Analog Rytm Loses Sync - Pattern Change

XanaX · 2015-04-28T06:28:06.000Z

^ as stated earlier by simon (who works for elektron) it would greatly help if you stated you’re exact midi chain, meaning what devices are you trying to sync it with and how are you setting them up specifically. i would also advise you to contact their support team and enter a ticket explaining in full detail the problem and you’re setup (midi chain & devices). this is the only way you may get elektron to look at the problem and hopefully find a solution.

by the way, since i think you also have a 909 it may be interesting to try AR’s DIN sync and see if that helps the sync issue (FWIW Analog Four had a Din Sync bug when it released but was quickly fixed with an update after reporting findings). As i mentioned earlier the 909 has a timing sync bug with it’s V1 firmware but syncs tightly via Din Sync.

Ess · 2015-04-28T08:59:48.000Z

Since these are quite serious claims, I felt that it was necessary to make some proper testing in the line of what is described in this thread.

I made a single click track on both the Rytm and Analog Four, with 3 different patterns of a full 1/4, 1/8 and 1/16-notes on the Rytm which I am cycling between during the recording. The Rytm is set to receiving MIDI clock and transport, and the Four is set to send. I am doing this in Sequential Change mode.

Here is a screen recording of scrubbing and zooming into the recording in Ableton Live:

For a precise measurement, I opened the recording in Audacity. I took three screenshots from the very first click to later in- and the middle of the recording.

audacity_0.png2416×1726 136 KB

audacity_1.png2416×1726 139 KB

audacity_2.png2416×1726 134 KB

As you can see, the delay is around 95-99 samples for each example.
Samplerate is 44.1khz

XanaX · 2015-04-28T09:36:14.000Z

so around 2ms jitter/latency which isn’t too bad. of more concern is the claim that “busy” patterns and “direct jump mode” make the sync choke to what i suspect unreasonable amount (+6ms). i think you should do further testings as it seems others are having similar sync issues:

http://www.elektronauts.com/t/syncing-rytm-and-a4/9050page:1

Ess · 2015-04-28T09:57:09.000Z

Very well then.
Next test conducted with two patterns with every track (except for click) filled out on all 16 steps, parameter lock memory maxed out as well.

0.png2550×1698 147 KB

1.png2550×1698 139 KB

2.png2550×1698 148 KB

10 - 15 samples difference.

Dataline · 2015-04-28T10:01:12.000Z

amiga303 · 2015-04-28T10:16:54.000Z

big Kudos to the mods, thanks Simon

XanaX · 2015-04-28T10:28:47.000Z

@simon: so you’re actually getting less latency with a busier pattern?! that doesn’t sound logical at all or am i missing something?

@amiga303: you stated earlier in this thread:

i did loads of tests and sync lags happen with “busy” patterns and with fast pattern jumps.
busy: dsp heavy stuff: loads of fx, param changes, etc.

so again i’m confused… are mods using a different OS version? is the lag only with “direct jump”?

why is simon’s test in contradiction with what others are saying?

xidnpnlss · 2015-04-28T12:07:47.000Z

Thank you Simon for your official contribution. Frustratingly, these claims have not been addressed properly and/or brushed off by Elektron in the past. I’m glad to see someone adressing them.

While you’re doing tests, I for one would like too see the Direct Juaddressed. could you try sequencing tracks of varying lengths in INF mode? And since you’re INF, see what happens when you use Direct Jump to change patterns? Thank you for your time.

clunky · 2015-04-28T12:35:28.000Z

well someone needs to record

an example of the sync issues

they are having

and post em up

Ess · 2015-04-28T13:16:27.000Z

""XanaX"":

@simon: so you’re actually getting less latency with a busier pattern?! that doesn’t sound logical at all or am i missing something?

No – I am getting more latency with a busier pattern. ~95 samples vs ~110 samples.

I’m not so sure about the OS versions. I know the Analog Four I used for this has an internal Overbridge version, but the Rytm might have the current public OS, I’m not in the office right now so I can’t confirm that. Anyway, I don’t think the devs have been optimizing MIDI sync the last couple of months.

I think my test contradicts what some people are reporting because my MIDI chain is as clean as it gets, and that is usually otherwise the culprit. MIDI is a messy protocol!

Changing the pattern length on the fly with separate track length on Rytm does cause a loss of sync, this is a known issue and is being worked on.

XanaX · 2015-04-28T13:34:19.000Z

ok that makes more sense, i must have misunderstood your “10-15 sample difference” comment on the busy pattern test. anyways glad to know you guys are working on timing sync optimisations under certain scenarios. i’m assuming an update will roll out with OB release next month and hopefully tighten sync in various degrees (i’m also guessing this is why AR wasn’t demo’d with OB during messe).

xidnpnlss · 2015-04-28T13:53:25.000Z

""Simon"":

Changing the pattern length on the fly with separate track length on Rytm does cause a loss of sync, this is a known issue and is being worked on.

FINALLY an official word that it’s being worked on. Up to this point, there was nothing definite. Thank you!

XanaX · 2015-04-28T14:21:57.000Z

@BTS_WRKNG: yea sounds like we’re on the same boat. i really want one since i tried it (it sounds bloody fantastic) but i need rock solid timing with a drum machine. the whole point of going elektron is to have “busy” p-lock patterns and doing live wizardry so the sync loss is a deal breaker for me… i do have a feeling an update is around the corner but won’t pull the trigger until it meets these requirements. as for the 909, i use din/fsk sync which is tighter then midi imo

AdamJay · 2015-04-28T15:22:49.000Z

""xidnpnlss"":

[quote=“Simon”]Changing the pattern length on the fly with separate track length on Rytm does cause a loss of sync, this is a known issue and is being worked on.

FINALLY an official word that it’s being worked on. Up to this point, there was nothing definite. Thank you!
[/quote]
yes. Thank you!

XanaX · 2015-04-28T18:19:30.000Z

""BTS_WRKNG"":

Yeah, for me it’s both the specific issues that are out there and also just waiting for a sign of ongoing commitment to developing and improving the product. As we know, Tempest had both problems: specific bugs/issues and a general feeling of being underdeveloped or incomplete.

Hopefully Elektron will do better! To be determined…

oh yea i agree on that too. as cool as the AR is, it certainly feels a bit rushed/unpolished in parts of it’s functionality. things like retrigger recording, mute states memory (as pointed out in the SOS review), track multipliers/dividers, extra LFO and of course more machine engines feel like a necessity on a unit of this caliber… i’ve been following the AR dev and an update seems long overdue. again i’m hoping they roll out a strong update addressing some of this stuff sooner than later!!

amiga303 · 2015-04-28T18:26:07.000Z

""XanaX"":

@simon: so you’re actually getting less latency with a busier pattern?! that doesn’t sound logical at all or am i missing something?

@amiga303: you stated earlier in this thread:

i did loads of tests and sync lags happen with “busy” patterns and with fast pattern jumps.
busy: dsp heavy stuff: loads of fx, param changes, etc.

so again i’m confused… are mods using a different OS version? is the lag only with “direct jump”?

why is simon’s test in contradiction with what others are saying?

hi xanax
what i have done: measuring sequential play and direct jump (assuming sequential and pattern chains work the same) when AR is a midi slave.

direct jump: havent tested direct start mode. I assume direct start mode is slippery too.
sequential play was sober for me
direct jump mode not. busy patterns and fast jumps quickly yield to sync problems.

i dont bother that limitation. i’d opt for a optimization. like a limiting interval when AR is slaved, eg. 1 jump per sec max.

thanks simon for that track length bug indication. dont want to be picky, still i wonder if a dev will look into the direct jump issue when time allows.

XanaX · 2015-04-28T18:58:24.000Z

thanks for clearing that up amiga… so sequential mode doesn’t choke even under p-lock / data strain? according to simon’s earlier test, busy patterns add about 10 samples of latency which isn’t huge and i guess would be hard to discern by ear.

now about direct jump mode. i’m curious about one thing, this mode allows you to switch patterns on the fly instantaneously right? so unless you have perfect robot timing, it makes sense that changing a pattern in this mode might make you loose sync (especially at high BPMs), or is the sync loss so abnormal that it certainly is a clock choke?

if it’s the former, then perhaps elektron should look into making the direct jump switching a short but non-triplet quantized transition (like in Tempest 16 beats mode were you can set direct pattern switching to 1/8th or 1/16th transition) that way a pattern switch would never risk falling out of the 16th grid and sound too messed up when synced. just an idea.

AdamJay · 2015-04-28T19:44:24.000Z

""XanaX"":

now about direct jump mode. i’m curious about one thing, this mode allows you to switch patterns on the fly instantaneously right? so unless you have perfect robot timing, it makes sense that changing a pattern in this mode might make you loose sync (especially at high BPMs), or is the sync loss so abnormal that it certainly is a clock choke?

if it’s the former, then perhaps elektron should look into making the direct jump switching a short but non-triplet quantized transition (like in Tempest 16 beats mode were you can set direct pattern switching to 1/8th or 1/16th transition) that way a pattern switch would never risk falling out of the 16th grid and sound too messed up when synced. just an idea.

The bug in direct jump is that it gets put back as much as an entire 1/16th note. Sometimes the jump is in perfect sync, though.

Ess · 2015-04-29T07:25:26.000Z

Direct Jump causing a loss of sync is also a known bug and is on the to do-list.

Bin · 2015-04-29T07:53:39.000Z

[quote=“” Simon""]
Direct Jump causing a loss of sync is also a known bug and is on the to do-list.
[/quote]

Is the problem of patterns not changing until the next loop around if triggered near the end of a pattern on the list too? And an option to remove the 150ms limit in sequential mode?