Author |
Message |
TMA-1
Joined: Jun 29, 2014 Posts: 4 Location: USA
|
Posted: Mon Jun 30, 2014 12:29 am Post subject:
Creating a polysynth |
|
|
Over the years I have bought and sold most of the analog polys within my price range. I have found myself unsatisfied with their tone. However, I cannot afford a Prophet 5 or a CS80.
So I decided: I know how to solder. I can read diagrams (mostly). I've built guitar amplifiers, effects pedals, and synth modules. Why don't I build my own polysynth? Others have done it.
So I was hoping someone could give me some pointers on this quixotic project. These are the aspects I'm not sure how to tackle presently:
Keyboard scanner / voice allocation. I can't seem to find any circuits to use.
Power supply. Everything modular I've built has been +/-15v power. Is this practical in a keyboard poly? I'm not against an external power supply, I don't need to take this on tour.
Appropriate circuits. I had planned on using this SSM2040 clone by Rene Schmitz, but as I said above, I have no idea how appropriate +/-15v is for a self-contained polysynth.
Aftertouch. I would like to have aftertouch, but it's not a deal-breaker. I think it would probably be wisest to build the synth without it first, then add it later.
Scavenging parts. Which brings me to my next question. DX7s are a dime a dozen and have aftertouch. Would it be possible to steal the keyboard scanner / voice allocator and use it? Obviously the DX7 is digital, but my understanding is that many key scanners use CMOS or TTL. Can those only drive DCOs, or could I control VCOs with it as well?
The CS80 VCO. I found this CS80 VCO clone, but unless there's something wrong with my brain it only has a sawtooth out. Did the CS80 have waveshapers? If I go with this VCO, what waveshaper makes sense? I also just noticed it seems to use +/-15v, so I guess that answers that question.
Thanks in advance for any advice you can give me. I'm sure I'll come up with more questions!
EDIT: As an aside, I know I'm mixing the Prophet 5 filter with the CS80 oscillator. I had planned on adding multiple filters later, but I figured I would take this one step at a time. If you think there are more simple VCOs and VCFs I should use on my first project, please do recommend them! |
|
Back to top
|
|
|
gdavis
Joined: Feb 27, 2013 Posts: 359 Location: San Diego
Audio files: 1
|
Posted: Mon Jun 30, 2014 10:56 am Post subject:
|
|
|
By the time you finish this project, with the time and money you will have put into it you could have just saved up for a prophet 5. It's not easy and it's not cheap.
If what you want is the sound, do you really want to be concerned with such mundane things as a keyboard scanner? For my project I just receive midi and use another keyboard controller. Voice allocation is done in a micro controller/processor.
I don't know if Oberheim is what you're looking for, but there's a OBX voice clone project over at muffwiggler called crowbx, you might want to look into that. The same guy is also starting on a CS-80 clone.
If you really want to build something, start small, maybe a few modules or something. A poly synth is a whole other beast. You can find service manuals for the old analog polysynths online that contain complete schematics and lots of description of how they operate. Spend some time studying those to get an understanding of what's involved.
+/-15V is fine, you'll also need 5V for the digital logic.
There are many filters, that's largely a matter of taste. There are many VCO's also, for a poly you'll want one that's very stable and compact. _________________ My synth build blog: http://gndsynth.blogspot.com/ |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon Jun 30, 2014 3:00 pm Post subject:
|
|
|
Ditto what gdavis said.
Stability and thermal compensation is critical IMO or the voices won't sound the same. Even the filters need to be tempco "helped".
I've read from other folks who've tried to build analog polysynths. They are usually underwhelming, not patchable, expensive, heavy, power mongers and large.
At the risk of heresy, I'll ask: Have you considered a digital polysynth? Poly is much much easier to do in the digital domain. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
elmegil
Joined: Mar 20, 2012 Posts: 2177 Location: Chicago
Audio files: 16
|
Posted: Mon Jun 30, 2014 9:57 pm Post subject:
|
|
|
Even if you don't do a full digital polysynth, it seems to me that a synth that already provides the features you want viz polyphony and aftertouch, going through a MIDI converter with appropriate programming (you could set up a dedicated one to cover whatever control voltage outputs you might imagine with any of several microcontroller options) would be a far simpler means to the keyboard end of things. |
|
Back to top
|
|
|
TMA-1
Joined: Jun 29, 2014 Posts: 4 Location: USA
|
Posted: Tue Jul 01, 2014 2:42 am Post subject:
|
|
|
I think I'm going to get my feet wet by building two or three Thomas Henry voices, since they are so simple and sound very good to my ears. There is no reason to try to build a hybrid P5 / CS80 right out of the gate.
elmegil wrote: | Even if you don't do a full digital polysynth, it seems to me that a synth that already provides the features you want viz polyphony and aftertouch, going through a MIDI converter with appropriate programming (you could set up a dedicated one to cover whatever control voltage outputs you might imagine with any of several microcontroller options) would be a far simpler means to the keyboard end of things. |
Yeah, I thought about it overnight and it makes much more sense to build a rackmounted or desktop poly than keyboard poly at this point.
What would you recommend? At this stage of the game, I don't think I should get involved in anything other than pitch and gate, but I don't know a lick of code, and I'm completely out of the loop on microcontrollers. I've heard Mutable are doing some cool things, but I don't know what the applications are there. |
|
Back to top
|
|
|
elmegil
Joined: Mar 20, 2012 Posts: 2177 Location: Chicago
Audio files: 16
|
Posted: Tue Jul 01, 2014 5:55 am Post subject:
MIDI 2 CV |
|
|
I have Hexinverter's MIDI2CV with alternate firmware provided by rpocc@muffwiggler. Here's the current end of the thread discussing it: http://www.muffwiggler.com/forum/topic-93855-140.html. It needs a PIC programmer to use the alternate firmware, but I understand he may do a new release with that firmware instead.
I also have built my own, jumping off from Thomas Henry's MTS-100 but using an Arduino programmed AVR as the processor, and using only CV/Gate/Trigger outputs (so I cut a few pices out of the output side). That was a good learning experience, and I posted in a fair bit of detail about it on this forum somewhere, a couple of years ago. |
|
Back to top
|
|
|
TMA-1
Joined: Jun 29, 2014 Posts: 4 Location: USA
|
Posted: Tue Jul 01, 2014 3:09 pm Post subject:
Re: MIDI 2 CV |
|
|
elmegil wrote: | I have Hexinverter's MIDI2CV with alternate firmware provided by rpocc@muffwiggler. Here's the current end of the thread discussing it: http://www.muffwiggler.com/forum/topic-93855-140.html. It needs a PIC programmer to use the alternate firmware, but I understand he may do a new release with that firmware instead.
I also have built my own, jumping off from Thomas Henry's MTS-100 but using an Arduino programmed AVR as the processor, and using only CV/Gate/Trigger outputs (so I cut a few pices out of the output side). That was a good learning experience, and I posted in a fair bit of detail about it on this forum somewhere, a couple of years ago. |
Whoa, this is great! Thanks so much for this link. |
|
Back to top
|
|
|
m.o
Joined: Jul 05, 2014 Posts: 44 Location: Sweden
|
Posted: Sat Jul 05, 2014 11:22 pm Post subject:
|
|
|
Hi,
I have been slow-poking a poly-synth kinda project for the last 6-8 months or so.
I don't know if it'll ever be completed but I'm learning things all the way so it's all good even if I fail or realize it's not possible for some reason.
(My project is based off of the mid-80's 'cheapo-polys', one of the characteristics of those were they basically had only Oscillator+Filter+VCA in hardware, all envelopes and LFOs etc were software-controlled.)
Almost all polysynths and most (monosynths with patch-memory) use a CPU to generate all the needed CVs (control voltages).
To generate the CVs you need a DAC (digital to analog converter) with sufficient resolution (a MIDI2CV essentially).
A common technique to get voltages out to all the destinations (for a 6 voice 1-osc synth this can be 20-30, for a 2-osc synth the number increase to around 50 or so) was to "multiplex" the analog voltages at a decent frequency (at least a couple of 100 times per second).
Scanning front panel controls as well as scanning a keyboard is done in similar ways (but reading voltages instead of sending them ofc).
My biggest advice so far is:
- Download service manual with schematics for as many synths as you can find; Oberheim OBX / OBXa, prophet 5, Roland Junos, Akai Ax 60/80.
Study these carefully and try to understand (at least on a higher level) how they do all the "tricks".
- Check out the MFOS (music from outer space) website for much insight in all things synth-circuit.
To try and address some of your questions:
Keyboard scanner / Voice allocation: These can often be separate tasks, although I think some actually do simple hardware voice allocation schemes this is well suited to be done in software. Keyboard scanning; This is not something I'm trying to do (going for midi for keyboard control) so I don't know very much but as I understand it is often a matter of having the keyboard arranged in a 'matrix' and scanning through this to detect the states of all keys.
Power supply: Sure, many synths use +/- 15V, but you'll usually also need other (lower) voltages here and the, +5 for the digital bits, sometimes there is +/- 5 or 6V used in certain circuits. Some have +-15 or more globally and the have local filtering and regulation to lower voltages on the voice card.
Circuits: Give it a go, especially if you can do it SMD (there are a lot of transistors in that circuit). The important this is that you get something that works and you can practically build 5-6-8 of and get consistent. Filters seems to be the most component-intensive part of a voice (I've done a SEM-like filter based in lm13700's).
Scavenging: I don't think you can isolate and "steal" the scanning/voice allocator from a DX7, the keybed with aftertouch sensor yes, but the "intelligent"/active bits are certainly done on the CPU/Mainboard.
CS80 VCO: There are examples of wave-shaper circuits readily available (images.google.com if your friend ), the circuit you linked there I think is an implementation of the core principle only (saw).
That circuit is also without either linear-to-expo converter or any temperature compensation.
I think it is possible (maybe even pretty clever, I think korg used it) to build a synth that uses a linear voltage for frequency, but this I think brings the problem that you need to double the voltage for each octave and the precision can be a problem over the whole scale (not sure exactly, I've gone for 1V/octave exponential scale).
If you're going to do this I think you need to do it with the attitude of learning, to prove that you can do it, not expecting cheap/fast/good
/ rant off |
|
Back to top
|
|
|
TMA-1
Joined: Jun 29, 2014 Posts: 4 Location: USA
|
Posted: Sun Jul 06, 2014 2:49 am Post subject:
|
|
|
Thanks for your reply m.o! I am a big fan of Ray Wilson's site, I have looked at a lot of schematics there and will undoubtedly be ordering parts from him for some of the Thomas Henry circuits. I'm having a heck of a time finding appropriate PNP transistor pairs though! And I haven't even given the Rene Schmitz 2040 schematic a close look, I just know there are a ton in the circuit, as you noticed.
SMD soldering is not really something I want to deal with... I know there are many parts that are getting impossible to find in through-hole, but I just like working with it so much more.
I think I am definitely going to go with the updated firmware MIDI2CV, I agree there's no real reason for me to try to tackle key scanning/voice allocation. So I gave up on the idea of scavenging a keyboard.
I figured I could find a waveshaper circuit to use, I just wondered if anyone knew the exact schematic for the Yamaha one, as I wanted to get as close as possible to the sound of the CS-80 waveforms. But I think I may give up on that as well and wait for the CS(R)-80 to be done. Big thanks to gdavis for pointing me to that.
I would love to see or hear your project, if you have any videos or pictures or anything! How are you powering yours, by the way? I have a power supply which outputs +/-15v and +5v, but I don't know the number of VCOs, VCFs, VCAs, envelopes, etc. that it can power at once safely. |
|
Back to top
|
|
|
m.o
Joined: Jul 05, 2014 Posts: 44 Location: Sweden
|
Posted: Sun Jul 06, 2014 1:25 pm Post subject:
|
|
|
TMA-1 wrote: | Thanks for your reply m.o! I am a big fan of Ray Wilson's site, I have looked at a lot of schematics there and will undoubtedly be ordering parts from him for some of the Thomas Henry circuits. I'm having a heck of a time finding appropriate PNP transistor pairs though! And I haven't even given the Rene Schmitz 2040 schematic a close look, I just know there are a ton in the circuit, as you noticed.
SMD soldering is not really something I want to deal with... I know there are many parts that are getting impossible to find in through-hole, but I just like working with it so much more.
I think I am definitely going to go with the updated firmware MIDI2CV, I agree there's no real reason for me to try to tackle key scanning/voice allocation. So I gave up on the idea of scavenging a keyboard.
I figured I could find a waveshaper circuit to use, I just wondered if anyone knew the exact schematic for the Yamaha one, as I wanted to get as close as possible to the sound of the CS-80 waveforms. But I think I may give up on that as well and wait for the CS(R)-80 to be done. Big thanks to gdavis for pointing me to that.
I would love to see or hear your project, if you have any videos or pictures or anything! How are you powering yours, by the way? I have a power supply which outputs +/-15v and +5v, but I don't know the number of VCOs, VCFs, VCAs, envelopes, etc. that it can power at once safely. |
(Let's see if I can manage to upload pictures...)
I haven't recorded any sounds yet, but I have done a couple of "integration" tests, the first in the picture was just 4 oscillators (at the top) and 4 VCAs, I use an Arduino for midi input (lower right), DAC control (the breadboard under the arduino) and multiplexing (the blue board, I use a CD4051 for 8 outputs).
I used this to test out my arduino code, and it actually worked eventually (sawtooth only though).
I have since started using another VCO design, and I'm on the second 'major' version of my vcfs...
I think my biggest setup so far uses less than 400 milliamperes, so far I'm just using a lab supply with dual rails, I run everything on +- 12V partly because it feels safer testing together with my eurorack modules..
Regarding parts;
For OTAs (used in many filters and vcas) you often see 3080's, these can often be raplaced with LM13700's (which is an improved double version of the same) which is still sold in through-hole.
Transistors; The 3046 is a popular choice in many circuits (it's a 14-pin ic with 5 NPN's of which 2 are matched) - these are hard to find in DIP, but are available in the 'largest' SMD size SOIC/SO-14, soldering this size is quite doable. You can get adapter boards for these where you solder them onto a small pcb which then breaks out the pins to standard through-hole spacing (I think Sparkfun has these).
I recently discovered the BCM-847-BS (NPN) and BCM-857-BS (PNP), these are 6-pin packages that has two matched transistors inside them, and they're something like $0.50 each - the bad new is that these are only available in really small SMD packaged (SOT-363), but I have actually managed to etch adapter boards for these as well. Soldering them is really fiddly, but it actually worked - the trick was to first solder them on but this left a blob shorting the legs, but with a quick application of a narrow soldering-wick it actually became ok (see second picture).
- I think these would be perfect for Schmitts' 2040-clone.
- I now use these in my 'second generation' designs, VCO, VCF and VCA.
- Have also found temperature -compensation resistors in SMD size (about $1-$5 each), so I made a version of my adapter boards with pads for these beside the SOT-363)
The alternative that many people do is to take normal through-hole trannies and hand match them using a contraption like the one described on MFOS.
Oh, and another tip for circuit examples it to check out the "EFM synthesizers", I use a version of the 3531 VCA both as dedicated VCA and for CV-controlling resonance in my VCFs:
http://synthdiy.fonitronik.de/forums.html/efm/
Description: |
Early integration test without VCF |
|
Filesize: |
97.72 KB |
Viewed: |
158 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
Description: |
|
Filesize: |
172.73 KB |
Viewed: |
178 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
Description: |
|
Download |
Filename: |
adapters.pdf |
Filesize: |
5.56 KB |
Downloaded: |
180 Time(s) |
|
|
Back to top
|
|
|
elmegil
Joined: Mar 20, 2012 Posts: 2177 Location: Chicago
Audio files: 16
|
Posted: Sun Jul 06, 2014 1:59 pm Post subject:
|
|
|
m.o wrote: |
The alternative that many people do is to take normal through-hole trannies and hand match them using a contraption like the one described on MFOS.
|
There's a much simpler circuit for matching here:
https://home.comcast.net/~ijfritz/MiscProj/transmat001.pdf
I've wired this up across two 4PDT switches and a socket, and it works great (when you have patience for doing the matching in the first place). No PCB necessary, though some folks on Muffwiggler have created one as well (google for Ian Fritz Transistor Match PCB and you should find it pretty fast). |
|
Back to top
|
|
|
jackdamery
Joined: Apr 26, 2010 Posts: 75 Location: UK
|
Posted: Fri Jul 25, 2014 6:44 am Post subject:
|
|
|
Welcome to the club!
Keyboard scanner / voice allocation
I learnt about this from this tutorial. It's not velocity sensitive, but if you want to go down that road you might as well just use MIDI from a dedicated keyboard controller. If not then the best way to go is a microcontroller, which you will have to use anyway for MIDI in.
http://www.codetinkerhack.com/2012/11/how-to-turn-piano-toy-into-midi.html
One thing I learnt was that an Arduino can do almost everything! I bought an old farfisa organ keybed and used a shift register and Arduino to scan the keys and then convert the notes to MIDI out, but also CV out from a DAC for a VCO as well as a digital envelope for a VCA/VCF through another DAC.
Power supply
As someone mentioned previously check out the service manuals of old synths. +/-15v was pretty universal although lower voltages were sometimes applied to various bits of the circuit esp. 5v for digital ICs. Even DX21 used that power supply. You can buy OEM dual power supplies suitable for embedding from most larger electronics retailers, speak to a rep if you need help choosing.
Project management
I think if you're serious about building a polysynth you should approach it focused. It shouldn't have the all over the placeness of a modular synth.
The idea is to get a design that works (i.e. a complete voiceboard) and then make it as small as possible, then duplicate it for however many voices and build the control hardware/software around that.
For inspiration check out this tumblr
http://polysynth.tumblr.com/ |
|
Back to top
|
|
|
rhdf
Joined: Jul 11, 2014 Posts: 39 Location: Sweden
|
|
Back to top
|
|
|
m.o
Joined: Jul 05, 2014 Posts: 44 Location: Sweden
|
Posted: Fri Jul 25, 2014 7:43 pm Post subject:
|
|
|
I realize some of this might sound like I trying to discourage, it's not meant like that, just venting some of my own frustrations - In the end I kind of enjoy the problem solving and figuring things out.
You also need to eventually consider how things are going to come together both mechanically and electrically.
When you go from just a few modules that can easily be handled to multiples of 4-8 this becomes an issue already during testing/development.
If you're doing voice-boards, how are these going to be mounted/connected toether, do you use a large "motherboard" or are you going to use ribbon cable/connectors? How is power going to be distributed?
It's also easy to forget "auxiliary" circuits, you need to have electrically controlled switches (or multiplexers) for waveform selection, all places where you need to control signal level needs a VCA of some sort.
@rhdf:
Those SOT-457 will definitely make your life easier, I think I might order some too next round
Aren't there also larger "diode arrays" available with like 4-5 diodes in one capsule (14-pin)?
For my own project, well it's not moving fast, but I did get some time last couple of weeks to fiddle about with it.
For my new oscillators, I think I've got the hang of the tempco values now, still might need another revision of the board (It might be necessary to add a fine-tune trimmer in addition to the "main" one).
The new filters has taken quite a lot of time too, I've had trouble getting the resonance control right.
Last couple of days I've been working on the arduino control program and added a couple of test-modes to make calibration easier.
I have been sketching (in KiCAD) on a complete "voice card" for a 2-osc synth setup, but it tends to become pretty large (hard to get everything onto a 10 x 16 cm board) and I hesitate also because I know from experience how I tend to discover small but critical changes after I've committed and etched a bunch of boards... |
|
Back to top
|
|
|
rhdf
Joined: Jul 11, 2014 Posts: 39 Location: Sweden
|
Posted: Fri Jul 25, 2014 9:03 pm Post subject:
|
|
|
m.o wrote: |
Aren't there also larger "diode arrays" available with like 4-5 diodes in one capsule (14-pin)?
|
It might be some out there, but i have no clue if the specs of the separate diodes are "OK". Some of the ones i've seen also have strange internal connections. Not 4-5 in series or just parallel but rather 4 with common anode and so on.
m.o wrote: |
I have been sketching (in KiCAD) on a complete "voice card" for a 2-osc synth setup, but it tends to become pretty large (hard to get everything onto a 10 x 16 cm board) and I hesitate also because I know from experience how I tend to discover small but critical changes after I've committed and etched a bunch of boards... |
Why not split it in 2 and put them on top of each other. Double surface-space (and kind of modular, if you want to rebuild something you only have to redo that card) |
|
Back to top
|
|
|
|