Beads' crackling noise (CPU overload, software fix ready very soon!)

Eurorack
#1

Hi Émilie!
First, I’d like to congratulate you on the new module. It’s so creative and wonderful sounding. I’m looking forward to spending more time with it.

I have a slight issue with my Beads and I wonder if anyone else has encountered this.
When a stereo signal is patched into the module, w/ signal 100% dry and density not at noon I can hear a crackling noise. The crackling noise responds to density, so I assume that it is some artefact from seeds being sown.
This behavior does not happen when a mono signal is patched.
Here is a recording of the issue. Density starts fully CCW and at the end is at noon. (I have EQed the recording so the high crackling is immediately apparent)

Has anyone else encountered this?
Thanks!!

1 Like
Beads - release announcement
#2

Interesting. I can’t reproduce it on the few units I have here:

  • You don’t have grain trigger generation enabled on channel R, correct? It would sound different, but just checking.
  • Is the CV assigned to the dry/wet parameter? Does the crackle react to other grain parameters? (theory: small offset on the CV input causing the signal to be 99.9% dry).
  • Is the reverb off?
  • Which quality mode?
  • How does it react when you manually lower the gain of the input signal?
#3

Thanks for the quick reply!

You don’t have grain trigger generation enabled on channel R, correct? It would sound different, but just checking.

I don’t think so. Although, if you’re referring to the latched grain gen mode, then yes, the behavior doesn’t occur when it’s off. When it’s latched, then the crackle is constant, when pressed, the crackle also appears.

Is the CV assigned to the dry/wet parameter? Does the crackle react to other grain parameters? (theory: small offset on the CV input causing the signal to be 99.9% dry).

The CV was assigned to FB, but the behavior continues whatever it is assigned to.

Is the reverb off?

The reverb was off, but increasing it does not change the issue.

Which quality mode?

Bright digital. It seems like the crackle disappears when in other modes. Yes!

How does it react when you manually lower the gain of the input signal?

The volume of the crackle changes in tandem with the gain, but is still audible even at the lowest gain.

#4

Thanks for all the info, I’ll look into this! The first challenge will be to try to reproduce it. Obviously the EQ makes it more obvious, but maybe a direct recording would make it easier for me to know what I’m looking for.

I’m confident that it’s software (it would happen in the other modes otherwise) but there are just so many mysteries about it – why is it in bright digital mode only? why with 100% dry only? and where does it come from since the grain triggering signal exists only in the software realm? Odd CPU overload thing?

1 Like
#5

Actually, the crackle continues even when the signal is wet. I just realized that as I was testing.
I’ll make a raw recording later and post it here (or elsewhere?).
Thanks again for helping sort this out!

#6

Ok thanks, that gives me something to work on. My theory so far is that on some units (for which reason? Subtle variations in the crystal or the PLLs clocking the converters?) the code takes slightly more time to run comparatively to the codec rate, and since parameters like the number of concurrent grains were tuned to approach 99% CPU use, we can go over that limit. This explains why it appears when DENSITY is high (voice-stealing and grain parameter computation code running more often), and in the most CPU intensive scenarios (48kHz on both the dry and wet signal, and stereo). The fix will probably consist in sacrificing one concurrent grain :smiley: or scratching my head for one more code optimization somewhere.

I’m currently working 13 hrs a day with shipping and taking care of the production of the next runs, but I’ll look into it, I’m confident that it’s going to be addressable in a firmware update.

5 Likes
#7

I have identified the issue. There is indeed a “sour spot” when the grains overlap often enough to cause many calls to the grain-stealing code, and are triggered fast enough to cause many calls to the grain initialization and random parameter sampling code. This causes a very mild case of CPU overload (the last sample of the left channel is not rendered).

There is a lazy fix of course (reducing by 1 the number of concurrent grains), but I hope to find something more elegant. I’ll keep you updated!

10 Likes
Beads - release announcement
#8

Thanks because I have the same issues here : a lot of crackles in Bright digital (not others mode) without touching any knobs on both dry and wet (see picture). Of course I check my sound source (ears to tapo with feedback) : no crackles in direct sound from source.
We can hear the crackles in this demo video (it’s’not me but same crackles)

3’13, 10’ and at the end.

I’m happy to see that I hadn’t a broken Beads and that you’ll solve it in firmware update.
Thanks a lot for’your quick support with your 13hours works. I can wait, no problem !
Right now I’m enjoying it a lot with the clouds quality mode and it’s AWESOME !
A legend is born :star_struck:

1 Like
#9

Just to make sure I’ve totally nailed it, it’s only when the module operates in stereo, and on the left channel only, right?

1 Like
#10

I made some several tests:

  • left input only : no cracking
  • right input only : no cracking
  • stereo input : cracking mostly on left output but also a little on right ouput (but not everytime on right output). I also tried to mix the left input and right input with two manual vca before entering Beads : both the same if I mute left or right input.
    Craking starts with density knob from 3 to fully CW or from 9 to fully CCW, fully dry, dry/wet or fully wet.
    No feedback, no reverb (the same if I add them).

I have the same crack (only in wet) when I move or modulate the time knob in delay mode (size infinite) but I think it’s’normal because we always have weird sound when moving a time delay :grinning_face_with_smiling_eyes:

2 Likes
#11

I forgot to precise : no problem at all with the internal wavetable.
This mode is so a madness ! :ok_hand:

2 Likes
#12

Yes, the issue only occurs in stereo, and only on the left channel.
All of my tests conincide with @Benamad 's results (thanks for the confirmation that I wasn’t the only one!).
Here’s an un-EQed audio sample of the “sour spot”:

And Émilie, thank you again for all of your fast and friendly help. I know how busy you must be now and I really do appreciate it.

1 Like
#13

Forgot to say that I didn’t have crackles with square grain enveloppe (shape fully CCW), they appears with the others enveloppe (less with the last one fully CW).
If it helps …

#14

Yes square envelope is gentle with the CPU :smiley: 1.0f most of the time.

1 Like
#15

Another lazy fix consists in overclocking the MCU to 500 MHz. Naaah, won’t do that :smiley:

Oh god, I hope there won’t be any “use at your own risk” alternative firmware overclocking the MCU to dangerous rates for the sake of more concurrent grains or a higher sample rate.

3 Likes
#16

Thanks for the warning :smiley:

#17

ouch, in that case you might have to rethink your generous future support policy.

#18

Imagining a cursed future of people trying to liquid cool their modular synths and trying to install arch on the modules

6 Likes
#19

I don’t know if it’s just me or not but im getting a lot of pops and crackles when im in delay mode. At all qualities all density all feedback and all size values. Tried lowering the incoming signal level but that also doesnt help (read the manual) Is this problem related or is it another problem? Or am i the “bug”?

1 Like
#20

The problem we talk about here is only in hi-fi mode, stereo, high density, shape around 12 o’clock and size around 1 or 2 o’clock.

Not sure about the pops and crackles you mention, but couldn’t it be simply that you use a percussive or reverse envelope shape?