From eli@UX3.SP.CS.CMU.EDU Tue Feb 6 15:55:10 1996 6 Feb 96 18:55:04 +0500 6 Feb 96 18:42:25 +0500 Subject: matrix-6 trivia Date: Tue, 6 Feb 1996 18:14:56 -0500 (EST) From: Eli Brandt Organization: Dr. Beddoes' Pneumatic Institute Last weekend I was trying to tune my 6r to a 19-tone equal-tempered scale, and discovered something interesting along the way. I'd always assumed that bipolar signals (like LFOs) ranged from -1 to +1, and unipolar signals (like envelopes) were 0 to +1. Nope. They have the same total range; unipolar signals are 0 to +2 (or bipolar are -1/2 to +1/2, but that's ugly). For example, when you use a ramp to make a one-semitone pitch slide during the attack, you don't use the (+46) modulation that the manual lists as being a one-semitone strength. And this is why a square LFO doesn't cancel an envelope during its negative sections. In retrospect, this has been confusing me for quite a while... Anyway, if you're using a bipolar signal, here's a table for modulation of pitch -- notice that the even-numbered entries are straight from the table in the manual. semitones mod 1 42 2 46 3 50 4 53 5 55 6 56 7 57 8 58 10 60 12 61 14 62 16 63 I got these by retuning, matching the oscillators with osc2 detune, shifting osc1 up a few semitones, and matching osc2 with ramp1 modulation. All of these matches were good to within a few cycles per second at C5. If you're interested in tuning the thing to non-12 scales, read on... First I tried tuning the octave purely by ear, but that turned out to be too hard for my ears. I found that to cancel out keyboard tracking takes 4(-63) mods by KEYB, and a (+42) to clean things up. So the keyboard is hard-wired to mod the oscs by 4(2/3) - 1/12 = 31/12. (To define these mod strengths, I'm assuming that oscs have a 1 "volt"/octave response. A (+63) mod creates a 16-semitone interval from a +2 signal, so it has a strength of 16/12/2 = 2/3.) For 19-tet, we want to reduce the tracking by a (19-12)/19 fraction. So we need downward mod by keyboard with a strength of (31/12)(7/19) = 0.95175. Well, 0.95175 = 0.66667 + 0.28509 - 0.00658, which means that (-63) (-57) (oh well) will pretty much do the job. It turns out that (-63)(-57)(-29) is the right combination, as far as I can hear. Doing this to both oscs eats six mod routings... (Is this a perverse thing to do with an analog synth, or what?) -- . Eli Brandt usual disclaimers . . eli+@cs.cmu.edu PGP key on request . ."You probably use this word -- sound -- quite frequently, but did you . . ever actually think about nature of `sound'?" -- Casio VZ-10 manual .