The Turing Machine, designed by Tom Whitwell, is one of the most popular Music Thing modules. And like all the Music Thing modules it’s open source, so when Ben Rufenacht (of the LMNC Discourse group) decided he wanted a Kosmo format version, he copied the MT repo and laid out a Kosmo-size PCB and panel. Then Sebastian Beyer decided he kind of liked Ben’s design but wanted to tweak the front panel layout, so he made his own version. Which I liked better, but I still wanted to rearrange things a little, not to mention convert the panel graphics to something more like my own style, so I made my own fork and added my changes.
And then I built it. Here is my version of Sebastian’s version of Ben’s version of Tom Whitwell’s module.
Between me, Sebastian, and Ben, we made only a few very minor tweaks to the original design: Adding a current limiting resistor on an output that lacked it, adding an external protection diode to the input of the analog switch, and replacing ferrite beads on the power rails with reversal protection diodes. Oh, and I used a DAC0808 in place of the specified DAC0800, and larger resistors in series with the LEDs since I was using super low current ones.
One thing that changed along the way was the name. Music Thing asks that any derivative designs not use the original name or the Music Thing name, and while I think that’s probably meant just for commercial derivatives, and that Whitwell wouldn’t care what we called our Kosmo versions made for ourselves and a few friends, Ben did go ahead and name his version “Halting Problem”. Not that the module has anything to do with the mathematical Halting Problem, but then again, it has nothing to do with Turing machines, either.
What it does, under either name, is it runs bits through a shift register 16 bits long, under control of an external clock. Bit n also gets fed back into the first bit, where n is switch selectable from 2, 3, 4, 5, 6, 8, 12, or 16. When a bit shifts from the nth bit back to the first it can be copied as it is or it can be complemented, and that’s determined by comparing the output of a noise source to a threshold voltage. Set the threshold very low and the bit will always be flipped; set it high and it will never be; set it in the middle and it’ll flip half the time, meaning essentially a random bit gets shifted in. Or if you push up on a momentary toggle switch it’ll force a 1 into the first bit; push down to force a 0.
The first eight bits are sent to a DAC. Essentially they’re interpreted as a binary number and that number is converted to a corresponding voltage. That goes to the output jack. So it’s a CV sequencer, where the sequence can be affected by the toggle switch, and repeats exactly if the threshold knob is all the way up or down, but otherwise it evolves randomly.
The first eight bits also are available on a pair of pin headers (one with and one without a logical AND with the clock signal) which can be used to connect one or more expansion modules with a ribbon cable. Ben adapted two of the MT expansions and I did versions of those.
One was called Volts originally and I call my version HP Potentials. It just takes the first five bits as 0 or 12 volt levels, attenuates each with a pot, and sums the results. This creates a second CV sequence, different from but correlated with the main module’s output.
The other was Pulses and my version is HP Gates. It takes the first seven bits and sends them as pulses or gates to output jacks. Four more jacks are fed with logical ANDs of two or four of those seven bits. So it gives you a trigger sequencer with eleven channels, again correlated with the main module’s CV sequence.
I started on the Halting Problem a while ago, put it aside while I was breadboarding the pole mixing filter (and waiting for some 50k pots, of which I thought I’d had some but evidently not), and finished it up Sunday. Monday I calibrated it, and it seems to be working. Given the number of chips in the design (that number being 10) I was pleasantly surprised. Today I built both of the expansions.
The Github repos, if you’re interested:
https://gitlab.com/rsholmes/KosmoTuringMachine
Analog Output 