DIY synth

[Autophase]

hi guys, i thought I'd put this in the Brewery as not to clog up the lab.
I was fortunate enough to have a moog prodigy in my studio this weekend, and i loved it.
I was even allowed to take a look in side.
Its got me thinking, I'd love to build a simple 2 osc analogue synth with a nice filter, like the moog prodigy, but preferably in a box format (ie: No Keyboard) and with midi input.
Have you guys encountered something similar to this in your years of DIY, is there a forum like this but more specific to DIY synths?

Cheers.

 

[letterbeacon]

You could try the MFOS Sound Lab - http://www.synthtopia.com/content/2011/11/22/mfos-sound-lab-mini-synth-mark-ii/

www.electro-music.com/forum is the place to go for your DIY synth needs.  I went from knowing nothing to building this modular synth in no time!

 

[zayance]

Looks terrific, Nice job.

 

[geoff004]

I've got a Prodigy - it's great.
I think re-making something based on this synth wouldn't be too difficult.  The schematic is available online - easy to find in a search.
And you could probably reduce the size easily, especially if you replaced some through hole components with surface mount.  Power on the Prodigy is +/-15 volts, and not a lot of draw.  You could probably cram the whole synth in to a couple slots of a 500 rack if you wanted.

As for adding midi, I don't have a lot of experience in that department (though I did add midi to my 808).  My guess it that's not too difficult either.

 

[etheory]

I have three projects on the go at the moment. An API compressor, a Calrec Pre/EQ and a customized 4 oscillator Minimoog Model D. I guess you might be interested in the latter. More details soon.... I'll be building in full MIDI support also.

It will also extend (but not modify the original features) of the Mini quite a bit, especially what I have in mind for the OSCs (sync/cross mod/restart/AM/PM/FM per OSCILLATOR).

 

[letterbeacon]

Looks terrific, Nice job.

Thanks!

As for adding midi, I don't have a lot of experience in that department (though I did add midi to my 808).  My guess it that's not too difficult either.

You could try the Midimplant http://www.midimplant.com I've no experience with it but others over on electro-music have.

 

[ej_whyte]

Not quite as DIY, but still cool:

http://www.doepfer.de/DIY_Synth_e.htm

 

[living sounds]

I've looked into this as well, and basically concluded that none of the standalone DIY synths sounds as good as the great sounding vintage monosynths. There are some very cool DIY modular projects, but it's a lot more work and you'll end up with a modular. ;-)

If I were to build a DIY standalone synth this one would be it:

http://www.musicfromouterspace.com/index.php?CATPARTNO=&PROJARG=SOUNDLABMINIMARKII%2Fpage1.php&MAINTAB=SYNTHDIY&VPW=1373&VPH=625

Some people seem to have had sucess getting a sweeter sound by replacing some TL072s in the oscillators with different chips, too.

 

[etheory]

I managed to successfully simulate the early revision 1 Minimoog Model D oscillator last night in LTSpice, and, unbelievably, it freaking oscillated PERFECTLY (the almost-totally-discrete version - not the later revision op-amp one) - including all those lovely warm waveform imperfections you'd expect from an early revision mini.

More soon....

 

[living sounds]

I managed to successfully simulate the early revision 1 Minimoog Model D oscillator last night in LTSpice, and, unbelievably, it freaking oscillated PERFECTLY (the almost-totally-discrete version - not the later revision op-amp one) - including all those lovely warm waveform imperfections you'd expect from an early revision mini.

More soon....

Now you got me interested!

 

[PRR]

> you could probably reduce the size easily

But the bigger the synth the better.

letterbeacon is going the right way. Only should stick some fake knobs on the blank panels until he gets around to putting guts in them.

 

[etheory]

Here is where I've gotten to so far.  I love that the triangle contains the little glitch at its half-cycle and the saw wave contains a glitch at the top of it's peak (just like the original).
I've read all too many articles about people cloning the later ua726 boards and "fixing the imperfect wave-forms".... !!!! If I wanted perfect wave-forms I could stick with any one of the many many VST instruments I use.
For me such a mammoth project is all for me about ugly, raw, fat, analog imperfection as much as possible.  I love that even the simulation is already far from "perfect".

I don't have a lot of time to rush this to a physical prototype oscillator (as I recently started another full-time job for a bit), but I'll start a new thread soonish to continue this crazy journey.  The plan is to build a 4-oscillator (modified yes, but I like to think in the spirit of the original) version of the Model D early edition, with the R.A.-style filter (fully matched transistors throughout) and the early revision oscillators.  I know many have tried and not finished before, but I'm confident that I'll get there within the year.

 

[living sounds]

Great! The triangle wave is what the Moog sound is all about IMO.

Are the early Mini oscillators identical to the first Moog modular oscillators (901a/901b)? Some people have reportedly succeeded in cloning those, but there are no complete useable PCBs availible. And they'll never stay in tune. ;-)

Another important thing for the early Moog modular sound are the CP3 discrete mixers. Are they in the early Minimoog as well?


It's hard to decipher the low-res schematics - did you use modern components or would it require sourcing some of the elusive ancient parts.

Anyway, this is exciting!

 

[Jonte Knif]

I got a set of thyratron tubes yesterday and built a simple oscillator with them. I got a quite nice sawtooth sound with 4 decades range. There is somewhat "flutter and noise" present in the frequency, and the amount depends on the tube. It seems that the thyratron trigger voltage has certain amount variance. Fun, lifelike, human sound. The cleanest tubes were clean enough for everything, and the "noisiest" started to sound a bit funny and hysteric. 

Any how, since my tube filter and tube ring mod work very well and there should be no difficulties to build a good tube VCA, I'm also now into synth building, just can not say with which schedule. All non-audio frequency stuff will be with ICs, but for the sake of some credibility I try to keep the audio path 100% tube. (and a lots of iron)

 

[etheory]

Great! The triangle wave is what the Moog sound is all about IMO.

100% agreed - it's a sound that people forget about - but for basslines for modern music, it's a REALLY useful tone - specifically with that reset click, it takes it from mellow to in your face.

Are the early Mini oscillators identical to the first Moog modular oscillators (901a/901b)?

Not at all really - they seem to be very very different beasts from what I'm seeing.

Some people have reportedly succeeded in cloning those, but there are no complete useable PCBs availible. And they'll never stay in tune. ;-)

Not necessarily true.  If you build them using modern methodologies (stable PSU, star grounding, stabilization via good opamps etc. etc.) then I've heard you can get them quite stable.  The rarity of the monolithic matched-transistor arrays and diode arrays is the real issue for those things.  Basically with some careful and well-placed engineering you should be able to retain 100% the sound, and gain some more stability - but people are rarely game to spend the long hours of messing around to actually achieve this.

Another important thing for the early Moog modular sound are the CP3 discrete mixers. Are they in the early Minimoog as well?

I don't think so.  The schematic I am using is basically the FantasyJackPalance one from here: http://fantasyjackpalance.com/fjp/sound/synth/synthdata/16-moog-minimoog.html - which simply mixes each oscillator (after some passive resistor dividing volume normalization networks for the different waveform types) through 25K pots, then 33K resistors, into a 10uF cap to remove any DC into the discrete filter.  The mixing is "passive" and directly after the 33K resistors in series with the wiper after each pot.

It's hard to decipher the low-res schematics - did you use modern components or would it require sourcing some of the elusive ancient parts.

The existing schematic that I linked to (low res to avoid a large file-size) is basically a 1 for 1 for the original, except since I didn't have an LTSpice model for the 2N4058 transistor, I substituted a BC557C - which doesn't seem too different enough to cause issue.  I also didn't have an LTSpice model for the matched e402 JFET - but pretty much any JFET model seems to produce oscillation in this circuit, just changing the center/base frequency of oscillation for any given control voltage due to the different switch-on/off points for the different gate voltages.  In my existing sim I used a pair of 2N5484 - but am considering a modern surface mount JFET pair here instead for the prototype.  Also it appears the CA3046 would be reasonably substituted by the modern NTE replacements - but I need more research on that.  Aside from that reasonable substitutions exist for all other parts.  In fact with a complement of polyester timing caps (or indeed WIMA ones) and 1% metal film resistors, a modern build should kick the crap out of an old one for stability (the modern parts have much much better tolerances overall).

Anyway, this is exciting!

Thanks!  Glad you are excited - cause I sure am.  It's hard to maintain it for long though.  By day I work as a computer graphics programmer for the feature film industry, so that's a good enough distraction to keep progress quite slow.

 

[living sounds]

I'm certainly looking forward to anything you come up with! I just hope it can be build with all the modern substitutions and not end up sounding like a Voyager... ;-)

I recently bought a Micromoog, which uses lot's of "bad" op amps and is scaled down on many levels, but once you change a DC blocking electrolytic taking away too much low end for a bigger one this thing with it's single oscillator pumps out an amazing bass. And the envelope is as fast and punchy as the SH-101s.
Also, the service manual is the most complete one I've ever seen, taking care of every eventuality, explaining every single circuit detail.

 

[peterc]

The existing schematic that I linked to (low res to avoid a large file-size) is basically a 1 for 1 for the original, except since I didn't have an LTSpice model for the 2N4058 transistor, I substituted a BC557C - which doesn't seem too different enough to cause issue.  I also didn't have an LTSpice model for the matched e402 JFET - but pretty much any JFET model seems to produce oscillation in this circuit, just changing the center/base frequency of oscillation for any given control voltage due to the different switch-on/off points for the different gate voltages.  In my existing sim I used a pair of 2N5484 - but am considering a modern surface mount JFET pair here instead for the prototype.  Also it appears the CA3046 would be reasonably substituted by the modern NTE replacements - but I need more research on that.  Aside from that reasonable substitutions exist for all other parts.  In fact with a complement of polyester timing caps (or indeed WIMA ones) and 1% metal film resistors, a modern build should kick the crap out of an old one for stability (the modern parts have much much better tolerances overall).

I used BC550/560 for the transistors in mine. The FETs were J112s and the 3046 can be replaced by the current CA3146.
I bought my J112s from RS Components and the 3146 from my local component agency.
Futurelec have the 3146   
http://www.futurlec.com/Linear/CA3146Epr.shtml

The uA726 is not available, so it is a good idea to stay away from that circuit.

Also have a look at:

http://www.groupdiy.com/index.php?topic=12639.0
http://www.groupdiy.com/index.php?topic=44081.msg552717#msg552717

Peter

 

[buildafriend]

I built a sound lab ultimate, I would sell it. That build was insane, I'm done with that.

 

[etheory]

I used BC550/560 for the transistors in mine. The FETs were J112s and the 3046 can be replaced by the current CA3146.
I bought my J112s from RS Components and the 3146 from my local component agency.
Futurelec have the 3146   
http://www.futurlec.com/Linear/CA3146Epr.shtml

The uA726 is not available, so it is a good idea to stay away from that circuit.

Cheers Peter!

My intial tests seem to indicate that BC550C/560C's might not be the best choice for a modern Model D.  First thing first - yes they WILL work (as you clearly know ;-) ), but I don't think they actually work as the circuit originally intended.

The simulator shows significant waveform deviations at the precise oscillator reset phase.
The original contains some relatively intricate and complex wobbles and other nice deviations from smooth, but with BC550/560's it's an all-together cleaner and more "boring" waveform.
I don't think it's the best idea to use them as direct drop-in replacements.
Even though oscillation is maintained, the centre-frequency (i.e. the frequency of oscillation for any given input voltage) changes significantly, and the details of the reset pulse differ so significantly that I think it would "sound different".

Food for thought!  But, regardless, I'll bet yours still sounds kick-ass.  Did you modify any surrounding circuitry to account for the transistor parameter variations between BC550/560 and 2N4402 2N3392?

 

[peterc]

First thing first - yes they WILL work (as you clearly know ;-) ), but I don't think they actually work as the circuit originally intended.

Minis actually used different transistors in different versions, TIS96 (?) in the earlier ones, and the specs of the BC's are within spec, just better than the old ones. Jeurgen Haibles also used them in his Mini versions.

The raw waveforms sounded very similar to my Minimoog as well.

You could also try lower spec transistors like 2N3904/3906.

Remember also, simulators tell pork pies ....

Peter