[F] DIN Sync / sync24 timing issues

[Simon]

STATUS: Fixed in Firmware v1.1 - download it here: viewtopic.php?t=238
-----------


Short description of the issue: the DIN Sync (sync24) output of the Midronome, used for vintage Roland devices, is not perfectly in sync. Tested with TR606.

Setup (devices connected to the Midronome and how): Roland TR-606 connected to the ANLG plug, An.L setting configured to "din".

Steps to Reproduce: connect and press Play, listen to the metronome at the same time as the TR606

Expected Behavior: everything is in time

Actual Behavior: there is slight drift and sometimes offset, in particular when changing tempo

[Simon]

DIN Sync has been really hard to test, mainly because it was extremely hard to find beta testers with actual vintage Roland machines.
If you have a Midronome, have some vintage machine and wants to help please reach out

[beatbastelbaer]

Hi Simon,
the Midronome arrived today, the Firmware update was really easy and the Midi clock works out of the box.
I have a TR606 and tried to sync it with the Adapter cable (TRS - DIN) that came with the Arturia BeatstepPro (or Keystep37) and can tell you, that one doesn´t work. (works fine with the Keystep37).
I will solder a new one, first have to check the pin wiring diagram.
I will report as soon as I tested this.

[Simon]

Thank you so much

The wiring of the Midronoe just follows the specs from wikipedia: https://en.wikipedia.org/wiki/DIN_sync#Pinouts

Note that the DIN pins used by sync24 are not the same as the ones used by MIDI. So using a TRS to MIDI-DIN adapter has no chance of working.

The only adapter on the market available for purchase as far as I know is the one from OXI:
https://oxiinstruments.com/product/trs- ... k-adaptor/
https://www.thomann.de/gb/oxi_instrumen ... daptor.htm

The sync should definitely work (with the right cable), but the play/stop might be a bit out of sync - as mentioned in this bug.

[beatbastelbaer]

Hi Simon and to all DIN sync´ists,
I know that the TRS MIDI Adapter has a different Pinout ,
but the Arturia devices come with three Adapters,
two TRS - MIDI and one TRS - DIN sync , thats the one with black TRS Plug.
The cable is only available in a "spare parts pack" with 4 Adapters
https://www.dv247.com/en_GB/GBP/Arturia ... 015838-000
but the USB Y splitter alone is worth the price, the ultimate solution for eleminating ground loops,
but thats another story - back to topic:
Now I can confirm, the Arturia Adapter works very well
I had another problem ....

The sync seems to be in time, I started and stopped multiple times with different BPM´s,
made rapid speed changes for- and backward, no issues so far.
I will keep on testing, just wanted to clear the cable confusion.

[Simon]

Ok I see

Interesting information, and very strange you cannot reproduce the bug. Can you listen to the Midronome's metronome as the same time as the 606 to see if they are in time? Like I wrote in the bug, it should be out of sync by one tick (about 20ms) when you press PLAY on the Midronome - not every time, but often enough you can easily reproduce it.

[beatbastelbaer]

Hi Simon,
now I did some tests with audio recordings of 4/4 Beats,
the TR606 DIN sync device is on the left channel
the TR8S MIDI sync device is on the right channel
with e.g. Audacity you can see very clear if there is a offset
and here are some results:
- I cannot reproduce the start offset issue, Started many times, no offset
- when the Midronome is running with constant BPM there is no offset
tested with 60BPM 30 min 120BPM 20 min 250BPM 10 min no offset

BUT, when changing BPM there is a significant offset ,
tested with different up and down changes.

As I cannot upload the audio files here I will send them to your email adress.
Cheers
Frank
Please let me know if you need some more or specific tests

[Simon]

That's awesome thank you so much
Really helpful! I'll investigate what you sent and get back by email - then we can update this topic on the improvements

[justmull]

Hi Simon
I observe the behaviour you describe and have a 606, 707 and 808 here for testing if that helps.
Here's a recording at 118 BPM with the midronome panned left and the 808 panned right. I've included the individual tracks as well with another device midi synced to confirm it stayed in sync and just the DIN sync was drifting. The midronome has a loud hum but that's a seperate issue.
https://drive.google.com/drive/folders/ ... sp=sharing
Cheers!

[Simon]

Thanks Justin, I will look into that!

Frank was also particularly nice and gave me tons of data to look at. What I am worried is that it has nothing to do with the Midronome but these devices are old and just get out of sync? (my guess would be that they just miss some ticks)
Because of the way it is done it's near to impossible that the DIN sync clock is drifting compared to the MIDI Clock (or any other clock coming out of the device).

Could maybe you or Frank try to do the DIN sync but this time coming from another Roland device (for ex 808 syncing a 606) and see if you also see a drift?

[beatbastelbaer]

Hi Simon,
I made a test with the DIN 24ppq signal on one channel and a MIDI synced 4/4 beat on the other and I can see, that the Midronome sends correct 24 5V high peaks.
For more testing I synced the 606 with this setup:
Midronome -> MIDI -> Keystep37 -> DIN sync -> 606
and there is not the slightest drifting, very exact and tight timing

So, why is the 606 (and other Roland gear) drifting when synced with Midronome?
When we look with an oscilloscope at the Start/Stop 5V high signal, you can see the signal is not constant high, there is a kind of oscillated signal with constant frequency, I am nearly sure this is the origin of the drifting.
Maybe a condensator can help to smooth the high level ?

Greetings
Frank

[Simon]

Thank you Frank for looking deeper into this, yes there is definitely something going on here that makes it drift.
I'm sure it's solvable if you can get it to work with the Keystep37, I must have missed something somewhere

My biggest problem is I cannot get my hands on one of these here in Aarhus to test it with. But I heard similar reports of drifting with an RE303 which will be easier to get.

I'll investigate and get back to you, thank you both so much for your help!

[Simon]

It seems there is the same issue when syncing a Korg Volca using the ANLG plug. This actually helps a lot because they are much easier to get than these vintage machines

With the Volca it's easily solvable since it has MIDI as well. But I'm hoping I can get my hands on a Volca, find the issue, solve it, and that it will also solve it for the vintage Roland machines.

Simon

[beatbastelbaer]

Hi Simon,
is there any progress in your research according the DIN sync drifting issue ?
The DIN sync possibility was the main reason for me to purchase the Midronome.
Let me know if I can do some further testing,
cheers,
Frank

[Simon]

Thank you so much Frank, actually great timing because I'm literally going to look at it this week!

I have an idea of what the cause could be, I'd say it has a good 80% chance of solving the problem
I'll make a beta version of the firmware with the fix, it would be great if you (or anyone else) could test it and tell me if it solved the issue

Cheers
Simon

PS: and just FYI, I did get the Volca Beats last week, tried really hard to reproduce the issue but nah it just stays perfectly in time, no matter what mad tempo changes I make... I contacted the person who had an issue with it but I have not heard from him yet. If anyone has a Volca Beats and can get it out of sync with the Midronome I'd love to hear what you do

[beatbastelbaer]

Thank you Simon,
good news,
I am looking forward to test the new firmware update.
Frank

[justmull]

I thought the issue might be noise also appearing on the analogue out, so I got one of those ADuM3160 USB isolators that another user mentioned in the very loud hum thread. It fixed the very loud hum issue, but the din sync still gets out of whack. I've tested with an 808 and 707 now with the same result. Anecdotally I find it most easily reproducible at 118BPM but that could be nothing. Happy to test out a firmware if you want to send me a link.

(couple more recordings here if it's any help https://drive.google.com/drive/folders/ ... sp=sharing)

[Simon]

Thank you Justin, very happy to hear you were able to sort out the noise issue.
I think the sync issue is a simple signal issue, completely unrelated to the noise one.

I have much data and info from you and Frank now, I will investigate these days and send a beta firmware for both of you to test

Thank you
Simon

[Simon]

For more testing I synced the 606 with this setup:
Midronome -> MIDI -> Keystep37 -> DIN sync -> 606
and there is not the slightest drifting, very exact and tight timing

I've been testing and looking at things today, and this message particularly got my attention. So I looked at what the Keystep was sending, hoping to find the magical answer as to why you can sync your 606 with the Keystep and not the Midronome.

And gosh results were very surprising, I do not even understand how it can work!!
Basically, when syncing to MIDI, it seems the length of the DIN Sync pulses are based on the current tempo, to be exactly 50%. Problem is, when you do quick tempo changes, the pulses either get way too big or way too small compared to the tempo - to the point that if you go fast enough they entirely disappear! See the attached screenshots (FYI the second screenshot the clock was not even going fast - like 150bpm).

You can also see how the Midronome's clock follows tightly the MIDI while the clock coming out of the Arturia is a bit off (and the offset changes - not great But honestly I cannot blame the Arturia team, they packed SO much into that unit, and for such a small price, no wonder they did not have time to investigate minor clock issues).

This is clearly a bug on the Keystep, but I guess it's not only a specific setup (sync to MIDI and then sync out DIN Sync - the signal is fine when the Keystep is set to "internal clock") and also requires very quick tempo changes so I guess nobody ever experienced it. But maybe it should be reported to the Arturia guys


Anyways - apart from that I did find an actual BUG on the Midronome - a code bug, the very first one! Basically right now some pulses are skipped when doing wild tempo changes. So this will not solve the drift that you guys recorded when running at constant tempo. We'll get there

[justmull]

Here's some recordings of the din sync output from the midronome, 808 and a Squarp Pyramid
https://drive.google.com/drive/folders/ ... sp=sharing

The pulses from the midronome are a bit unstable in comparison and you can see the dropped cycle or 2 at 1:25.800 in the midronome recording (this would be where it goes way out of sync)

Some other irregularities are pointed at here

The Pyramid doesn't get the implementation exactly right either btw but it's clock is at least stable when syncing so thought it may be useful to include to look at

[Simon]

Ahhh Justin you found them just when I did

Basically yes, just like you did now, I found the issue: some pulses are skipped - also a bug! That explains the getting out of sync at tempo 68 or 118.
See attached screenshot.

That's huge progress guys, now I know what the issue is and I can reproduce it, the rest is on me, I just have to find a way to understand why those pulses are skipped... It seems it only happens at certain tempos, and happens about once every 2000 pulses

So, current status is:

• There is one bug when doing wild tempo changes, which might result in pulses being skipped
• There is another bug where some pulses are skipped every ~2000 pulses, at specific tempos like 68bpm or 118bpm

PS: Justin you seem to keep pointing at that noise, that's just a few mV, there is no way this could affect a 5V square signal - most gates are set to 1V or 2V, it would only be a problem if the voltage would be dropped below that threshold

[Simon]

This is now solved - beta firmware coming today or tomorrow.

[Simon]

Fixed in Firmware v1.1 - download it here: viewtopic.php?t=238

[justmull]

This is working great Simon! I've left it running for like 30mins and came back and still in sync. I've run it with audio sync as well and no issues. Thanks for fixing this so quickly - most appreciated!

[Simon]

Amazing to hear!

[beatbastelbaer]

Same results here,
tested it now for 2 hrs very deep with rapid speed changes, multiple start stop events, Iongtime test with the speeds were I found the drifting before.
The timing is 100% correct, not the slightest drifting or offset.
Congratulations Simon,
great work !

[Simon]

Awesome, and thanks for the in-depth testing!!