My new wind synthesizer project

This is what I have, to start with.

After having built and used the original Bergfotron, I have come to the conclusion that the modular format doesn't suit my working/playing style at all. It's simply too slow and tedious to try out new sounds on a traditional modular synth. I want something that is very quick to work with. I want to be able to try out new sound ideas by just turning knobs and flicking some switches. Therefore I have thought long and hard about alternatives. Actually, over the last five years I have made drawings and designs for no less than four synthesizer concepts. The AMORE system was to become a fruit of the first of these. None of these concepts were finalized or built but with the AMORE system I could start to build modules that could become part in a future synth. So subsequent concepts were naturally based on AMORE boards.

Too large or too basic

As I play wind controller, my new synthesizer should be optimized for wind controller playing. It should be portable and not as large and heavy as the original Bergfotron turned out to be. Ideally it should fit in a briefcase like the EMS Synthi AKS. Also, it mustn't bee too time-consuming and expensive to build. But on the other hand, it needs to have more than a bare-bones synthesizer voice.
My first design turned out too ambitious and hence would never fit in a briefcase, not even the largest kind. I realized that I had to think very hard about what modules and functions would contribute most to the sound in a lead/section instrument like a brass or woodwind instrument. Naturally it should not only produce imitative sounds but it should have the same musical function as those instruments traditionally have. So no need for polyphonic pads or sequencer style drones.
The second attempt did fit in a briefcase but I realized after a while that it was really too limited.

The third attempt was very interesting but I had chosen not to include a feature I really wanted because otherwise it would have been far too complicated. As it turned out, it was still a bit too much to fit in a briefcase.

The fourth and hopefully final concept was born when I devised a new, much simpler way to do the feature I had to leave out from concept three. Now you wonder what that feature is. Well, I'll tell you later. Right now I only note that this feature has it's own limitations on the synth's complexity. So I had to weed out some functions from concept three.

A great advantage of the AMORE concept is that you can test the modules and listen to their sound without building an actual synth. You just plug them into the AMORE Starter Kit. I also have the AMORE Exerciser so I can patch up combinations of two boards and the VCO and the LFO in the starter kit. This can give a good feel for which modules are the most desirable in the synth you are going to build. For me it of course meant that I first had to design and build all the AMORE modules that I might want to use in the synth.

What modules?

A solo-type instrument mustn't be too complicated. You must be able to re-create the patches quickly and anyway, there are diminishing returns from more modules at a certain point. And the front panel can't be too large if the synth should be truly portable. For this and a technical reason I'll explain later, there is a firm limit to the number of parameters the synth can use. This number was originally 16 but I was later able to extend it to 20 parameters. Doesn't that still sound very low? Well, since the breath pressure from the wind controller replaces an envelope generator, there would not be a need for one in the synth. So there you save at least four parameters. And the bite sensor in the wind controller is used to generate vibrato, so no LFO is needed for that. More parameters saved. Still an LFO could be useful for other things of course. It felt natural to chose the Minimoog concept, where VCOs can be used as LFOs and there are no dedicated LFOs.

But how many VCOs? As a solo instrument, you really don't need more than one. If you add a second VCO, you can do either dual-VCO sounds or use the second as an LFO, for audio modulation or sync'ed to the first VCO. This was my original plan.

Now we come to the filters. How many and what type? Since wind instrument sounds might need a noise component, there should probably be separate filtering for the noise. And some resonant filter for imitating the formant resonances in acoustic instruments. But the dual resonant filter uses up no less than six parameters! That's a problem. Maybe just use one then? Or something more powerful that gives better use of the parameters?

After playing around a bit with my VCP module, I realized that it could produce many different effects with few parameters. In addition to the classic sweeping phaser effects, it can produce comb filter effects that can sound vocal-like. I decided to gang it with a VCX module to be able to control the wet-dry mix. That would give a lot of possibilities with just three parameters. In the end, I simplified it even further. Testing revealed that both wet/dry mixing and feedback at the same time didn't add much sonic possibilities. So I'll use the same parameter to add either feedback or wet/dry mixing.
The VCP can give some of the possibilities of a formant filter and with the multimode filter set to bandpass, a further resonance could be added. So I decided to skip the dual bandpass filter.

After this there was still a couple of unused parameters. What more should I add?
I decided on a third VCO, to make the synth at least as powerful as a Minimoog. An extra VCO could be useful for section-type sounds. Also, you cold sync two VCOs for more complex solo sounds and use the third VCO for modulation. This is starting to sound interesting!

Finally I settled on a preliminary module list:

WVCO (short for Woodwind VCO)
DVCO (short for DeLuxe VCO, which I haven't designed yet. In the meantime, think of it as a second MVCO)
MVCO (Minimoog style VCO)
MVCF (Minimoog style VCF)
SVCF (SEM style multimode VCF)
VCP (my VC-phaser with a dedicated VCX for wet/dry mix)
MVCA (Minimoog style VCA as final VCA)

It's got to be expressive!

As a solo type instrument, this synth has to be as expressive as possible. The breath pressure should not only affect the volume. I decided it should be able to affect any parameter. Note number should also be able to affect all parameters, to balance the sound over the instrument's range. I also devised a parameter I call “pedal”. This can either be a wah-wah type foot pedal or a knob used for morphing the sound. So this controller must also be able to affect all parameters. All these modulations will need knobs on the front panel so that's another reason to keep the total number of parameters down.

For the synth to be truly expressive the mixing between the modules and the strength of certain modulation signals must also be voltage controlled. So this calls for a certain number of VCAs and/or crossfaders, which also use parameters. After some thought I arrived at two additional VCAs and three crossfaders (VCX).

Painless patching

I wanted to be able to create sounds quickly. From the old Bergfotron I had realized that patch cords are simply too slow and tedious. The synth had to be pre-patched. But wouldn't that make it too inflexible? Yes, it would. Especially if you want to cram the most possibilities out of a limited number of parameters (and hence a limited number of modules).

So what to do?

Well, I settled on a form of automatic re-patching system. A bit like the “algorithms” on the Yamaha DX-7. But here the algorithms would consist of different analog modules instead of the identical digital blocks on the DX-7.

But how many algorithms? Creating a switching system with a lot of algorithms would be complicated. Not to speak of communicating them to the user.

I decided to try four algorithms. But how do I know that they cover all the possibilities I want?

I drew some algorithms with the modules that I wanted to use. Then I made a list with the functions that I wanted and noted down for every algorithm, which of these requirements it satisfied:

VCOs divided between MVCF and SVCF (19, 24)
Mix WVCO and DVCO with balance control (13, 20)
VCO to MVCF FM (19, 20, 24)
VCO to SVCF FM (19)
WVCO to MVCF FM (19, 20)
WVCO feeding the VCFs in parallel (20)
WVCO to SVCF and DVCO to MVCF (19, 24)
WVCO modulates it's ”own” filter (19, 20)
DVCO modulates it's ”own” filter (19, 24)
MVCO to MVCF FM (20)
DVCO to MVCF FM (19, 24)
All VCOs to MVCF (13, 24)
SVCF before the MVCF (24)
VCOs divided between the filters and the phaser (19)
Both MVCO and DVCO to MVCF (13, 24)
Both MVCO and DVCO to SVCF (only) (-)
Noise to MVCF (13)
Noise to SVCF (20, 24)
Noise to VCP (19)
VCF and VCP filtering the noise in parallel (24)
SVCF can filter only the noise (20, 24)
VCP can filter only the noise (19, 24)
Wet/dry-mix on SVCF (13)
MVCF+SVCF+VCP in series, with wet/dry mix on SVCF (13)
VCP on WVCO only (24)
VCP not on WVCO (19)
Modulate the LFO bus (13)

The idea was to find four algorithms that together made as much as possible in the above list.
After 24 attempts, I arrived with four algorithms (13, 19, 20 and 24) that together satisfied all the requirements but one. Below you find block diagrams of the chosen algorithms (a line entering a filter module at the bottom means modulation of the filter frequency):

I decided that the one not satisfied requirement was not very important, so it seems four algorithms is indeed enough. But does this mean that I have arrived at the four best possible patches for the chosen modules? Maybe not. I think you can only determine this in practical use.

Better sound with sounder ideas

This all sounds very well thought out, you might say. But suppose the ideas don't work out the way I have thought?

This is a very real possibility. It's very likely that a function that I thought would be useful will rarely be used, whereas something more important have been missed. For that reason, it must be possible to change the synth without having to redo a lot of tedious work. The AMORE modules are already a good step in the right direction. But we all know the real work is in the front panels and other mechanical parts of the synth. So these parts must be highly adaptable for changes too.

To be continued...