MIDI Express XT SysEx corruption :(

[seedylee]

After many seemingly random problems with various hardware modules and software editors, I have come to discover that the (supposedly top-of-the-line) MIDI Express XT completely mangles some SysEx messages. Oh the amount of time I have wasted trying to diagnose various (seemingly unrelated) issues!

Doing some research, it appears this has been an issue going back to 2012, and it seems MOTU have no intention of doing much about it. I have raised a Techlink but never heard back.

Are there any known workarounds, or do I resign myself to the fact I need to offload this pile of junk and get something that works? Has anyone heard of MOTU are working on a fix? I saw another thread where a user was able to reliably reproduce the bug just a few months ago, and he sent his (very detailed) findings to MOTU. Any chance that after seven years they might fix this?

I’ve owned two bits of hardware from MOTU now, both bought new, and both have been incredibly problematic.


Thanks!

[monkey man]

It's been many years since I had my gear up-and-running, but back in the day I always sent and received sysex through my Express XT units one message or stream at a time.

If, say, I sent setup data for a project to multiple ROMplers simultaneously, one or more of the units would be quite likely to not receive the intended message/s properly.

I'm interested to hear if this works for you. HTH, mate.

[terrybritton]

They fixed the old sysex issue with a firmware update. Have you updated to the most current firmware? It is a very important one.

I have that unit and once I started bypassing USB3 by using a USB2 external port all my problems went away. It definitely does not like USB3. (I push it really hard every day, too, and use all the ports and a lot of sysex dumps into MidiQuest.)

Terry

[monkey man]

My units are too-old to be able to accept FW updates, Terry. Maybe that's why I'm stuck with the issue I outlined and the workaround of using only one port at a time for sysex streams.

Your observation makes me nervous about upgrading to a possible new desktop Mac this year or next 'though 'cause it'd surely not offer "original" USB2; it'd be supported under USB3, which would be "different", yeah?

[seedylee]

Thank you all for taking the time to read my post and your suggestions.

I am indeed running the latest firmware.

Someone on Gearslutz posted some really good observations of the issue a few months ago, so presumably other people are continuing to have problems. What they noticed is:


[quote]After generating thousands of valid SysEx messages, carefully looking at forwarded messages, comparing corrupted messages to the originals I finally discovered the exact edge case that causes the bug to appear.


If your SysEx message is longer than 130 bytes, and in the byte positions of 64, 128, 192, 256, 320 etc. (multiples of 64) you have the same value for consecutive positions the latter byte will be dropped.

[terrybritton]

My units are too-old to be able to accept FW updates, Terry. Maybe that's why I'm stuck with the issue I outlined and the workaround of using only one port at a time for sysex streams.

Your observation makes me nervous about upgrading to a possible new desktop Mac this year or next 'though 'cause it'd surely not offer "original" USB2; it'd be supported under USB3, which would be "different", yeah?

No need to worry - a USB2 USB port is very inexpensive, and works fine as an interface between the USB3 port on the computer and the XT.

I'm on Windows so my experience with Firewire has been very bad, even with the "oft-suggested" Texas Instruments chipsets on the boards.

Terry

[terrybritton]

Thank you all for taking the time to read my post and your suggestions.

I am indeed running the latest firmware.

Someone on Gearslutz posted some really good observations of the issue a few months ago, so presumably other people are continuing to have problems. What they noticed is:

After generating thousands of valid SysEx messages, carefully looking at forwarded messages, comparing corrupted messages to the originals I finally discovered the exact edge case that causes the bug to appear.


If your SysEx message is longer than 130 bytes, and in the byte positions of 64, 128, 192, 256, 320 etc. (multiples of 64) you have the same value for consecutive positions the latter byte will be dropped.

That is very strange sounding to these ears. Like I said, I shuttle sysex to and from my units many times a week using MidiQuest to the Ensoniq ESQm, Yamaha TX81Z, a Kawai K5000s and two Korg Wavestations (one an SR). Perhaps they ALL use smaller chunks??? (I seldom send large dumps from the units themselves. So maybe MidiQuest is more conservative in how it gathers its info.) However, 130 bytes is such a small amount that I really doubt this person's report reflects everyone's experience. Are you certain this report does not pre-date the firmware update that was made to fix the sysex problems?

TL;DR: I suspect you have a different problem.

Terry

[monkey man]

My units are too-old to be able to accept FW updates, Terry. Maybe that's why I'm stuck with the issue I outlined and the workaround of using only one port at a time for sysex streams.

Your observation makes me nervous about upgrading to a possible new desktop Mac this year or next 'though 'cause it'd surely not offer "original" USB2; it'd be supported under USB3, which would be "different", yeah?

No need to worry - a USB2 USB port is very inexpensive, and works fine as an interface between the USB3 port on the computer and the XT.

Thank you so much, Terry! Good to know, mate.

[seedylee]

Thank you all for taking the time to read my post and your suggestions.

I am indeed running the latest firmware.

Someone on Gearslutz posted some really good observations of the issue a few months ago, so presumably other people are continuing to have problems. What they noticed is:

After generating thousands of valid SysEx messages, carefully looking at forwarded messages, comparing corrupted messages to the originals I finally discovered the exact edge case that causes the bug to appear.


If your SysEx message is longer than 130 bytes, and in the byte positions of 64, 128, 192, 256, 320 etc. (multiples of 64) you have the same value for consecutive positions the latter byte will be dropped.

That is very strange sounding to these ears. Like I said, I shuttle sysex to and from my units many times a week using MidiQuest to the Ensoniq ESQm, Yamaha TX81Z, a Kawai K5000s and two Korg Wavestations (one an SR). Perhaps they ALL use smaller chunks??? (I seldom send large dumps from the units themselves. So maybe MidiQuest is more conservative in how it gathers its info.) However, 130 bytes is such a small amount that I really doubt this person's report reflects everyone's experience. Are you certain this report does not pre-date the firmware update that was made to fix the sysex problems?

TL;DR: I suspect you have a different problem.

Terry

Thanks Terry.


So in my case, the issue is manifesting when trying to do a SysEx dump from a Moog Sub 37 to a Korg Kronos or a Yamaha QY700. I am running the MOTU standalone to perform MIDI routing.


If I connect the Moog directly to the Kronos or QY700, a dump is 760 bytes and works correctly. If I incorporate the MOTU into the mix, the dumps suddenly go to between 753 bytes and 759 bytes - information is going missing.

Any ideas?

[terrybritton]

It may be the buffer size, number of buffers and speed settings. I know that in MidiQuest it gives different recommendations as being ideal for different pieces of hardware, so a "one-size-fits-all" setting might not be ideal in this situation. Can you change those settings anywhere?

MidiOx is a very dependable piece of software for the PC for doing sysex transfers and has the utility to change such settings - see if there is an equivalent for the Mac. Often smaller buffer sizes are the trick, but sometimes larger ones are needed. According to MidiQuest's Michael Lambie, a good place to start is to set the Buffer Size to 80000 and set # of Buffers to 10.

Also, with the MidiExpress XT in the loop, there are twice as many connections, so make sure your MIDI cables connectors are squeaky clean.

Terry

[seedylee]

Hi Terry,

There is no computer in the loop anywhere - this is purely hardware to hardware via the MOTU. If I do the routing via the computer, using MidiOX or similar, the transfer works. If I connect the hardware devices directly without the computer, the transfer works. It only fails when the MOTU is in between, and only between some devices. Most devices I have have Sysex transfers smaller than 130 bytes (it breaks the transfer into many smaller messages, which may be why they're successful).

All other MIDI functions are working great.

Based on that, any thoughts?

[terrybritton]

Hi Terry,

There is no computer in the loop anywhere - this is purely hardware to hardware via the MOTU. If I do the routing via the computer, using MidiOX or similar, the transfer works. If I connect the hardware devices directly without the computer, the transfer works. It only fails when the MOTU is in between, and only between some devices. Most devices I have have Sysex transfers smaller than 130 bytes (it breaks the transfer into many smaller messages, which may be why they're successful).

All other MIDI functions are working great.

Based on that, any thoughts?

Well, if the computer "talks" through the XT alright, but the hardware device does not, I can only surmise that the hardware device is doing something incompatible with the XT that the computer software corrects in the pathway.

But now I would give MOTU a call (they always pick up pretty quickly nowadays most times of the day): (617) 576-3066.

If you cannot manage the call let me know and I'll run it by them for you.

Terry

[seedylee]

That would be awesome; thanks Terry. I might do that tonight.

Last time I dealt with MOTU support to exchange my 828mk3 they were awesome. Going back to when I originally purchased it in 2011 was a slightly less pleasant experience

[seedylee]

If anyone wants to buy a like-new MIDI Express XT that doesn't work with SysEx from Moog, DSI or Yamaha synths, but mostly works for non-SysEx Data, hit me up. It also doesn't work properly with USB 3 ports, but it does seem to work okay via a USB 2 hub. I'm getting rid of my MOTU hardware due to a litany of ongoing, unresolved issues. I may replace it with an older Timepiece AV for standalone pstchbay duties, but I'm open to suggestions on that front.

[stratblue]

For anyone reading this, I sent the sysex backup of a Line 6 rack delay unit back to the unit and it was bricked by this useless bit of kit. I assumed it was the Line 6 delay and it happened again to a second unit which belonged to a client. Avoid.