Begin Primary Ignition
Sat 22 Jul 2017 by mskala Tags used: electronics, hardware, music, north-coastI'm very close to officially launching the North Coast Synthesis Web storefront. I have stock for the first product ready to sell. All that remains to do is file a few more sales tax registrations, and I expect to sort that out this coming week. I'm aiming for August 1 as the launch date. If you're reading this on my Web site you'll note I've replaced the long-running Chessudoku ad with one for my new store.
It has taken a lot longer to get to this point than I hoped and expected when I decided to start down this path. Looking back, I think the basic issue was that I, repeatedly, waited too long on various other people and events. The corporate registration; mail forwarding; and accounting setup all followed similar patterns: I'd contact someone, things would look good, then I'd wait and wait and wait for them before giving up and dealing with someone else. I probably shouldn't have waited so long in each case.
On top of all that, I'd simply not known how long it takes to make deals with Chinese factories, which very much slowed down my front-panel sourcing process. And even though I didn't directly need my personal effects from Scandinavia to run my business, I needed them to run my life in general, and I had been thinking I would get them in November. It turned out to be not until March, and that really put a crimp in my style. I was, no joke, running my whole digital life on an Intel Atom notebook computer for almost half a year and I'm just lucky it didn't get dropped, stolen, or anything.
So. At launch time I'll have one product for sale: the MSK 010 Fixed Sine Bank. You can see it on Modulargrid, and you can hear a demo of what it does on my audio server. You can also download the user/build manual, or the source code package.
Planned pricing is $24 for bare circuit boards, $160 for a full do-it-yourself kit, or $320 for an assembled module. Those are in Canadian dollars; U.S. dollar prices at current exchange rates would be roughly $19, $127, $254 respectively.
The next planned product will be the MSK 008 voltage-controlled octave switch. I call it that because I have to call it something, but it actually can be patched to do several other things as well, including wave-folding, flip-flop operation, precision adding, mid-side encoding or decoding, and so on. I've actually already started production on the first batch of these - solved what I think is the last engineering issue yesterday - but there are just a few parts I still need to source before I'll be able to finish any, and I have to write most of the manual, do another run of circuit boards, and so on. Probably the biggest sticking point on the octave switch will be the front panel, as was the case with the sine bank. I can try going back to the same place in China that I got the sine bank panels, but I may end up going with a German company instead. They're on vacation right now; next week I'll be placing a sample order. All in all, I figure it may be three or four weeks after launch of the Web storefront, before I can launch the second product.
Third product: the much-awaited MSK 007 Leapfrog filter. This is what I think of as the "flagship" product, and it'll be more expensive and is more technically innovative, but that also means it costs more time and money to gear up to produce it. At this point I think I'll be lucky to have those for sale by the end of the calendar year.
In my last posting I talked a bit about frequency shifters and discussed making a "Hilbert transformer," which would be the heart of one type of analog frequency shifter, as my fourth product. I was thinking that I could apply the leapfrog technology I used in the MSK 007 and that would give me a bit of economic moat, because it's a North Coast "competence" that my competitors are not able to match. However, I spent some time working through the math for that design and I was a bit surprised to discover that it doesn't really work. For reasons that are probably too involved to go into here, the advantages of leapfrog technology (especially: relative immunity to the effects of component tolerance variation), which I thought would be significant, don't show themselves in the wideband all-pass filters needed for building a Hilbert transformer, and a more conventional design would really be the best way to do it. That makes that whole thing less interesting for me. But at least I learned a bit more about the math side of it, and got acquainted with some new software in the process.
Instead, at this point I'm thinking about guitar synthesis. It's not easy to make anything that sounds like a guitar in a modular synthesis environment. First you've got the sound of a plucked string, which has what we call inharmonic partials in it - the spectrum isn't just all multiples of a single fundamental frequency, and that in turn relates to the fact that the sound source is more one-dimensional (the string) than zero-dimensional (the integration capacitor in an analog synth oscillator). You can get around that a little with a frequency shifter (but those are big and expensive in analog); or do something called Karplus-Strong synthesis with a bucket brigade device (only a little less circuitry than the frequency shifter, and it depends on special chips that aren't really made anymore, and it's only "analog" if you bend the rules for what counts); or give up and go fully digital. Then on top of that, if you want to do it in real time, you have to build six of them to play six strings at once. And figure out a way to control it, with modular control systems that are usually only designed for one or two voices. All in all, it'd be easy to spend $6000 on a basic analog modular system to play something that sounds like guitar chords.
My current thought is to build just the simplest circuit that could possibly sound guitar-ish, which seems to be a triangle oscillator overdriving a VCA that rounds the corners a bit, with a built-in envelope generator. I've done some simulation tests on this (link not promised to last forever - I didn't post it on the main audio server because it's work in progress) and I think it's good enough for rock'n'roll. Then I could put probably two of them in a 10HP module, and I have a fairly clear idea of how I'd want to build a chord controller module that could control three such two-string modules (necessarily digital, but that's okay). If I could fit the chord controller in 12HP the overall package would be four modules filling exactly half a row of Eurorack; still expensive but less than other ways of making similar sounds; and able to play guitar-sounding chords with a control system that plays nicely in a modular environment. And it'd have a reasonable entry path, because each of the string modules would be of some use without the rest of the package, and the controller could (with an expander instead of using the dedicated connections on the back) control other CV-gate voices. So depending on how things look after I've launched the Leapfrog filter, I'm thinking my fourth product may be some part of the guitar synth idea.
But that's not the only idea I have in the pipeline. A different one is a voltage-controlled filter based on inductors, which I was playing with in Denmark and haven't really looked at since. And something in the nature of "low-hanging fruit" would be alternate filter curves for the Leapfrog. This might be nice from a business perspective because in principle I might be able to sell an enthusiastic customer another new $640 module for every different filter curve without much additional investment in development. I can just do some (rather complicated) calculations, swap in some different resistor values, and get any number of other kinds of filters from the same basic design. I'd want to actually build one of each new curve and try it out before pushing it as any kind of official variant, though. Whether the alternate curves end up being just a section in the manual for advanced home builders to "mod" their projects, or separate products, or what, remains to be seen.
Extending the MSK 007 design to allow the curves to be selectable without building a new module for each one might seem appealing but wouldn't be easy; it would pretty much demand a fairly sophisticated digital controller to adjust all the many parameters in the filter core. One could certainly imagine something like the Rossum Morpheus with an MSK 007 analog core, but I'm not sure I really want to attempt building that very soon; I think people who really want such things are more likely to go with that all-digital module or something similar.
So that's the news from North Coast, and a bit of a look ahead. Watch for the store opening, and buy lots of modules.
0 comments
