The (truly) great thing about Emilie’s modules (along with 4ms, TiNRS, etc) is that they are open source, clearly coded and easy to follow. This allows anyone to modify their own modules to enable them to do what they want! It’s a great way to learn coding on embedded stuff…


I do hope Emilie is fabulously wealthy, with her generosity. I am concerned that she has only 3 more modules to go and what will happen next for her and mutable instruments? I heard this on a recent Divkid video the great filling the Tardis series! Learned so much from them. Still consider myself a beginner its been about 3 years for me, in this wonderful world full of so many helpful people and manufacturers and Mutable is my favourite!I have grouped all of mine together in one of my 9U 84hp cases, to focus on them and make the most of them and the very special places they can take you!

When Marbles is in shift register mode, is it possible to have the T side sync with the X side? I have yet to find a solution. The shift register mode is truly amazing and completely underrated. Marbles continues to surprise me.

Have you tried patching t2 into the clock in on the x side and using that as the clock?

I tried it out and it seems to sync up. I don’t know what I was doing before but everything seems to be fine. Thanks!

2 posts were split to a new topic: Marbles - new alternative firmware

Hi everyone,

Perhaps someone can help me, because I’m really struggling with the behaviour of Bias and Spread.

I’ve got Marbles going into an external quantiser. My quantiser is set to D dorian. Steps is set to 12 o’clock (just a slight nudge beyond 12) - unquantised mode. I’ve got Bias set fully anti-clock wise.

When spread is full anti-clockwise I get the root D repeated with no other notes. This makes sense to me.

When I turn spread to 9 o’clock and leave bias at full anti-clockwise however, marbles seems to want to avoid the root note at all cost. Is this expected behaviour? From the manual i was expecting the root note to be favoured, but it just doesn’t seem to be the case. The only way I can get the root note to be favoured is to put spread at around 2 o’clock, but this means that the spread of the notes being played is much greater than I want in this particular use case.

The manual doesn’t actually show the various permutations of spread when bias is fully anti-clockwise, but the implication is that the root note is favoured (and this would seem to make sense musically).

I’ve also tested it with Marble’s in built quantiser on C major, with Dejavu set fully anti-clockwise and it seems to never play C.

What’s going on? Am I missing something really obvious?

I assume you have set the range to 0…2V or 0…5V, and that by “root note” you mean 0V.

Then the behavior you describe is perfectly normal. By setting SPREAD at 9 o’clock, you instruct the module that you want a “bumpy” note distribution, and by setting bias fully CCW, you indicate that the bump should be titled towards 0V. So you get a shape like the orange curve here:

Which has a probability value of 0 for the minimum value (0V with a unipolar range).

Then the distribution no longer has a shape of a bump, and the value 0V no longer has a null probability.

What I suggest you to do: keep BIAS at 12 o’clock, and consider 1V (one octave above root) as your new root note.

1 Like

Ahh thank you @pichenettes I think that makes sense… I’ll experiment further

Edit: FWIW to get the behaviour I expected (which was what I needed for this particular use case), I followed pichenettes’ advice, then passed the v/oct cv through a precision adder to minus 1V bringing it back to 0V, passed it through Kinks full wave rectifier, finally onto my quantiser.

I’m wondering if there are any tricks to clocking Marbles with odd rhythms from sequencers rather than just 16th, 8th and 1/4 notes etc… I get very strange results when I try this. Any tips?

What sort of strange results?

I think marbles can learn even non standard rhythms in order to then divide/multiply/jitter etc but I think I’ve read before about a certain clairvoyance to it that requires some kind of a pattern so it can work it’s magic i.e. that chaotic or random clock patterns can cause issue.

However that shouldnt stop you from feeding it complex or weird clock patterns! After all they are (subjectively) more fun!

As for tricks, I’ve had great fun using something like Frames step out (i.e. a non standard clock) to clock the t side, with the green branches mode, and making sure that rate is at 12 o’clock in orange mode (so there’s no dividing or multiplying of the external clock), and having jitter at fully CCW. And of course you can always clock x side with whatever kind of pattern!

1 Like

I like to clock a clock multiplier (A-160-5) with 960 style sequencer*. Variations in input clock timing trips the clock multiplier up. Set it for triplets, and clock a note sequencer with the result. It’s very funky. Sort of like the Marbles catch-up jitter, but more on-beat.

*-an 8/24 step sequencer you can adjust the time period between each step.

Marbles is wonderful for random stuff. I would love to get another one someday - but one is incredibly powerful

Would it be possible for someone to create an alternate firmware that repurposes the x section’s clock input to be a reset input?

I ask this humbly and with hesitation realizing that there are a lot of suggestion posts for an audience that has been spoiled by already well thought out modules. I love Marbles, and I know that it’s meant to be a random source, not a sequencer, but boy is it a fantastic sequencer. I use it a lot with other sequencers as well and the only thing I find myself wishing for when using it is a reset input. I do know that it will automatically go back to the first step after no clock is received for five seconds, but there are times where I want to do this live without stopping the flow. When pondering how it might be done I thought about how I rarely ever clock the X section separately. I have zero programming skill or knowledge of the code so maybe this doesn’t even make sense. If it does, maybe it’s also possible to preserve the original functionality by having a button combo (don’t shoot me!) to change behavior.

1 Like

This has been requested many times. I’ll eventually think of a way to do it. But first Beads, then production issues, then sleep.


Awesome! Take your time and by all means get some well deserved rest. Greatly appreciate your consideration.

1 Like

I’ve been using Marbles for over a year now and am pretty familiar with it. I haven’t had many problems with it but right now I’m seeing an odd behavior. I’m working on a song in the key of D minor. I made sure to patch the 1/vo to my oscillator first with the Steps knob fully CW and the Spread knob fully CCW. Bias at noon. I have a slow blinking orange light indicating I’m in a minor key. I then tuned my oscillator to D and let Marbles do its thing from there. It’s mostly playing within the key except there’s an A flat in there that is not in the key of D minor. I’m so confused. I’ve never had this happen before. Anyone know whats up with this?

If you need a specific series of ‘notes’ (Marbles only understands voltages, not musical terms or scales), it would be best to program them in yourself, the factory default ones I believe are based on a root of C (0v) so the major/minor/pentatonic will not really hold true if you transpose the root.

Programming a scale

  1. Connect the CV and Gate outputs of a keyboard or MIDI interface to the SPREAD (5) and CLOCK (6) inputs respectively.
  2. Hold the external processing mode button [O] for 2 seconds. The LED above the scale selection button [J] blinks and indicates the active scale. This scale is going to be reprogrammed!
  3. Play a little jam in the scale you want to program. Fifty notes, or more, is the recommended length.
  4. Press the button [O] when done.

The module analyzes your jam to measure how frequently each note occurs. The least frequently played notes will be the first to be eliminated when STEPS is turned clockwise from 12 o’clock. The most frequently played note will be the last one to remain when STEPS is at 5 o’clock.

Note: it is also possible, at step 3, to play the scale in ascending order, instead of a long melody. In this case, the module will not know the relative importance of each note of the scale, and the gradual scale “carving” will not be performed: turning the STEPS button from 12 o’clock to 5’clock will not modify the scale.

One can take advantage of the way the module counts notes to program several subsets of the same scale along the course of the STEPS knob. For example if we play :


We’ll get, right after 12 o’clock, C D E F G A B ; then C D F G A (least frequent notes E and B are eliminated) ; then C F G (less frequent notes D and A eliminated) ; then C G (F is the next to be eliminated) ; then C (which is the most frequent note in the fragment).

I have two questions I’d like to put out there.

  1. Is there a preferred rate when sending a clock into the Marbles clock input? I noticed that my rhythms were wildly off when clocking from Pamelas New Workout at 1x speed but once i upped that to 4x speed everything seems to tighten up and sync better. Does this even make sense?

  2. I still tend to have trouble making complex rhythms with Marbles that sound in time with a metronome. Even if I’m using the t2 sometimes t1 and t3 sounds off to me even with Jitter down. Are there any tips for making complex rhythms with fills and changes that stay locked solid in time?

  1. Do you mean 1X and 4X as tempo modes of Marbles in general (the button under RATE) or as in the position of the RATE knob? Just in case it’s just that simple of an issue: remember that the RATE knob has to be at 12 o’clock for the clock input to remain unchanged. I usually clock my Marbles at 1X with a clock from Pam and the only scenario I can imagine where Marbles would behave as you describe is if RATE is not at 12 o’clock, but if you already got that out of the way it’s probably another issue.

  2. Which color mode is your t in? A good way to test this would be to e.g. set it to the green mode, then put the t’s bias all the way to the left and checking that Jitter is at 0: t1 and t2 should fire at the same time. Is the behaviour different? If it’s all good, you can start playing with the red mode for example, a simple patch would be to patch t1 to a kick, t2 to a hat and t3 to a snare. Then you can just let patterns play and decide if you want to trigger kicks or snares more often by using BIAS. You can also lock your patterns with the t DEJA VU flashing button.

Hope it helps?

Huh… interesting. I thought when clocked the rate knob becomes a clock divider/multipler. Is this not the case?

I’ve tried every mode. I basically would like to explore more complex patches. For instance I would like to send an oddly timed (not 4/4) clock into the clock input but marbles doesn’t really seem to like this. Or send LFOs into the bias for more movement. It just ends up sounding like a mess whenever i try any of this stuff.