.

4. On to the second schematic: now we have VCO1 & VCO2 going through a series of 4066 switches which look suspiciously like a 4-bit DAC ladder.

What are they doing? Digitally-controlled waveshaping? Not quite sure as the schematics are pretty opaque when it comes to designating actual functions of each section. Those 4066's are also switched via the main digital bus (D2, D3, D4, D5, D6, D7).

5. Next the VCOs go through the CEM3320 multi-mode filter then an LM13600 VCA..

My quick 5 cent, have not checked the design in detail!

1-2: It's a peculiar design, there is 4 VCO's LM331 controlled by a linear current composed of LFO/Bend/tuning etc. Each 331 out pulse is then
feeding a PLL configuration with LS624 VCO as the PLL's high frequency output who's clocks the A and B DCO's on the dual voice boards.

So it's a VCO-VCO-DCO chain somewhat similar in principle to OSCar the difference is OSCar sends it's final high frequency clocks to a counter that counts on a memory where waveforms are stored.

3:The DCO square wave clock's a waveform generator (LS193) who's wave seams to be a triangle by inverting counts by the XOR gates every
256 +1 count , kind of 9bit triangle and 8bit saw is generated by counting to 256 and then restarting. Square is just the MSB bit and PWM
is saw or triangle through a comparator. The waveform DAC is the XOR gates feeding the weighted resistor net.

The 10n capacitor and BC559 seams to be step smoothing and sync restorer and ramp kill. At first i thought it was the charge/reset ramp circuit but the cap was to large.

4: The 4016/66muxes is used as DCO volume selection.

But the elegance in the whole synth is the ADSR design.

Measure the ramp at collector of 145 PNP transistor to see if the steps is smoothed out or not.

sneak-thief · Post by **sneak-thief** » Sun Apr 05, 2015 4:40 am

brokensolderingiron - the Synthex has neither LM331's nor LS624's.

It does however take the modulation input into an LM311 comparator.

The LS193 analysis makes sense as well as the 4016/4066 DCO volume control - probably for the OscA/OscB balance.

dougt · Post by **dougt** » Sun Apr 05, 2015 5:58 am

skkatter wrote:Somebody on a different forum pointed out that you can't do cross modulation with DCOs, but you can do cross modulation with the Synthex, which is another plot hole I suppose.

Someone hasn't seen any of the Roland JX series I guess...

usw · Post by **usw** » Sun Apr 05, 2015 9:15 am

What roland called cross mod or metal in its jx serie has nothing to do with frequency modulation, it was just a pseudo ring mod, and frequency modulation of DCOs derived from low resolution counters would sound really bad indeed.

brokensolderingiron · Post by **brokensolderingiron** » Sun Apr 05, 2015 2:54 pm

sneak-thief wrote:brokensolderingiron - the Synthex has neither LM331's nor LS624's.

It does however take the modulation input into an LM311 comparator.

The LS193 analysis makes sense as well as the 4016/4066 DCO volume control - probably for the OscA/OscB balance.

sneak-thief - you have to read all of the schematic pages! I suggest sheet 5-10 and 6-10.

sneak-thief · Post by **sneak-thief** » Sun Apr 05, 2015 6:02 pm

brokensolderingiron wrote:
sneak-thief wrote:brokensolderingiron - the Synthex has neither LM331's nor LS624's.

It does however take the modulation input into an LM311 comparator.

The LS193 analysis makes sense as well as the 4016/4066 DCO volume control - probably for the OscA/OscB balance.
sneak-thief - you have to read all of the schematic pages! I suggest sheet 5-10 and 6-10.

Ack - my PDF was missing page 5-10! I overlooked that PCB because I had only worked on the oscillator side - and had assumed it just handled the upper and lower LFO's.

Jesus, where the hell does that clock bus come in? I imagine CP1 and CP2. Now I get it. This board likely has the weirdest poly oscillators of any in that era, and I've repaired synths like the OB-Xa, Expander, CS-70m, JP8, P5 Rev2 & 3, etc.

skkatter · Post by **skkatter** » Tue Apr 07, 2015 8:52 am

So, correct me if I'm wrong as I'm not the most technical savvy person in the world, but the conclusion is we're not 100% sure if it's a DCO, and to check this we'd need to measure the ramp at collector of 145 PNP transistor to see? So somebody with their hands on an Elka could check it pretty easily with an oscilloscope?

It does seem to be a pretty complicated voice board alright.

StepLogik · Post by **StepLogik** » Tue Apr 07, 2015 10:35 am

It is going to be interesting to see the approach they take if they actually decide to reissue the Synthex (I'm skeptical).

Part for part (as much as possible) recreation of the original? If so, then the price will be in the 5-figure "collectors only" territory as per the reissued Moog Modulars.

Will they try to modernize it? SMT, consolidate all those separate panel boards and panel mount pots into a single (or two) boards? Consolidate the voice cards into a single SMT board? Switching power supply to reduce bulk? Cheaper switches and keybed? Try to implement more of the digital logic in software to reduce component count? Dave Smith has proven this approach pretty well, but the final result is still a fairly expensive synth.

Start saving your pennies now!

pricklyrobot · Post by **pricklyrobot** » Tue Apr 07, 2015 11:33 am

I'll just wait for the Behringer version.

revtor · Post by **revtor** » Thu Apr 16, 2015 7:01 am

Reissuing a synth with 1000 IC's? They could probably get rid of 75% of them with some sort of big FPGA, and then they PS requirements could be much less..

I wonder if they'll make any changes, inverted envelopes maybe?

~Steve

MOD WIGGLER

MOD WIGGLER

Elka Synthex is back!

Elka Synthex is back!