My thoughts about a Matrix
My favored matrix is 8x8. 8 switches should be integrated that act as phase under AC and as an inverter with DC.
Switching the Matrix is only possible via MIDI and not from the module. Thus, the module will remain small, the costs are reduced significantly and there is no wear for any parts!
I came to the decision for 8 (instead of 16) because a 16x16 matrix is four times as large as a 8x8 matrix, and already two 8x8 will likely meet the requirements of my practice. Both the controller hardware integration as well as the MIDI handling is much less sophisticatedat 8x8 (64 pads) than 16x16 (256 pads). Nevertheless, a 16x16 matrix isn't off the table.
For improvised music a 16x16 matrix is perhaps intellectually too demanding, one would look too much/long and think - for what is, as we know, no time when making music.
First, I had not noticed who gave me this incredible openness and debated deepened with me this module - it was Graham Hinton (Hinton Instruments), truly an icon of synthesizer history.
Here i will copy his current posts from the forum:
Graham Hinton:
Sun Mar 06, 2011
Development is underway. I have designed two systems, one is based on the existing SwitchMix 8x8 pcbs and so will be about the same size and depth, the other is expandable to 16x16 in a 3U frame. When Jochen has the 8x8 type working with his software we will say more.
32x32 is going to be huge and expensive. I have built 128x16 with VCAs integrated on the inputs and outputs before (The Real World Automated Send Matrix) and it was two rack cabinets and £50k. I also designed the AMEK MICE which provided 64 channel MIDI muting for some AMEK and TAC consoles with remote controlled switches already built in, so the real cost was hidden in the mixer. Not to mention the cost of all the cabling.
I would question if you really want 32x32 because as matrices get larger the redundancy increases more so. You might get better value by having two smaller matrices or non-square ones with a few common rows and columns as the cost is proportional to the number of switchpoints. That was a technique used by the BBC Radiophonic Workshop in their Hypercard controlled project which was never really finished before they were closed down. There is an article describing it in the Feb '89 Sound on Sound called "MIDI Futures at the BBC", it does not appear to be online, but you might find it at a library. I'm sure that Mark Wilson would agree with me that after building it he then knew how not to do it, which is a luxury you have when you are spending other peoples' money. It went the way of Hypercard and that is a lesson to be learned: an object like this is a long term investment and it should not be based on short term flavour of the month computer software that will vanish with the next OS upgrade.
I don't see any point in making comparisons with other failed products or projects apart from using them as an example of atrocious ergonomics. Computer control is providing the necessary protocol for full control by an external computer, not putting a PIC in a box. Neither does putting a crosspoint chip in a box constitute a summing matrix. If an object like this does not work well and as expected it is worse than not having one and a waste of money. It is also very easy to get totally distracted by the control system and forget about the quality and accuracy of the signals being controlled. Obviously I'm only saying that because I haven't forgotten them and will be publishing a full specification.
