Why is finding a stable MIDI clock so hard? 

Expecting a Wintergatan reference: "So can we play tight music?"

Wow! I learned two things - the concept of "yak-shaving" ("raka en jak" in Swedish I suppose) and also why the MIDI clock on my Ableton Live performs so poorly compared to the MIDI clock from my trusty old Boss DR660. Great video and I would love to see the followup with the deepdive into the RD-9 and TR-909 comparison.

Enjoyed this very much. I liked seeing how you methodically went through this process with increasing understanding and troubleshooting. Being frustrated by MIDI Clock is almost a right of passage, but most people don't get to understand it the way you have. FWIW, when you must sync to the DAW and use devices that support MIDI Clock, it's far better to use a device like CLOCKstep:MULTI, Multiclock, etc. It's exactly the reverse of what you've done to make a reliable test. The clock output from the DAW is audio and goes to a device that can convert it into a much more stable MIDI Clock and then distribute it. Throw in a few other protocols like CV and Sync24, and you have Sync Hub for lots of gear. Even on OSes and DAWs that can output a reliable enough MIDI Clock, there can come a point where the project grows to such an extent that MIDI starts to suffer under load and these devices become well worth it.

this is the best explanation of midi clock and testing on the internet

Very interesting and informative, Thank you for producing and posting this video

Lennart, many many thanks for compiling all this information and walk the extra miles of analyzing things in detail. I like how you´re turbonerding on this topic (and please take this as a compliment :D). In my setup, everything revolves around ableton link, ethernet based, and I wonder what the good options are to translate that into MC sent to a variety of hardware devices. Currently I use Bitwig, create an audio tick track via ERM plugin and send that to the multiclock device from the same company. It may be smarter to have that one as the master when doing hardware based music only, but I don´t like switching the whole sync setup between producing and performing. Again, many thanks for this useful and valuable information, I´m looking forward to see more once you find the time. Good things take time!

and i thought behringer was rubbish. waiting for the next video.

Thanks for doing all that tedious and “boring” work 🙃 The end results were quite interesting.

Awesome! Thanks

Thank you Very much for investigating this ♥️♥️👍👍

Expert sleepers Silent way is good..so is their USAMO ..we have both in the band.

I don't want to pressure you to make the next video but I did enjoy this one a lot.

Why are the BPM values different for RD-6 and TD-3? Everything else in your table was going for 120 BPM but RD-6 is 127 and TD-3 is 117? That's off by quite a bit.

How i deal with this, knowing that ableton is shit with midi clock: i record each part and drag it so that it works in the mix. Or i work dawless and use one drummachine as master. Or i use the ERM multiclock and abuse the timing for each channel a bit to make it sound so i want to dance to it.

I'm impressed by those results from the Behringer devices. I can guarantee that having a Mac does not necessarily mean you get good MIDI clock out of it - I have done similar timing checks and found that MIDI out from my Mac via a MOTU MIDI Express 128 (over USB) gives spurious regularly recurring signals, when the only thing going through is MIDI clock and not MTC etc. I heard that part of the problem is the USB implementation as the data is packeted, but I don't know enough about that side of things to comment. Interestingly, machines we revere today as being 'tight' often have fairly sloppy clocks. The 808 is a discrete hardware clock and is absolutely not fancy in any way at all; the LinnDrum uses a 555. The sense of tight timing, I think, comes from every event happening simultaneously - when all the sounds hit together, even if each clock pulse is not 100% on beat, the hits are unified and not staggered as they would be if triggered using MIDI notes on a sound module, for example. If your kick and snare hit at the exact same time, does it matter if they're a millisecond off? Also, regarding my surprise at those Behringer figures - I've done some work with DIN Sync24, and as far as I've tested, none of the new gear - not one example - complies with the old Roland DIN Sync spec. The DIN Sync timing of Behringer and Arturia units with a DIN Sync option (out, not in) is abysmal. This, it seems, is because they don't reset the clock on Start, whereas the old Roland spec does: look at the service notes for the 606/808 etc and they explain that the clock runs continuously, even when the device is stopped. When Start is pressed, the clock is held low, there's an approx. 8mS gap to allow the device to reset fully, and then the clock starts again. There's a separate Run/Stop signal which is low for stop, high for run. Receiving devices hear the start line go high, and wait for the re-started clock to appear. Timing is tight and there's a consistent lag of 8mS after hitting Start. The problem with new units like Arturia and Behringer is that the clock is not reset, which means the lag on start varies according to the tempo of the clock, and little attention seems to have been paid to whether it's high or low at any given point. Rubbish, really. So I'm suprised their MIDI timing is so good!

Oväntat spännande! Har nog alltid känt att midiklocka varit instabilt och oförutsägbart vilket nog både berott på både instabila klockor men också noll koll haha. Har numera en MIDIhub som sköter klocka till allt i studion och det funkar fint sen nåt halvår tillbaks :)

Thanks Lenhart. Happy to see my TD-3 can be of use again from now on. What are your thoughts about a plain Arduino setup? Arduino tutorials are basically build around a MIDI port, 2 resistors and 3 jumpers which I like to experiment with but have never really had the opportunity to pass through a rigorous timing test ... but your knowledge of electronics is better than mine and maybe you can foresee noise issues that could set the timing off without a proper filtering module ...

Are these measurements just the clock send of each device? My interpretation is that the feel/etc comes also from how a drum machine/etc syncs to an external clock while still playing its own rhythm (not acting as a sound module). I would think that this would be measured by the output sounds, and a comparison between the RD & TR could be recorded in one pass with identical patterns output from each machine as a mono feed routed hard left & right. Very interesting results nonetheless, I would suggest maintaining the fun in content - there seems to be plenty of burnout on the platform lately. Cheers!

Thanks a lot, very interesting! It got me thinking that, could it be that some famous drum machines - like the original 909 or 707 - get their famous groove from having an inaccurate MIDI clock? The probability distribution you showed did not look gaussian to me - what if it is the unique shape of the distribution of clock irregularities that is the main contributing factor to a "unique groove"? This would have been a super cool follow-up video!

Promo>SM 💞

get ERM multiclock and forget about any midi timing issues

The reasons it’s so hard to find a stable MIDI clock is because the government is lying to us and the earth is actually flat 😂😂😂