Another VCO that works

[gstopp at fibermux.com]

     Hi all,
     
     As you may remember a couple weeks ago I was concentrating on the EN 
     tri-square VCO as my primary home-built VCO option. Well now that I've 
     made two and am waiting for the parts for more (and for list members), 
     I've gone back to look at the runner-ups.
     
     I was able to get the Chroma VCO to work properly. Yes, it was cockpit 
     error. Let me just say that this VCO is a lot more sensitive to solder 
     flux effects than the others for some reason (charge pump maybe?).
     
     This design uses a 3046 NPN transistor array to charge a cap in an 
     integrator which gets monitored by an Exar 4151 charge pump IC which 
     detects the sawtooth peak and dumps a charge back into the integrator 
     cap to reset the sawtooth. A tempco resistor is used in the Chroma 
     design at the input to the exponential converter - I did my tests with 
     a regular metal film type, and as with the tri-square VCO there was no 
     problem with drift in a stable environment. Useable frequency range is 
     from a couple hertz up to over 20 khz. Tracking is quite good over the 
     entire audio range.
     
     The sawtooth shape has a problem, as pointed out before, such that the 
     start of the rising ramp has a step in it which starts out small at 
     low frequencies and eats more and more into the ramp at higher 
     fequencies until most of the ramp is gone leaving narrow pulses at 
     frequencies just above 20 Khz. This step is proportional to the charge 
     pump's finite cap discharge time. This waveform glitch means that this 
     VCO is not really useable as the basis of a general-purpose 
     multi-waveform VCO. About the only other wave you can get from this 
     sawtooth is a pulse, however the pulse width modulation will be 
     limited to a range starting at the sawtooth peak and working down in 
     reference voltage to the half-way point on the sawtooth ramp, because 
     if you go lower than this you run the danger of hitting the step and 
     that will cause a jump in pulse width. Also this "danger zone" will 
     get bigger as the frequency goes up, but it will at the same time get 
     less important to the sound since the harmonics will start leaving the 
     audio range.
     
     In a good general-purpose VCO it is desireable to allow the PWM 
     voltage to approach the triangle (or well-formed saw) from outside its 
     amplitude range, sweep through the range, and then go beyond the range 
     once again on the other side. The sound of this is of course silence, 
     followed by narrow pulses, through a square wave, back to narrow 
     pulses, back to silence. This is what you want in modular systems. The 
     Chroma VCO, however, only needs to go from narrow pulses to a square 
     wave in the "protected" environment of the guts of a polysynth.
     
     However the point must be made that this VCO has a very low parts 
     count. In the Chroma schematic there are two VCOs sharing the NPN 
     transistors in a single 3046, plus dual op-amps that share functions 
     as well. I copied this on a vectorboard and ended up with two VCOs 
     that use a total of six IC's in less room than one of the EN 
     tri-square VCOs. I did the rat's nest thing again and plugged the VCOs 
     into my modular Moog. Here are my thoughts:
     
     * The sawtooth sounds fine in the range where waveshape-related timbre
       is most important (low audio to mid audio).
     
     * The VCOs track very well over the entire tuneable range
     
     * Multiple VCOs (four or more) tracking each other with sawtooth
       waveforms is a really bitchin sound
     
     Therefore I am thinking of building a module that provides multiple 
     VCOs that track together providing only sawtooth waveforms, kind of 
     like a limited version of the modular Moog's 921A Oscillator 
     Driver/921B Oscillator group. This module will take up minimal panel 
     space, just enough for tuning knobs, a couple of FM inputs, and the 
     sawtooth outputs.
     
     Regarding the solder flux thing: when I built my first EN tri-square 
     VCO, it worked great even without de-fluxing, except below about ten 
     hertz. I de-fluxed the circuit and it became perfect at all 
     frequencies. When I first built the Chroma dual VCO, I cleaned off the 
     solder side with a pencil brush and isopropyl alcohol. Then I 
     performed my tests, with less than total success. Then when I built 
     the second tri-square VCO it had really rotten waveforms up to about 
     50 hertz. After checking all components and connections, I decided 
     that solder flux may be the problem so I really blasted it with 
     industrial stuff. It was better, but not perfect. The next day I 
     looked again and it was performing even better than the first 
     tri-square VCO! Knowing all of this I then really blasted the Chroma 
     VCO circuit, then dried it with a heat gun, and sure enough, it 
     straightened right up performance-wise. So here's some guidelines:
     
     1. Build the VCO
     
     2. De-flux it with a good industrial de-fluxing solvent (available at
        electronics stores in spray cans). Be wasteful.
     
     3. The temperature drop caused by spray can de-pressurization will
        cause moisture to condense on the circuit, so that it will not
        perform well right away.
     
     4. Either let the circuit board dry completely (about a day) or force
        the issue with a heat gun, hair dryer, hot sunlight, etc.
     
     Oh yeah before I finish I should mention that I modified the VCOs to 
     add hard sync inputs. I did not use the sync circuit as shown in the 
     Chroma schematics, however. The Chroma uses some additional 
     transistors to provide an additional discharge path to VCO 2's 
     integrator. As I was looking in the databook for the 4151 I noticed 
     that the sawtooth peak is established by a comparator reference pin 
     that is tied to +5 volts in the Chroma, but was created by a 20K-10K 
     resistor divider in my circuit since I am only using +/- 15v supplies. 
     Well I just added a cap to this reference point to allow an external 
     edge to change the 4151 discharge reference, thereby forcing early 
     reset of the sawtooth. Works great.
     
     Thanks to Ric Miller for the 4151's he sent me. I think he said he had 
     access to a whole bunch of these for real cheap (right, Ric?).
     
     Later,
     
     - Gene