Towards Automatic Music

Published on Sep 24, 2016 karl welty

"The designs are based on cold cathode gas discharge tubes (neon NE-2's and XC-18's).

The XC-18 “amplifier/oscillator” is a three terminal trigger tube circuit based around the topology of a transistor common emitter amplifier. It has a master drive knob, a bias knob, provisions for self oscillation and the components of an NE-2 LFO. Both LFO and self oscillation functions are on switches.

I was surprised to discover that it “triggered” from external input synchronously. Waveforms of frequency X would cause it to oscillate at that, or harmonically related frequencies. This characteristic is threshold adjustable using the drive and bias potentiometers. It could gate on and off based on the incoming signal. I could control it from a previously built neon module, I could run it from the audio output of a laptop.

Previous experiments with counting circuits and attempts at gating had yielded some interesting projects, but all of them had functioned in a subtractive fashion. An LFO would modulate an oscillator by stealing some of its voltage causing it to drop in pitch, if used as a gate the oscillator would ramp up to its note. I had yet to create an additive control voltage source whose output could be adjusted, and have a variety of steps of differing voltage. I'd been aware of a project schematic I'd pondered a year or so ago, from a 1959 electronics magazine. It had baffled me then but seemed worth another look. The voltage divider after the lamp is really part of the big secret. A portion of the drive voltage goes to ground keeping the lamps counting in a stable pattern (based on input conditions) and the other portion drives a CV bus which feeds the transistor oscillator. Since all of my previous attempts had been serial I chose to give this parallel idea a try.

Building up the logic portion of the board and testing showed that such a device could output a stepped varying voltage of between about +2vdc and +14vdc. Previous builds with neon oscillators had shown that such a low swing (compared to the +130vdc lamp supply) barely tickles a neon relaxation oscillator... perhaps a major second, maybe a third, but even then only at lower frequencies. It was clear why the project schematic had a transistor tone generator. I chose to veer from my pure neon path for the sake of an experiment, as I knew that the XC-18 module would function if driven by a solid-state master oscillator... I had a bag of MilSpec LM-118h opamps, seemed good fodder for a test. Found a simple circuit and built a prototype, found another interesting vco circuit and wasted a week or so, learned the thing I'd already built was infact a vco awaiting CV, tried it... and yup, it changed pitch with CV input. I made some mods to the circuit to widen its range. It was time to start pulling all the sub-assemblies together.

The original 1959 project was intended as a bench toy. A device which would create a series of pitched patterns which would cycle and alternate over time. The capacitor section can be re-drawn in the shape of a six pointed star and any given lamp “point” has three caps leading to its rail. It is not a standard sequential counting circuit. The wide range of the resistor values in the original were intended to further throw off its timing and behaviour. Its a primitive computer based on a state table comprised of capacitors feeding lamps, the resistors determine the voltage on a lamp leg and/or the amount allowed to ground through two paths... one of which passes past a transistor vco and causes it to alternate in pitch.

In my re-design (instead of fixed resistors) I added a on-off-pad switch to each lamp leg and an input pot. They are fed via a bus from a master pot with an associated med-lo-hi switch. Widely varying voltages can be fed to any combination of the six lamp legs, which can emphasize their duration in a pattern or otherwise alter the manner the lamps fire. After the lamp is a 470k/51k voltage divider each leg of which also has a 1N4148 diode in series. The 470k path buses to ground allowing the lamp to function more normally. The 51k path goes to another on-off-pad switch and associated pot, bused to a master pot and med-lo-hi switch and sent to a 1/4” jack on the panel. The control voltage output of any given lamp leg is a function of is combined input drive and cv mixer output settings. They are interactive and adjusting the device can take a bit of patience. The variations available are vast.

I substituted my opamp based oscillator for the transistor variant shown in the original. It has a CV input 1/4” jack, tuning and range scaling pots, a volume pot and 1/4” output jack. It generates a sharkfin squarewave and is rather pleasant in its own right. The waveform works quite well for triggering the XC-18 module."