Marbles


#165

1/ is there a limit to the number of scale “variations” on the upper half of the Steps knob?

7 variations are pre-computed and laid out on the second half of the knob.

why do pots have such a large dead-band/backlash?

On Marbles, I have added artificial deadband on the DEJA VU knob (you can see both illuminated knobs blink in this zone), and near 12 o’clock on STEPS.

Plaits, Stages, and all the other pots on Marbles do not have any particular tweak to their response.

I switched to a different potentiometer supplier in early 2016, for parts with a longer life-cycle and more consistency in rotational torque between batches. There’s no difference in response curve, maybe more backlash.


#166

I’m at a moment with Marbles where it seems like a good opportunity to learn something using different modules. I’m assuming there’s a way forward here.

Basically I use too little modulation, and the modulation I apply tend to be at best something like a clocked lfo, which is nice, but I’m trying to get a grip of what sample & hold entails. So with some additional extras for this patch I have a Kinks, Blinds, Maths and Cold Mac unpatched. Funny how I under-utilise my plumbing.

I would like to modulate steps from 5 to 12 o’clock and spread from 12 to 3 o’clock. I it would be nice for this to occur every four bars would. It’s roughly what I’m currently doing with my hands for this patch. I’d rather have machines do the heavy lifting.

Thank you all.


#167

You can do that with a clock divider, then routing the output of the clock divider to an attenuator (Shades, a Blinds channel or one of the hundreds of available other options!).

A cool module for introducing long term variations (every beat, second beat, bar, second bar…) is the Doepfer A-115.


#168

Ok!

I’ll dig into Teletype to set something up.

Marbles is too cool. Thank you.


#169

I just had an idea for an alternative spread mode, maybe one of you code-savy folks will be intrigued:
Holding the spread button switches to chord mode where t1 clocks x1 to select a root note while x2 and x3 play notes of a chord appropriate to the selected scale and root note within that scale. on every clock pulse from t2 and t3, x2 and x3 decide which note of the chord they want to play e.g. All outputs can sometimes play the root note or any combination of root, third, fifth and octave while spread inverts the chords counterclockwise and spreads the notes apart in octaves clockwise.


#170

7 variations are pre-computed and laid out on the second half of the knob.

Ok, then there is something I don’t get. For instance, here is a sequence I feed to Spread CV and Clock (commas are just for readability, all notes entered in a single sequence):
C, CD, CDE, CDEF, CDEFG, CDEFGA
That’s 6 different occurence numbers, so it should fit on 6 positions of the Steps knob.

Then, putting Spread fully CCW and scanning the scale with the Bias knob I get:
C, CD, CDE, CDEFG, CDEFGA
That’s 5 variations; it skipped one. Whenever I record a scale with >5 variations it squashes it to 5. I tried reading the sequence a few times in a row, same result.

Plaits, Stages, and all the other pots on Marbles do not have any particular tweak to their response. I switched to a different potentiometer supplier in early 2016, for parts with a longer life-cycle and more consistency in rotational torque between batches. There’s no difference in response curve, maybe more backlash.

It must be the pot then. For instance on Bias I have a dead band of approximatively 5% of the pot’s run; same thing for all the knobs on Plaits and Marbles AFAICT. Is it within specs? it seems a lot to me.


#171

What do you mean by deadband? At the beginning and end of the course? This is normal for pots with linear taper, see the curve:

Nothing happens in the first and last 5%


#172

The algorithm for creating the variations for the scale works as follows: note frequencies are computed and scaled so that their maximum is 255 (so the most frequent note(s) has a score of 255).

The scale variations are:

  • All notes.
  • All notes with score > 16.
  • All notes with score > 32.
  • All notes with score > 64.
  • All notes with score > 128.
  • All notes with score > 192.
  • All notes with 255.

#173

What do you mean by deadband? At the beginning and end of the course?

Deadband = the abstract idea corresponding to the mechanical notion of backlash: the fact that it takes a certain movement of the input (the knob) in one direction to make the output (the parameter) move; this amount of movement is the deadband.

No, not at the beginning and end of the course, right in the middle. Here is a video: https://www.youtube.com/watch?v=2GRdtLoGC0U

Making a large knob movement in one direction has an effect (pitch), whereas small movements have no effect.


#174

Ah! I see that the X pots (SPREAD, BIAS, STEPS) have more filtering than the others. That must be it. I’ll check tomorrow how much this can be lowered in software.

Nothing like that in Plaits though.


#175

If the filtering in question is linear, that can’t be it. As you see in the video moving the pot slightly has no effect at all.


#176

The filtering is not linear.


#177

Same thing for Plaits (the deadband seems a bit smaller, but still very present):


#178

I’ll check tomorrow when I have access to a module, but there’s no similar non-linear filtering code on Plaits’ FREQUENCY knob.

This non-linear filtering is needed on Marbles because even a tiny deviation of the SPREAD, STEPS or BIAS value can be “catastrophic” and cause a note to change in the sequence.


#179

The manual specifies that when Marbles needs random data, it gets it from a true hardware random source. Can we know more about this component?


#180

Chapter 24 of the STM32F405 datasheet.


#181

Already posted this on MW, but after I found this forum exists I figured this would actually be a better place to ask:

I love my marbles, but I cannot get the “grids” mode into phase with my clock signal.
I use a multed signal coming from an acme4 clock, it’ll sync just fine, but not in phase, ie the kick drum channel rarely has the downbeat.

I tried using clock signals from different modules, octocontroller and rcd (both with different divisions), a pico logic module, the clock out from my varigate 8 but all yield the same result: sync, but no correct phase. What am I doing wrong?

I wonder what clock signal people use that achieves a correct phase…

thanks!


#182

The clock signal is literally just a pulse, so it won’t make any difference to the pattern phase what clock source you use.

I don’t think it’s possible to do what you’re trying to do. This kind of thing is usually achieved with a Reset input which caused a sequence to reset to the first step when a pulse is received.

This is common on linear sequencers, but Marbles doesn’t really have any concept of a fixed “first step”, which is why I guess it has no reset input.


#183

You could probably achieve what you want by running the clock very slowly, and trying to plug the clock cable into your Marbles at just the right time.

It will take a bit if trial-and-error, but once it’s synced correctly, it should stay in sync, provided the “Grids-like” mode always produces patterns of a fixed length.


#184

If drum pattern generation is primarily what you want Marbles for, I’d recommend getting yourself an actual Grids module. It’s really fun!

And it has a Reset input.