Another module experiment

[gstopp at fibermux.com]

     Hi List,
     
     I've managed to throw together another module. This one falls into the 
     "I've wanted to build one for twenty years" category - it's the 
     "Quasi-Digital Bi-N-Tic Filter" as shown in Electronotes issue #92, 
     page 15.
     
     This filter is a 2-stage 8-tap commutating (multiplexed capacitor) 
     filter in a state-variable configuration. The input signal is summed 
     with the output plus the Q pot wiper, and is fed into 2 stages of 
     8-tap capacitor analog muxes made from 4051's. The Q pot is tapped 
     from the output of the first stage. The module's output is the output 
     of the second stage. The analog muxes are clocked by a counter that is 
     driven by a high-speed VCO. There is also a pot called "Bandwidth" 
     which is a dual unit, with one half at the input of the first stage 
     and the other half at the input of the second stage.
     
     For my circuit I used a 4046 as a VCO, driving a 74HCT163 synchronous 
     4-bit counter, whose A,B,C outputs drive the A,B,C inputs of the 
     muxes. For the muxes I used 74HCT4051's. These chips run off of plus 
     and minus 5 volts since they have an 18 volt max terminal-to-terminal 
     rating. For the integrators of each pole I used CA3140's, and the 
     audio input summer and output buffer are an LM358 dual. I also used an 
     LM358 for the CV input summers for the VCO (two inverters in a row).
     
     Power was provided by a small linear suppy at +12v, +5v, -5v, and 
     -12v.
     
     Assembly was un-eventful, aside from the usual "duh" mistakes. Didn't 
     blow anything up this time.
     
     For the audio input I used a single op-amp LFO producing a square wave 
     at about 80 hertz, amplitude about 3 volts peak-to-peak. For 
     monitoring I drove a pair of headphones directly from the output 
     op-amp (which actually works rather well, in case you don't know that 
     trick).
     
     The sound? Weird. I came to the conclusion that the "Bandwidth" pot 
     and the VCO frequency pot have useful ranges that are a small 
     percentage of their full rotation. I suspect that they are 
     inter-related in these ranges so I don't want to jump to conclusions 
     yet. The VCO is set up to run from zero hertz to 2 Mhz in one sweep 
     from zero to +12 volts on the CV input, so I expected that for that 
     particular control. Also the dual "Bandwidth" pot is 250K instead of 
     100K so I'm not surprised there either. The Q pot can cause 
     oscillation if turned up all the way, if you can call it oscillation 
     (let's just say the output gets into this "high-amplitude craziness" 
     mode). It has the same "max feedback is minimum Q" characteristics as 
     the Q pot on a state-variable design.
     
     Anyway at the least weird setting the filter sounded somewhat like a 
     flanger with high resonance when the clock frequency is swept. You can 
     also tell by listening that there are several notches (or peaks?) in 
     the swept spectrum. Overall the effect is very clangey and bell-like. 
     Many inharmonicities and alias-like noises, quite grungey. More 
     experimentation is needed with a wider range of input types (like 
     noise, etc.).....
     
     - Gene
     gstopp at fibermux.com