VCOs - testing over the weekend

[gstopp at fibermux.com]

     Hi all,
     
     Here's some follow-up on my recent VCO experiments.
     
     As you may have read in my spoutings last week I had the most success 
     on the test bench with the Electronotes tri-square VCO using the 
     CA3080/matched pair. Although I wanted to get off to the local stores 
     to buy the right parts for the other VCOs I didn't get the chance over 
     the weekend but I did mess with this one some more. After the bench 
     tests with the scope it was time to do the tests that mattered. This 
     VCO has all the proper components with the exception of the tempco 
     resistor, which I will go into later. The resistor used is a standard 
     2K 5% metal film type.
     
     I set the VCO up next to my Moog modular and hooked it in (yah it 
     looked like a rat's nest). Tweaked it up to 1v/octave and tracked it 
     with two other VCO's on the Moog. Results:
     
     * Tracking - excellent. Zero beats to reference tone at all octaves. 
     Of course this takes a few dozen trimpot tweaks on the 1V/octave 
     trimmer. High note tracking in the 10k-20khz octave sounded crystal 
     clear, in comparison to a VCO with a high frequency track adjustment 
     that is not trimmed right, for example. Low notes also tracked 
     beat-free down to subaudio.
     
     * Stability - excellent. I set the test up, tweaked it, messed around 
     a bit (loudly), and then went out in the back yard and did some 
     yardwork. Came back in about an hour later, still in perfect tune.
     
     So - great, but what about these expensive parts?
     
     The VCO remained in tune even without the tempco resistor since the 
     room temperature did not change. According to Electronotes the 
     un-temperature-compensated exponential converter will vary in 
     frequency to the tune of 1/3 semitone per 10 degrees fahrenheit 
     (about 5.5 degrees C). Now this may be acceptable to some users who 
     have their machines set up in controlled environments; it would be 
     most noticeable in machines that have calibrated front panel markings 
     for tuning. However, many systems (especially homebuilt ones) have a 
     COURSE knob and a FINE knob and that's it as far as panel markings 
     go. This works for me and as such I get the feeling that a bunch of 
     these VCOs in a system will track and be stable for at least the 
     duration of a session if not longer. So I would advise the use of 
     tempco resistors if you want PERFECT VCOs but other than that you 
     should be fine without them. I want to check into the possibility 
     that a different type of resistor (like an old carbon junker) has a 
     yukkier tempco than metal film and may be better for this part.
     
     As for the MAT-03 precision matched PNP pair - this thing is 
     essential. I tried a pair of random 2N3906s in its place during my 
     bench tests - no attempt was made to match them. The VCO worked fine, 
     but when I put my finger on them the frequency changed on the scope 
     by a noticable amount due to body heat. Doing the same thing on the 
     MAT-03 produced no frequency change at all. I did not try to match 
     transistors after that and do not intend to since I know I can get 
     the MAT-03.
     
     Circuit description:
     
     I started this VCO with just the exponential converter and tri-square 
     oscillator. After this was seen to work well I added a sawtooth 
     waveshaper which is based on a level-shifted triangle wave going into 
     an invert/don't invert follower driven by the square wave. The pulse 
     width threshold summer also provides the offset for the triangle in the 
     sawtooth converter, so that PWM also results in the sawtooth shape 
     changing in what Electronotes refers to as "symmetrized ramp 
     modulation", which is supposed to be similar to pulse width modulation 
     in sound but with a different harmonic spectrum. At one point the 
     sawtooth becomes a double-frequency sawtooth so the sound morphs into 
     the second harmonic as it passes through this point.
     
     Feature list:
     
     * CV summing node for frequency control, 100k = 1v/octave 
     * Linear FM input summing node
     * Triangle, Square, Sawtooth, and Pulse waveform outputs 
     * All waveform outputs +/- 5 volts, zero centered
     * Pulse Width/Symmetrized Ramp modulation input summing node
     
     Semiconductor list for the VCO in its current implementation:
     
     (3) TL082 Dual FET op-amp for summing amps and buffers
     (2) 748 uncompensated 741 for rectangular waveform comparators 
     (2) CA3080 transconductance amplifiers for the VCO guts
     (1) MAT-03 matched PNP pair for the exponential converter 
     (1) 2N3819/PN4391 FET for the sawtooth converter
     (4) 1N914 diodes for the VCO guts
     
     Next thing to do: sync input experiments
     
     TTFN,
     Gene