Request: Shift Register

[electrofux]

Hi,

maybe something like that already exists or can be stitched up easily.

What i need is a device which has 8 CV ins to register 8 values but has the possibility to rotate the values on a clock pulse and then outputs the Byte value and/or the rotated single outs. At best the device should chainable to make up an 16 bit shift register.

I was able to recreate something like that with a ton of Thors, a Mesher 8, Modselsors and Truth but it has become very heavy. Tried to figure if i can use a PSQ to do this but i am stuck at the moment.

As a background, i want to rebuild the Turing Machine Modular device.

[Loque]

Hum...i am not sure i really understood what you are trying. You want 8 values and each by each should output on a pulse signal? I would do this with Thor's sequencer in step mode.

Maybe i misunderstood or missed som einfo here or you want to input the values from external or something and rotate them than? Maybe this could be done with Korde or Shape or any other sequencer. Sounds like you need a stepper (Thor), somthing to record the value step by step, and finally play them back again step by step. Yea, i would try this with Thor as stepper and some sequencers like Analog Sequencer or something else, which can hold a CV value (in best case from a CV signal)....

[electrofux]

A shift register holds in case of an 8 bit one 8 values eg (which could be a Gate Signal coming in to the 8 CV Ins):

10101010

The dezimal output in this state would be 170

on clock it shifts all values 1 slot

01010101

Which would result in a decimal value of 85

The result-value could be dependant on the summing function (either internally or externally and in this case a bitwise summing) or one uses the 8 CV outs as a pattern generator but this is afaik how eg the Turing machine generates a Note Value at a given time. So each clock pulse a new note value is generated, 8 values for 8 steps.

BUT the Turing machine also has a probability circuit which decides if a Bit should be flipped or not over time. On midlle probabilty everything is completely random (50% chance of a flip). But at zero Probability the current register is locked (0% chance of a change) and thus generates the same 8 different notes.
So you can let it play on middle seting and when you like something you turn the knob to 0 and thus lock the pattern.
Therefore increasing the Probability also slowly (and pretty organically) alters the melody as there is only a low chance that a step gets altered. On 100% probability the 8 step pattern turns to a 16 step patterns where the second 8 steps are an inversion of the first.

I have build a combi that behaves like this but the Shift Register Part takes up the most space or device count. I know it is an esoteric request but shift registers are quite cool for generative music if you dig into them.
The problem is not the register part. Every bit can be represented by a 2 step thor sequence which outputs either 127 or 0 depending on a gate trigger to make it flip flop. But the hard part is to make this set of 8 values shift on clockpulse. Atm i feed the 8 values into a Mesher 8 and reorder them to the 8 outputs which as is i said involves alot of modules including the comparably heavy Mesher itself.

[ScuzzyEye]

I had thought about doing a LFSR device. My plan was to make it configurable from 1 to 32 bits. And allow for selection of which bit(s) feeds back to where. I was going to have it run at audio rates, but I see CV would also be a good idea.

I just don't know how many people would be interested in such a thing.

BTW, the Noise channel in my MS-34 chip synth, is implemented as an actual 8-bit LFSR modeled off the one in the Texas Instruments SN76489.

[MannequinRaces]

Lectric Panda’s PSQ has a shift register sequencer.

[electrofux]

Lectric Panda’s PSQ has a shift register sequencer.

Yeah i know. And i have a deep feeling that you can peel a Turing Machine out of it. Cant say i didnt try lol. But in the end i ddint have success getting what i wanted.

[electrofux]

I had thought about doing a LFSR device. My plan was to make it configurable from 1 to 32 bits. And allow for selection of which bit(s) feeds back to where. I was going to have it run at audio rates, but I see CV would also be a good idea.

I just don't know how many people would be interested in such a thing.

BTW, the Noise channel in my MS-34 chip synth, is implemented as an actual 8-bit LFSR modeled off the one in the Texas Instruments SN76489.

I have also seen some freaky stuff with Shift Registers at audio rate.

And well, it is for the nerds, that is for sure.

[emilng]

Lectric Panda’s PSQ has a shift register sequencer.

Yeah i know. And i have a deep feeling that you can peel a Turing Machine out of it. Cant say i didnt try lol. But in the end i ddint have success getting what i wanted.

Have you tried using the PSQ shift register sequencer in Klee mode?

From the manual: "This mode maps the sources Shift Register to multiple sequencer positions and sums the sequencer values together. Mixed at low Chaos Setting, it creates slowing morphing patterns."

[electrofux]

Yeah i know. And i have a deep feeling that you can peel a Turing Machine out of it. Cant say i didnt try lol. But in the end i ddint have success getting what i wanted.

Have you tried using the PSQ shift register sequencer in Klee mode?

From the manual: "This mode maps the sources Shift Register to multiple sequencer positions and sums the sequencer values together. Mixed at low Chaos Setting, it creates slowing morphing patterns."

Yeah, i tried. But the source of the Squencer will allways be the Shift Register Source and there you can select any selection of bit distribution but you dont have direct access on the individual bits which is needed for this. And the Sequencer Pots also cannot be controlled via CV. I have turned this upside down but cant get it to make what i am looking for.

There is something very cool about how the turing machine is designed as compared to other random generators. One thing is the evolving type of melodies it produces where you hear that the base melody often remains over alot of bars and only slowly alters in another direction. That is something i dont get out of Quad. I think it is because the turing machine alters its current state based on what is there and Quad recalculates everything from scartch on each pass. The Klee mode can probaly do something similar but i dont think it can be made lockable but i have to check the chaos parameter as i have have just found out that in the Mod Matrix there is a parameter Called SHift Register Probability hidden.

Edit: made a quick test. When you set Chaos at 100% and make Mod Matrix Knob 1to Moldulate Pulse 1 Shift Register Probability you get that shifting and evolving bit sequence which depending on the probability changes from subtle to drastic. However the bits are not actually registered/unregisterd because when you turn probability back to 0 the pattern also changes back to its original setting. I think it is not really a register in that sense.
On the turing machine at 0% probability it stays however the pattern was before and on 100% you get a double length pattern (as the second pass will be an inversion of the first pass). Everything in between evolves more or less.