PAiA 9700s Page

I decided to start a page on the 9700 modular synth. There seem to be very few resources out there for builders of this kit – I just finished assembly of mine and I’m starting to calibrate and test, so I thought it would be useful to put a permanent page together to publish what I find.

The kit is available from PAiA here.

After I finished the build it wasn’t immediately apparent how powerful the modular synthesis approach really is. I’m used to the more traditional keyboard synthesizers like the Juno 106 or Jupiter 8. These offered a set of “modules” in a fixed configuration and let you tweak the various parameters to get the kind of sound you wanted. The modular approach is like lifting the lid on the Juno, grabbing hold of the innards and squishing them together like jelly. You can still tweak the parameters, but now there are no rules as to how one module can influence the other – you can even mix ‘n’ match audio and control signals. So the modulator could be used as another oscillator and mixed in at some point.

Or not.

Now that my eyes have been opened to this new way of thinking I keep getting flashes of inspiration of how I can re-route the signals – something I never got with the software simulations.

---

Testing the VCA ADSR Function

I had few problems with the kit, the VCF had some problems initially, but removing it out and re-seating it in the case got it working. The VCA seems to have a slight problem in that the L+R mix control is biased more towards the L side than the R. That is, when the knob is a t 12 o’clock the R signal has disappeared completely. I need to scope out the two opamps around connection points C+D to find out why.

In the meantime, I got Scott from PAiA to verify these scope traces of the ADSR function. I connected to the scope to the envelope generator output and adjusted the four controls one after the other…

---

Test Patch #1 – Weird Pulsy Sweepy Thing(tm)

The first thing I need to add is my test patch – I was goofing around the other night and accidentally put together something that sounded really good. Here is an image of the front panel:

Click to open patch image

This patch uses the VCO modulator to create the rhythmic PW pulsing effect, then the VCF modulator in cycle mode to create a simultaneous slow rising and falling sweep. The MIDI Trigger and Gate outputs are patched through to the ADSR in the VCA.

Note: I accidentally wired the switch backwards on the VCA, so even though it is set to ‘Cycle’ it’s really in ADSR mode.

And here’s the patch in action…

---

Driving the Pitch CV from the Theremax

I hooked up the Theremax theremin to the CV of the VCO just to see what would happen.
(The oscilloscope is a general USB module from Measurement Computing.)

The only problem is that the CV out from the Theremax is in the 0-5v range, whereas the 9700 expects a 0-10v signal. So I might have to rig up either an op-amp, or maybe just an optoisolator to do the conversion. Means I’ve got to find 10v from somwhere inside the 9700…

UPDATE: Turns out I was right about the control voltage difference. I must have missed the memo on the PAiA forum, but Scott posted me the link to this article on the Theremax “booster” circuit that makes the output more compatible with the 9700.

http://www.paia.com/talk/viewtopic.php?f=6&t=102&sid=675992a51e0446f381b85d33499d7df6

---

Problem with VCA mixer

It looks like I’m still having a problem with the VCA L/R mixer. The R channel is only coming through at about 50% and there is appreciable noise on this channel (and not the good kind!).

The following video shows the problem. In it I’m sweeping from L to R between 2 identical triangle waves. The R output is noticeably attenuated.

UPDATE: Turns out I had wired 2 of the ‘X’ terminals on the front panel incorrectly (blue and black wires). This image shows were they were, and were they should be. Moving them along fixed the issue. Thanks again to Scott at PAiA for helping out.

---

Tuning the oscillators

I just spent the last week in a cabin in the depths of Michigan. I took the 9700 with me along with my laptop running some multitracking software. It was my goal to record some music just using the synthesizer and simple effects.

The first thing I tried was Equinoxe Part 7 by Jean-Michel Jarre. This had some interesting elements in it and required that I build a CV processing circuit to scale and offset the pitch bend CV control signals coming from the MIDI2CV controller:

One of the problems I discovered early on was that tuning both oscillators to track correctly was difficult. A really useful tool for this is available on this page:

http://www.techmind.org/audio/

Download the Frequency counter application, this will tell you how close the scaling is on your oscillator. The best way to set the scale is to tune the root note to ‘C’ then play one octave above and note how much shift there is in the upper C. If the shift is positive (sharp) then actually set the scale to make it sharper (counter intuitive I know) then drop back down to the lower C and use the front panel pitch control to move the whole range down.

Doing it this way I was able to close the octave gap to only about +3% error quite quickly and get both oscillators to track musically over a 2.5 octave range. I’m sure with more time I could have got this closer.

There was a further problem in that some of the effect tracks in Equinoxe 7 operate over more than 4 octaves, and so keeping them in tune required sampling and scaling the note range.

I’m currently planning on building a CV controller that can parse MIDI and “listen” to the output of both oscillators in a feedback loop, adjusting the CV out to get the full tracking 8 octave range. I was going to build it from scratch, but then I found this excellent page that has almost everything I need already written. The 16bit DAC chips are expensive, but I was able to sample a couple for free.

---

UPDATE – Tuning app (copied from the blog posting I made in February ’10)

The PAIA 9700s is a compact analog modular music synthesizer, available as a kit. It took me about 3 weeks to get everything constructed and installed in the case.

After building the next task is to set the scale trimmers on both voltage controlled oscillators so that they stay in tune over the largest range of musical octaves. This process involves going “back and forth” between the scale pot and front tuning knob until you get both oscillators tuned. It’s a difficult task as it’s not always obvious how the oscillator tracking is moving around when you alter the scale.

I found that the best way to set the scale is to tune the root note to ‘C’ then play one octave above and note how much shift there is in the upper C. If the shift is positive (sharp) then actually set the scale to make it sharper (counterintuitive I know) then drop back down to the lower C and use the front panel pitch control to move the whole range down.

On my PAIA 9700s page I mentioned a neat tuning utility that could help with the process. I contacted the author; Andrew Steer, as I was hoping to incorporate his source code into something I was creating to help tune my PAIA 9700s synthesizer. He sent me the routine that detects pitch – it was in Borland C++, but it turned out not to be too difficult to replicate the code in C#. I added MIDI support and a graphing function to assist in the tuning.

Now all you need to do is connect the first oscillator output to your mic in, click “Start” to see the signal in the oscilloscope window, and adjust the mic input level to get a clean signal.

Then you click start test, and it will send MIDI data out of the default MIDI port (which you should connect to the input of the 9700) and read the frequency output. Once it has completed a pass of oscillator one, it will prompt you to connect up oscillator two, and repeats the test. Finally it plots a graph of the two oscillator responses against an ideal curve. (Log plot is also available).

In this way you can see the changes as you make them.

Download the exe here.