Using turbo midi of the TM-1 query

Elektron Gear Other Elektron Gear
#1

Hey all! I’m sorry if this is old ground, but I have a question about the TM-1 and the usefulness of the turbo MIDI spec.

I am sequencing an Akai S5000 from my OT. I work at high BPMs and send a lot of MIDI data. I understand that MIDI is a serial protocol (so no two hits may occur at exactly the same time), and that the density of the MIDI stream affects latency of ‘concurrent’ events, resulting in perceived timing jitter. I am looking for a way to minimise this jitter.

To be clear: it is MIDI note and CC data that I am primarily interested in tightening. I do not use MIDI clock to sync other sequencers. Even with MIDI transport/clock send disabled on the OT, it doesn’t take much note density for the jitter to be apparent.

My S5000 has the latest OS and exhibits extremely tight MIDI timing when sent a simple stream of note ons. It is not to blame. The issue is the limited bandwidth of the MIDI protocol.

The OT is the MIDI master clock and I am currently sending its MIDI out into an iConnectivity mioXL:

OT MIDI in -> mioXL MIDI in
mioXL MIDI out -> S5000 MIDI in

If I purchase a TM-1, and connect both the OT’s MIDI in and out to it, I believe I can negotiate turbo MIDI on the OT. This should allow the OT to transmit serial MIDI data at > 31.25kBaud, thereby reducing the delay between serial events.

My hope is, that I can connect the TM-1 to one of my mioXL’s USB MIDI ports, so that the TM-1 can transmit MIDI data into the mioXL using the faster USB protocol, thereby mitigating the serial jitter between the OT and the mioXL. That then leaves the problem of serial jitter between mioXL and S5000…

The S5000 only has DIN midi and doesn’t support the turbo protocol (and neither does the mioXL, so it’s moot), but the S5000 does feature two separate MIDI input ports. I can use the mioXL’s routing to distribute critical note events across the two MIDI ports, to mitigate the jitter. The additional benefit will be that MIDI data sent to other synths etc will be tighter at the point of entry into the mioXL, so it will be able to trigger devices connected to individual ports with greater timing accuracy.

Does this sound like a reasonable approach? Or is my logic flawed?

#2

Welcome to the forum!

I don’t think this will work.

The Octatrack supports Elektron’s Turbo protocol but the mioXL does not, so the OT cannot negotiate any higher MIDI transfer rate. If the TM-1 is recognized by the mioXL, it would only pass messages at the standard MIDI rate.

Even if you were to split the data out from the mioXL to two independent MIDI Out ports, you would see no improvement in timing delays in messages sent by the OT.

(By the way, I wouldn’t use the word “jitter” to describe the latency that is characteristic of MIDI data transfer.)

If you are sensitive to timing of MIDI events, I recommend that you sequence the S5000 from either:

  • a hardware sequencer that has two independent MIDI Out ports;
  • a software sequencer that has good integration with a USB-MIDI interface.

In either case, the limited transfer rate for the MIDI messages going into the S5000 will (as you know) be the limiting factor.

#3

I don’t think it matters that the mioXL doesn’t support the turbo protocol.

I was proposing connecting the TM-1 to the mioXL using USB, not DIN MIDI.

The interface between TM-1 and mioXL would be entirely USB, which ought to be able to transmit a denser stream of events.

#4

I thought “jitter” is more frequently used to describe undesirable variable timing discrepancies, whereas the latency in a system is typically fixed. The timing discrepancy I describe varies with the stream event density, and is undesirable. What would be a more accurate term?

#5

OK. Even if you get the TM-1 to negotiate with the OT, you will still be limited to 1 x the MIDI transfer rate when you send the data to the S5000 over each of the two standard MIDI connections. So the maximum benefit you could possibly get would be 2 times the transfer rate.

1 Like
#6

I would describe jitter as a fluctuation around a nominal rate, of the kind often observed in transmission of MIDI clock messages.

I think the idea of latency being a fixed value is associated with audio data transfer. In general, I think that latency can be variable Whether we call the phenomenon delay or latency, we know that the source is the limited serial data transfer rate in the standard MIDI (1.0) protocol.

#7

it does matter; you’ll be stuck with the “standard” MIDI data speed. and the TM-1 is connected to the OT via MIDI. as @PeterHanes says, that’s still your bottleneck.

#8

Your assertion implies that connecting the OT to a USB MIDI host via the TM-1 yields no improvement in MIDI bandwidth. Which is in direct contradiction to the TM-1’s advertised purpose.

@PeterHanes - thank you, that’s super helpful. I think we’re on the same page. If the TM-1 speeds things up by at least 2x, and the mioXL’s USB performance is “perfect”, then the theoretical max improvement I could get is 2x (by virtue of dividing the event stream evenly across both ports on the Akai).

This might be worth a try. Certainly cheaper and less disruptive than buying a Cirklon, or converting my entire DAW-less rig to CV/gate.

#9

I’m just going by me trying to transfer samples with the TM-1 last week. it negotiated a transfer rate of 10x MIDI when the Turbo button was pressed. but it transferred samples at the exact same rate whether the button was pressed or not (took the same amount of time for the same sample to transfer). because I was using Sysex Librarian, not C6. which, as far as I know, is the only thing that will properly talk to the TM-1 to negotiate a faster rate.

yes that’s the advertised purpose of the TM-1. but both sides have to know how to properly talk to one another to make that happen.

go ahead and get one (but don’t pay $250 or something insane please!) and report back. but I don’t think it will do what you’re hoping it will.

#10

Ah, I hadn’t even heard of C6. Thank you. My assumption was that the TM-1 was simply a class-compliant USB MIDI host interface… but your experience suggests that something proprietary has to happen on the host side in order to benefit from the increased baud rate on the DIN side. Curious.

#11

I agree.

(Interesting that you’re talking to two Cirklon owners in this topic.)

1 Like
#12

If the TM-1 were just a USB-MIDI interface then it would be worthless for what you want to do because it would not be able to operate the Turbo protocol.

But the TM-1 was specifically designed to implement the Turbo protocol and can initiate the negotiation of transfer rate. However, I don’t have a mioXL to try the connection to the TM-1 to see whether the mioXL will accept the MIDI messages at Turbo speeds via USB.

#13

Sorry, I meant “class compliant USB MIDI host interface with the ability to utilise speeds > 31.25kBaud over the DIN connections, provided the connected device supports Elektron’s turbo protocol”.

Unless the C6 software makes use of a proprietary driver for the TM-1, I see no reason why the mioXL shouldn’t be able to handle the USB side of things, whilst the TM-1 negotiates and maintains a turbo connection to a DIN-equipped instrument.

After all, the TM-1 works “out of the box” with DAWs as a USB MIDI interface, no? And in that scenario, surely it’s possible to initiate turbo mode to speed up transfer of MIDI data between the instrument and the DAW? That’s certainly the impression I gained from the TM-1 manual:

IN ANY PROGRAM YOU CAN ALWAYS TOGGLE TURBO MODE WITH THE TURBO BUTTON!
USING THIS FEATURE THE TM-1 POTENTIALLY ENHANCES ANY MIDI APPLICATION.

#14

it does work out of the box and is class compliant. and I think the mioXL could likely handle turbo mode just fine if it was programmed to do so. but it’s also possible (and likely) that it was written with, basically, logic that says “communicate at this rate because that’s all the faster standard MIDI can handle.” and not logic that even considers going 10x faster (the Turbo midi rate).

#15

@chiasticon I don’t think the mioXL needs to be aware of turbo mode. It will just see USB data packets containing MIDI data. The TM-1 is the thing that takes USB MIDI data and serialises it for the DIN MIDI out port - it just happens to do it at a faster bitrate than standard MIDI. Same thing for MIDI data from the OT to the TM-1: the TM-1 “reads” bits from the DIN cable faster than regular MIDI, but it still has to pack the resulting data into a standard USB data packet for a class compliant host to handle.

Apologies for repeating the ALL-CAPS quotation from the TM-1 manual, but:

IN ANY PROGRAM YOU CAN ALWAYS TOGGLE TURBO MODE WITH THE TURBO BUTTON!
USING THIS FEATURE THE TM-1 POTENTIALLY ENHANCES ANY MIDI APPLICATION.

I don’t want to get into a circular argument. I think we’re all using more than a little conjecture here. Would be nice if somebody from Elektron could chime in here authoritively (I did email support last week but haven’t heard back).

#16

I think you’re overlooking a very important word: “POTENTIALLY.” but maybe you’ll get lucky and it’ll work. grab one and have at it!

#17

I’ve pulled the trigger on a TM-1 (~£90). Not too stupid a price for an experiment, and I suppose I can always resell it. I’ll report back in a couple of weeks once it’s been delivered and I’ve had a chance to test. Thank you both for your insights!

2 Likes
#18

Make sure to connect both midi in/out to get the turbo button to work

2 Likes
#19

Update!

The TM-1 arrived. TLDR: it works.

I set up two identical programs in a multi on the S5000, one on MIDI channel 1a, and the other on channel 2b (using the second MIDI port).

I set up the mioXL to send odd channels to the S5000’s first MIDI port, and even channels to its second port.

I connected the OT to the TM-1 and ran the TM-1’s USB port to a spare USB host port on the mioXL.

I then set up a pattern on the OT, with two identical tracks. One sending on MIDI channel 1, the other on channel 2.

Before engaging turbo mode, there is audible phasing due to the delay between serial MIDI events. The delay varies with each hit, so the phasing varies in perceived pitch.

After engaging turbo mode, the phasing increases in pitch as much as 2x, and the variation in tuning is much reduced.

From this we can deduce that the TM-1 provides at least a 2x speedup in data transfer between the OT and the mioXL.

Additionally, when sequencing polyphonic hits on a MIDI track, the flamming is dramatically improved.

I know this isn’t a scientific test, but trust me, it really does yield a noticeable improvement. I am very pleased. YMMV though.

5 Likes
#20

interesting results! glad to hear you’ve gotten some improvement with it, and that it sounds like it’ll work for you.