Has anyone successfully cloned the 921b Osc.?
I wonder if the module would work without the uA 796 Sync section? (a rarecomponent …but allowing for +12v connection and various resistors to SAW out.)
Check MOS-LAB, located in France. Said to be among the best: http://www.youtube.com/watch?v=SMTouJJP51A
yes…familiar with these thanks…but I’m more interested in the home made attempts..
You need the companion 921A in order to get v/oct response
The 921 would be a lot easier to clone than the 921A/921B, parts are easier to get.
I’ve seen 921 clones that are based around the CEM3340 VCO, that isn’t even a faithful clone.
A common 1496 ring modulator IC will work in place of a uA796, I think.
Oh ??? thanks Kevin..thats given me a new lead..I was going to emit the ua796 but now thats something else to follow up.
The 921b and 921 are very similar but deviate around this chip..but both complicated circuits…quite a marathon.
You might consider just building the base design of the 921/921B and omit the sync features.
While it’s usable and nice to have, it’s overly complicated and isn’t worth the trouble for the amount it’s usually used.
Remember, no matter what design you build, you will still need a tempco resistor and high quality cermet trim pots.
Fwiw, when I was building up Hans Zimmer’s huge Moog, he had a system delivered that was previously owned by UCLA.
That system had some DIY 921A/B sets that were built by JL Cooper.
On the boards it said “Moog via JLC”
So it IS possible to dupe them, but they’re not an easy project and certainly nothing for a beginner.
I very much suggest prototyping them beforehand on breadboard or other temporary fixture.
It is also high recommended that you borrow someone’s 921(s) to compare things to.
Schematics and photos can only go so far.
Good luck!!
thanks again Kevin for your kind help …the 1496 you suggested is a replacement for the 796 by the looks of it so its down to different case but I should sort it OK..that was a great tip. I would like to stick to the original Moog module if possible. Ive begun the design and have taken Moogs original layout and circuit plan (this Ive done from the few photos Ive gleaned..I wont be able to get a hands on of a 921b here in the UK) but the top layer tracks are proving elusive as the components cover the copper in most places but if I work from the schematics I should be able to get there eventually. (I hope) The 921a is a doddle but this 921b is the most difficult Moog Ive cloned so far but it will be superb to have the 901s and 921s in my system. I’ll keep everyone posted but I reckon this is gonna take a good 6 weeks or so. (then theres the set up!)
I just found this topic about Moog 921b clones..
Nice info from you guys !
I’m in a System 55 clone project myself since a while..
The clone won’t be a ‘perfect’ clone but VERY similar
estheticaly speaking. I’ll also add some new modules
of my own like this MIDI interface:
http://www.arcenson.com/projects/modular/M100%20MIDI%20Interface/
I already started a 12 VCO’s job aimed for a futur MIDI
guitar controler.
So I needed:
4 x 921a VCO drivers clones (called M103A in my system)
http://www.arcenson.com/projects/modular/M103A%20VCO%20driver/
and
12 x 921b VCO’s clones (called M103B in my system)
Not yet uploaded on my ftp but for info:
My 921b clones have the exact front panel design as 921b’s
except for the RANGE steps..
Mine are 32’,16’,8’,4’,2’,1’ instead of the original LO,32’,16’,8’,4’,2’
AND one major thing is the inner circuit…
The VCO CORE is an exact copy of the famous
MOTM UltraVCO MOTM300 completely PCB redesigned
with stuff I removed and stuff I added around for front panel component uses.
For copyright reasons I won’t be able to share any schematic
original & modif but some photos will be available.
To be followed..
The following link shows some modules ready to share.
http://www.arcenson.com/projects/modular/
I’ll be uploading my 921b VCO’s detailed files very soon
in this ftp space.
Anybody who has the same Moog system 55 cloning goal
are very welcomed to share ideas!
Cheers,
J-Pierre
Hi – interested in your cloning of the 921B/921A pair – i have almost finished doing this and had a fair amount of success so far. I still haven’t got the sync circuitry with the multiplier going completely and had to “tweek” the design a little bit to as i just cannot get a dual fet at anything approaching a reasonable price. You can see the rather slow progress here: http://forbinthesynthesizer.blogspot.com.au/2010/06/it-lives.html I am just in the throes of getting back into it and will try for a dual foot print for the multiplier along with fixing up the transistor array stuff up! Got a nice e-mail from a lady at Moog music that they are quite happy for me to post the layout (single sided and reasonably easy to etch your self + 2 x regulators so can run from ±15V) when i had finished as long as it didn’t say moog all over it and purport to be from them. She basically said that they weren’t really very interested in anything prior to Bob getting the company back in the mid 90’s…
Sorry, my last post had bad links to my projects pictures & files..
Here it is again when the corrected links. Just have a look..
Sorry again.
JP
I just found this topic about Moog 921b clones..
Nice info from you guys !
I’m in a System 55 clone project myself since a while..
The clone won’t be a ‘perfect’ clone but VERY similar
estheticaly speaking. I’ll also add some new modules
of my own like this MIDI interface:
http://www.arcenson.com/projects/Modular/M100_MIDI_Interface/
I already started a 12 VCO’s job aimed for a futur MIDI
guitar controler.
So I needed:
4 x 921a VCO drivers clones (called M103A in my system)
http://www.arcenson.com/projects/Modular/M103A_VCO_driver/
and
12 x 921b VCO’s clones (called M103B in my system)
http://www.arcenson.com/projects/Modular/M103B_UltraVCO/
My 921b clones have the exact front panel design as 921b’s
except for the RANGE steps..
Mine are 32’,16’,8’,4’,2’,1’ instead of the original LO,32’,16’,8’,4’,2’
AND one major thing is the inner circuit…
The VCO CORE is an exact copy of the famous
MOTM UltraVCO MOTM300 completely PCB redesigned
with stuff I removed and stuff I added around for front panel component uses.
For copyright reasons I won’t be able to share any schematic
original & modif but some photos will be available.
To be followed..
The following link shows some modules ready to share.
http://www.arcenson.com/projects/Modular/
Anybody who has the same Moog system 55 cloning goal
are very welcomed to share ideas!
Cheers,
J-Pierre
So, your Moog© 921b clone is really an exact copy of MOTM© Ultra Vco©, but for copyright reasons you won’t post pics of the pcb? 
The panels look good!
In looking at the specs for your MIDI-to-CV interface, I noticed there is only 12 bits of resolution for a 10 octaves range. That will result in some notes being of approximate value. Wouldn’t have been better to use at least 14 bits, or better yet 16 bits for such a wide range ? 16 bit D/A converters aren’t that expensive these days, are they ?
With 12 bits, you have 4096 different voltage values divided by 120 semitones (10 octaves), which yields only 34 possible different voltage values per semitone. Even though you can get a precise 1 volt per octave out of this, some notes in between will suffer from imprecise voltage for accurate pitch.
With 16 bits, you’d get 65536 different voltage values divided by 120 semitones, which would allow 546 different possible voltage values per semitone. You’d be assured of perfect pitch for every single note !
Anyway, if I was making a MIDI-to-CV interface, I would make sure to build the absolute most precise one I can. Because what’s the use of having ultra precise and stable VCOs if only to send them approximated pitch control voltages ?
My take on it.
Although, your panel and PCB look mighty professional there. 
“So, your Moog© 921b clone is really an exact copy of MOTM© Ultra Vco©, but for copyright reasons you won’t post pics of the pcb?”
I saw that too.
They will be a decent oscillators, but it won’t be quite the same as real 921s.
Every circuit has natural characteristics where only the same design and execution can yield the same results.
921 sync is especially unique to the 921, but also the motms are probably quicker to process modulation.
They likely have a bit lower noise floor overall too.
But I do understand that finding / affording real 921s is difficult or impossible.
The Forbin board is probably closer to a real 921, albeit a bit bigger board.
I wish he’d have asked me for dual FETs, I could have saved him some grief.
Hi cloners,
reading the threads concerning the 921 clones and studying the schematics I’ve not understood yet why a “ringmod” - chip is used for syncing. Hard sync is easy: just reset the oscillator core in the middle of the cycle by an externel reset signal. Soft sync is a bit more tricky, determine the amount of impact of the external signal on the oscillator core. Soft sync can be achieved by a pot and/or transistor solution.
So a question to all 921 owners (originals or clones): What does a sync of a 921 do differently compared to hard or soft sync? If you need the modulator circuit just for “leveling” the external signal there might be easier solutions to achieve this.
With 12 bits, you have 4096 different voltage values divided by 120 semitones (10 octaves), which yields only 34 possible different >voltage values per semitone. Even though you can get a precise 1 volt per octave out of this, some notes in between will suffer from >imprecise voltage for accurate pitch.
You will reach the ‘in between’ 34 voltages values per semitone only if you use the pitch wheel on your MIDI keyboard to go
lower or higher the played notes. I did that in my tests and the ‘staircase’ effect in the rising and lowering
of the notes are not noticable even with 34 steps max. Smooth action !
And most of the time the MIDI notes that are received generate between 0.000 and 5.000vdc
in fixed steps of 1/12vdc per semitones.
So again the semitone’s 34 voltage steps are used only in pitch wheel demands or MIDI modulation.
JP
I understand what you mean, jpdesroc.
But I’ve just realized that there was one important word missing in my last post about voltage resolution per semitone: scaling.
With only 34 possible voltage values per semitone, the scaling of the whole 10 octave range, although correct at each octave extremities, might be off on a few notes in between them. Yielding an approximate keyboard scaling, instead of perfect pitch for every single note.
I have the same problem with my Korg Mono/Poly and my MFB Kraftzwerg mini modular. They both use 12 bits D/A converters to produce pitch CV, and it’s impossible to scale them perfectly for each semitone accross their many octaves range, even though they both can produce perfect 1 volt per octave when properly calibrated. Some notes in between octaves are not producing perfect pitch. Some notes are off by as much as +/- 6 cents !
That might not seem like much, and unless one has a " perfect pitch ear" it won’t probably be noticeable on its own. But when played against other instruments that are better scaled, you can clearly hear beatings between notes on some, and not on others. It’s much more noticeable, now that my Minimoog D has an internal optical key contact system (Kevin Lightner’s Opto-Key) that is using 16 bits D/A to produce pitch CV across only 44 notes ! Allowing a spectacular 1489 different possible voltage values per semitone !
I know it seems like overkill for so few octaves (3.5), but you can be sure of absolute perfect pitch for each and every single semitone with that. No approximation here!
I simply don’t like the idea of settling for something like 2.354 volts when I can have exactly the 2.358 volts required, that’s all.
Especially in 2012 with all those powerful and precise digital chips at our disposal for cheap.
I attempted to faithfully clone the 921b a few months ago..I spent many many hours re-creating the track pcb (double sided)
from whatever photos I could get..the most difficult Moog module Ive ever tried..tracks run under components , one has to deduct
routes etc. But I think I just about did it..in contrast the cloning of the 921a was quite easy.
But so far the 921b “rests in peace” stone cold nothing from the saw wave. After many late nights testings I think I’m coming to the conclusion that the 2N 3954 I got from flea bay may not be an actual 2N3954..nothing on the casing..I could be wrong but isolating the saw wave part of the circuit, its still dead. I even transplanted the 2N3954 into a copy Mini Moog osc I made earlier.. dead.
I tried 2 J102’s in the 921b…dead. But experimenting with my working 901bs Ive found that unless parts are identical to Moog’s you get bad upsets in results…2N3954’s are unobtainable..thats the problem..where-ever Mos lab gets them from who knows?
I have 2N3954s. Even some 412s and other dual fets.
Hi Analogmonster
The Moog sync is really quite different from the “others” in that it implements what is usually called a phase locked loop. It sort of measures the difference between the frequency of the “sync to” source and it’s own frequency and then generates an extra voltage that will drag its oscillator to the same frequency as the “sync” source. It is considerably more sophisticated approach than just forcing a reset of a sawtooth as per usual hard sync circuits but comes at the cost of a multiplier. For a more detailed explantation with the maths google “phase locked loop”.