Another VCF Finished

[gstopp at fibermux.com]

     To make Juergen feel better about talking too much about SSM2040's, 
     I'll bring up a different filter. Some may remember that yesterday I 
     said that I had finished building a 4-pole VCF based on four CA3080's, 
     but had not tested it. Well this morning I snuck over to my bench in 
     the corner and fired it up.
     
     You know the weather here in Southern California is quite nice, today. 
     It's the kind of day that makes me glad...
     
     Oh, sorry. You probably want to know how the filter sounded....
     
     It's wonderful, wonderful. It's warm and clean and quiet. I used a 40 
     hertz square wave as a test signal. With no resonance you get the 
     characteristic loud full amplitude signal, with a very soft and smooth 
     low pass sweeps. With resonance turned up half-way, the volume drops 
     like all good 4-poles do, and a nice clean harmonic emphasis starts to 
     show up. With resonance turned up most of the way, the harmonics are 
     picked out very cleanly and distinctly as the filter is swept. I have 
     realized that one of the things that makes a good 4-pole sound (such 
     as the 2040) is how cleanly individual harmonics in the input signal 
     can be picked out, as the filter is swept, when resonance is turned up 
     a lot. Even with a high amplitude signal input (+/- 8 volts, or 16 
     volts peak-to-peak) the filter did not distort. If resonance is turned 
     up further, the filter breaks into oscillation as a very clean and 
     controllable sine wave. Too much resonance after that point of course 
     will create distortion on the sine wave.
     
     There is a single offset trimmer on the first 3080 in the chain, and 
     this seems to work just fine for trimming the whole filter's DC 
     offset. On the 'scope everything looks extremely good from stage to 
     stage, very clean and undistorted. I have tapped the second pole as a 
     quadrature output (like in the Electronotes circuit) and when I make a 
     lissajous pattern on the 'scope it comes out as a flawless, perfect 
     circle, indicting low sine distortion.
     
     The filter frequency sweep range is from 3.5 hertz to about 13 Khz. 
     This seems a little low on the high end, and probably the high end 
     could be pushed up a bit by putting in smaller caps in the 
     integrators. Doing that will probably bring up the low end at the same 
     time, which would be okay to do for audio purposes (3.5 hertz is a 
     little low for audio). However, for quadrature applications (such as 
     quad circular panning) 3.5 hertz is a little high, so really to use 
     this VCF for both applications it may be necessary to switch in caps, 
     maybe with an analog switch, relay, or rotary switch.
     
     I really think that this circuit works so well because the 3080's I'm 
     using are very good parts. They are of a recent batch from Harris, so 
     the manufacturing process is probably quite refined by now after all 
     these years. Also the use of CA3140's as integrator buffers is never a 
     bad choice, because you know for certain that if there are any 
     problems in your circuit, it ain't due to buffer errors even with 
     teeny weeny capacitors. To be safe I buffered the two outputs (sine 
     and cosine) with an LM358 due to the low output drive capacity of the 
     3140's.
     
     Anyway I think that this design is a good alternative to scrounging 
     around for SSM chips, as long as you don't mind all the extra 
     soldering. This circuit is different enough from the same design in 
     Electronotes #41, so I'll draw up an updated schematic. If you don't 
     want to wait, just use the same circuit with 3140 buffers instead of 
     the FET's, and a pair of 2N3906's or a MAT-03 instead of the AD821. 
     BTW the same filter is also in the Preferred Circuits Collection, as 
     option number 3.
     
     - Gene
     gstopp at fibermux.com