Digitone Arp + Soundlock bugs and crashes

Firstly I would like to acknowledge that I have been lurking here a lot over the last 2 and a half years. Really appreciate how these machines and this community have rekindled my love for making music.
I had gotten bored with the classical instruments I was trained in during my childhood and wasn’t making much/any music anymore. I had discovered and bought an OP-1 in the hopes of doing something about that and although the OP-1 didn’t really do it for me, it did lead me to discovering the DT and Elektron in general. After some research I quickly got an OT instead.
Long story short: a year later I had acquired every machine Elektron has made except for a DT (and the MS/MC). “Worse than heroin” is a phrase that comes to mind haha. It even made me pick up audio-programming (which wasn’t that much of a stretch since I was already relatively fluent in low level c/c++). Unfortunately last year forced me to focus on some other stuff, but I have been at it a lot again the lost couple of weeks, and I’ll do my best to participate/give back/share a bit more with the community, instead of just lurking here. :heartpulse:


So to start taking myself up on this, I have discovered this Mystifying Digitone crash:
I have a (somewhat) complex pattern using scale per track that, through sound locking, uses more than than 4 arps, for repeating hi-hats, simulating retrigs and fitting all my drums on a single track. The other 3 tracks use a single arp.

so the setup:
T1 : drums, soundlocks on almost every step, some with arps, some with prob-trig, trig-15 has soundlock with arp and 3:3 probs — scale: 16/16 x1
T2: no soundlocks, arp always on, no prob-trigs — scale: 16/16 x 1/4
T3: no soundlocks, arp always on, no prob-trigs — scale: 16/16 x 1/4
T4: no soundlocks, arp always on, no prob-trigs — scale: 64/64 x 1/4

master length 256

9/10 times this pattern completely crashes the machine into a locked state and a high pitched tone, the moment track 2&3 repeat. so at 64 of m.len. Once in a while it happens mid-track at either m.len 32 or 48 and once in while it plays just fine only to crash on the next iteration.

And now for the mystifying part, there is a couple of things that I can do to keep it from crashing.

Doing just 1 of these things keeps it from crashing:
1: mute 2 tracks, doesn’t matter which ones.
2: disable 2 of the arps on either T2/T3/T4
3: bring master length down to <=32
4: and, definitely the weirdest, disable the 3:3 probability on trig 15 of T1.
4a: changing it to 2:2 seems to fix it as well. 1:2 and any of x:3/x:4 also crash. haven’t tried any less frequent probs. these other (crashing) probabilities do seem to influence the point in the loop where it crashes.

things I tried to resolve it:
1: copy the pattern to a new pattern
2: copy the pattern to a new empty project
3: completely rebuild the pattern from scratch
4: upgrade from 1.30A to 1.31

yet the mystery persists. :mag:
USB cable is not plugged in and the machine operates just fine otherwise.

It’s not really a big problem since I’m not deeply invested in the 3:3 probability of trig 15 on T1 of a single pattern :joy: It’s just that the programmer in me recognizes that something is probably not working as intended here.
Willing to do some more bug-hunting/pinpoint what happens here if anyone has any suggestions. Otherwise I’ll just send it in as a bug report as is.

5 Likes

Really awesome bug hunt!

To me it basically sounds like you’re overloading the machine, you’re putting to much things in there so it strains the CPU causing it to crash…

The reason I say this is because the solutions you have found all point to actions that are less taxing on the CPU. Also my thesis is further proven by the fact that each solution is very random and seem to not show any correlation between the different functions.

The solution is probably Optimizing DN so that certain functions isn’t as straining on the CPU…

1 Like

That’s my first guess as well, sounds like OP might have succeeded in overtaxing the CPU somehow.
I’m guessing that doing something like turning off the reverb and otherwise leaving the pattern unchanged could help prove it?

Probably. Reverb trend to be very taxing on CPU’s in general so thats a good guess

Yeah, that was my first guess as well when it first happened, but the more I looked at it the more that hypothesis fell apart for a couple of reasons, mainly:

1: These guys clearly seem to know what they are doing in the audio-coding department and you would expect a maxed-out CPU to result in a dropped/incomplete audio-buffer and not in a frozen machine.
2: Solution 4a: the fact that just changing the probability from 3:3 to 2:2 on trig 15 on T1, fixes the problem is really weird and if anything that taxes the cpu “more” during one iteration of the pattern.

Especially point 2 now makes me wonder if it might be a bug in the sequencer part of the code.

I hadn’t tried “turning off” the effects since you can’t really do that on the Digitone (besides completely turning down the volume of the effects and dialing all the track sends down to zero) and I assume(d) that it still processes them regardless, simply because that’s the easiest way to make sure, from a programming standpoint, that overtaxing the cpu never becomes a problem “because you turned on the effects”. Simply put: The easiest way to make sure the effects can always be on, is to always have them be on :joy:

nevertheless I just tried it anyway, even disabled all the effects sends in the audio-routing menu. To no avail. :face_with_monocle:

1 Like

If the Digitone runs a real-time operating system this is not necessarily true.

But I agree this is a bug, because in either scenario overloading the CPU should not be possible.

1 Like

You’re totally right and after doing some digging that seems to be the case. Kind of had my head up Xenomai which I have been playing around with a lot lately. (which is awesome btw, if anyone reading this cares about that sort of thing)

I know it’s just a figure of speech but nothing is worse than heroin. Nothing.
Digitone is addictive though. I had my guitar running through Zoia and into the DN last night and it was a beautiful ambient thing.

3 Likes

Yeah :slight_smile: , It was indeed as you say a matter of speech. And even though opiates, also to my younger self’s surprise, never really had any substantial/meaningful pull on me since they never really did/fixed/relieved anything for me personally, not even in the temporary sense of the word, I do have a frame of reference for what you’re communicating. :pray:t2: Didn’t mean to downplay the pull/hold that that class of compounds can exert on a person. Probably should have known better considering this is a forum filled with musicians :smirk: (to be totally honest I might not have been entirely sober while I wrote that either)

and yes out of all the gear I acquired that year, I was definitely least gassy about the digitone, which turned out almost exactly the other way around once I got it. If it hadn’t hard-coupled kits to patterns and if the arp&the transpose could be forced to play in scale, it would have been kinda perfect. Wouldn’t know what else to want from it… :thinking: … random, the arp also definitely lacks random.

I was the same exact way. I always dismissed the Digitone and then one day it just appealed to me for some reason. I just started messing with the arp last night so wasn’t aware about the lack of random. The fact that I can use it as a sound card in a pinch really impresses me. Now all I need to do is figure out how to route OT to external inputs while at the same time being able to sample DN into OT and use DN as audio interface without phase issues or nasty feedback. I know it can be done somehow, I just have to figure it out. And no offense taken, don’t worry about that. I have to be able to joke about the subject or I will fall victim to it again.

2 Likes

Yeah I can accept the kit per pattern coupling as a design choice, I don’t like/agree with it, but that one seems fair. The lack of the random arp&and transpose in key feel like blatant omissions to sell more machines. Which I get from a commercial perspective, but as someone that bought most of them its kinda jarring to have to setup a box twice the size next to it, just because of the commercial need to motivate others to buy into the same nuisance, while simultaneously losing some of the awesome tight sequencer<->sound-engine coupling which made me want these things in the first place. :man_facepalming:t2:

Yeah I think I saw you wonder about something similar in a different topic. Assuming I understand what you want to do: You can definitely do this with overbridge, I just checked that for you, just dial IN L and IN R to zero on the master page and record the input channel in your DAW. That way the input signal coming from your OT into the DN won’t be audible through the main outs of the DN and therefore won’t create a feedback loop into the OT. Even though you dialed it down to 0 on the master page, the signal will still come through in overbridge as a separate channel that you can then record.

The only limitation I see there is that you can’t apply the effects on top of the inputs on the master page of the DN, unless you apply none of the effects on the internal tracks that you send to the OT, because that would create a feedback loop for the effects. :thinking:

Now I’m 95% sure you can’t do what you want with the class compliant usb drivers since that output is stereo and as far as I know there is no way of creating the separation that you’re looking for.

(And happy to hear brother, from where I’m standing that is the only way :relieved:, whatever the subject)

1 Like

So I started working on a new pattern, again putting drums on T1 and while I started on adding a baseline to T2 I noticed something. The soundlocked arps of T1 affect the arp on T2 while voice stealing is off. This is also the case on the pattern that completely freezes the machine, I just hadn’t noticed it before, because there is a lot going on in that pattern audio wise.

So now I’m wondering if it is related to this:

Made a video showcasing the problem (same as in post liked above)

Hey @purplewizard I saw that you commented on my other post with the arps and now I seem to have the same (or atleast a similar) mystical problem as you have with the crashes. My Digitone freezes with a high pitch beep and then the savings I’ve done in that session seem to disappear. Its really annoying and It happens periodically in different projects. I have done a factory reset to no avail. Because the pattern I worked on disapears, I can’t really troubleshoot. But I didn’t have the master length at 256, I believe maybe 64, and I didn’t have all Arps running. But I did have drums on pattern one with different soundlocks with different probabilities and with diffrent arps on diffrent sounds, and an arp running on pattern 2.
Have you been able to demystify it so that you can avoid it from happening? Does your saving from that session disapear?

Hey, @Kaffekopp sorry for the late reply, been a bit preoccupied, but to answer your question:
No i have not had any disappearing data. I have just send in a ticket to elektron documenting the bugs and some video examples so, lets see what they say. For now if you want to avoid the crashes avoid soundlocking arps all over the place, that’s the only way I can get it to actually freeze.


Besides the two bugs described above I found a third one. I had noticed it many times while jamming live, so I was never in a position to bug hunt while it happened, but:

Adjusting any variable on the arp of the track sound will effect sound parameters (pitch at least) of soundlocked sounds. Sometimes temporarily and sometimes this change stays active until the sequencer is stopped. It seems that the temporary change happens when the soundlocked sound has its arp activated and the ‘‘permanent’’ change happens when the locked sound has its arp disabled; although I haven’t vigorously tested that particular hypothesis.

temporary change:

permanent until sequencer stop

1 Like

@Kaffekopp @djst and anyone else who might be concerned about this: I have had some back and forth with elektron about this and they have confirmed these bugs and are looking for a solution.

4 Likes