Osmose ce drops midi connection

Osmose CE 61, macOS, Logic Pro. Intermittently the Osmose stops sending note events to the Mac. Logic sees nothing; MIDI Monitor (Protokol) also sees no note events on any port. A macOS reboot always fixes it. Nothing else does.

What I’ve ruled out:

•	USB / enumeration is healthy. ioreg -p IOUSB -l -w0 shows Osmose CE 61, vendor Expressive E, registered, matched, active, and UsbExclusiveOwner = MIDIServer. So the Mac has the device and CoreMIDI owns it correctly.
•	The connection is live and bidirectional. When I quit Logic, the Osmose display reacts. So Mac → Osmose MIDI is flowing. It’s specifically Osmose → Mac note data that’s missing.
•	Power-cycling the Osmose doesn’t help. Neither does unplugging/replugging the USB cable.
•	CoreMIDI-side resets don’t help. Killing MIDIServer, removing greyed-out devices in Audio MIDI Setup, and a full logout/login all failed. Logout/login brought every other MIDI device back but not the Osmose.
•	Only a full macOS reboot restores it.

Given that USB and CoreMIDI both look correct and only a reboot fixes it, I suspect this is Osmose-side runtime state rather than a Mac problem — possibly haken local control being auto-switched off by a DSP ping (CC117) arriving on the Haken port from Logic, which a reboot would reset. But I haven’t confirmed that, and I don’t understand why a reboot of the Mac would reset state on the Osmose, which is what makes this confusing.

Do you have a Hardware synths that you could connect to the Osmose instead of the Mac? To Take one variable Out of the equazion(and replace IT with another one :slight_smile: )

very odd, you’d expect unplugging the usb cable to basically reset the connection from the macOS side, and similarly rebooting the Osmose to reset any state it has.

if you just run Protokol and not logic, does the same thing happen?

as you say, Id wonder about some kind of midi loopback perhaps causing an issue, as you say, something like turning local control off - but why that would not be resolved by restarting things?
also, if it was a loopback, id expect it to happen very quickly, within a few seconds.

anyway running just protokol, no daw, means you shouldn’t have any possibility of a loopback.

does this even exist on the CE?
the CE doesn’t have the Haken board, its just using its own SoC to do the midi.

I guess they could have implemented for similar reasons to haken, its basically an auto loopback detect. ( which Id prefer they didn’t have - lol)

Id send in a report ticket to Expressive E, its a new product and no doubt has various tweaks in the firmware from the Osmose, so something odd may have crept in.

1 Like

No clues in any of the macs log files? I only ask since you are clearly more than capable of digging into the tech details more than most users.

Any clues in a MIDI monitor thats kept running before the crash even happens, to see what the last messages were in both directions before things went wrong?

Hmm did not try protokol only, possibly Logic does too much, good one. Then again, if i tear down the usb and midi stack and bring them back up, logic should not be able to do silly stuff with midi. Also i cannot phantom any service “swallowing” midi so that protokol not see it but at this point i will try anything.

Yeah, i did do a bit of digging but boy there is so much stuff and these services “ hide” themselves with fun names. But you’re right, i need to do some digging there.

Nothing at all. Osmose plays fine for days. I have the (bad) habit of leaving it in all the time (protokol and other tools also) so i should capture some event if there was one. Now i have a sneeking suspicion that its midi stream gets kill/stopped/mangled after thr mac wakes up - perhaps relates to macos turning on nap or drive sleep but that would have of course nothing to do with midi … you’d think.

I always ‘sleep’ my macs, they pretty much are only ever rebooted for a macOS update.
I’ve never had a midi device (including Osmose (non-ce) ) stop sending midi.
similarly, your lower level checks, seems to show the usb device is still connected.

frankly, for whatever, reason, I think most likely its something the Osmose CE firmware is just not sending midi any more… or perhaps, its some how corrupting the usb devices record on the mac side.

I get that none of this makes sense, as rebooting osmose, unplugging usb etc should all reset it, but given this is a new product with new firmware, the main suspicion has to go to the ‘new kid on the block’ :wink:

btw: one other thing that might be interesting to try… do the MIDI DIN ports continue to work?

if you really want to get low level, you could start looking at things like wireshark, I believe on macOS that does usb capture, see whats going on when you try to reconnect usb.

on the other side, whilst I dont check it regularly, Im on the “Expressive E Osmose” facebook group, and I’ ve not seen any reports of this there…
have you seen others reporting the same issue elsewhere? is it a ‘known issue’ ?

also how do you connect it?
directly to mac or via an adapter or usb hub? if so perhaps theres a conflict there?

whilst CE is new, so always a possibility of bugs - it’d seem odd if others were not experiencing what seems like a pretty serious issue.

This is my conclusion/gut feel also. It is almost as if a command was given/interpreted to shutoff MIDI.

Haven’t tried the MIDI output ports - I have to find something to plug in it first (haha). I did think about Wireshark (I use it for other projects) but as with the logs, there is so much data to learn what are the interesting bits. Will probably put that off till I am screaming and throwing stuff.

I haven’t seen anyone report this on the FB side and yes I did write up a ticket for expressive-e to look at, will see. I connect straight to the Mac (studio, M2) and I tried front and back ports (there is a slight power difference). I also have an M4 mini (not used for music) used for “measuring stuff”, One of the devices connected to the mini delivers a high data stream and is every so often interrupted - traced that down to a Mac port issue and a usb hub solved that problem. So perhaps Apple was a bit cheap with the Studio M2 too and I may need to stick a hub in between.

Just on the ‘seriousness’; if folks are cycling power on their Mac/pc I bet they would not see this. Secondly, it does not happen very frequently, I can go on for days before the Osmose no longer responds.

I sheepishly admit I do development work on that same M2 machine (so easily could screw up all sorts of OS related processes). yet in my defense, no other usb/midi connected device fails, just the Osmose.

when the CE fails, do other usb midi controllers still work?
this would indicate coremidi subsystem is still ok, or if its dead for all devices.
(you could try IAC but a hardware usb midi device is better)

do you have a another mac/pc?
if so, when it fails, try plugging in the CE to that, without power cycling the CE, to see if its sends midi to that.
(odd bit is removing cable and plugin in again should be tearing down the macos stack)

I assume the CE is only using standard driver? (class compliant midi),
its not using a new driver for the Ctrl-E side is it? (e.g. on a different endpoint)
if it was, then it could be that its that driver thats failing and thats blocking midi being sent out.
much more likely as the usb core midi one is rock solid )

as for reboot vs sleep, I suspect most mac owners use sleep, as its so reliable (unlike windows ;)) - if its powered (full) on 24x7 that might be less common.

it sounds like the sort of problem, that if you have enough users, then its going to get reported a reasonable number of times, esp. as rebooting is pretty inconvenient. so it’ll be interested to know if expressive-e tell you its a known issue, or they have other reports already.

whats also kind of odd, is whilst they firmware was upgrade for CE / Ctrl-E, you’d think the underlying midi stack (code) is very similar to what they have in place for Osmose… so is reasonable mature these days.

btw: if you plug it in via a powered hub, that would likely prevent any possible issues related to power saving over usb… not a problem Ive seen on macs, but you never know.
(then again, I always use powered hubs :wink: )

Actually, you just made me realize there is a specific USB-C for powering the unit on the back. I should use that when it fails to see if it brings back midi. It would be weird though, why would the connection with Logic Pro not fail and just midi note on/off etc fail due to power? hmmm.

On the Mac there are no drivers to load for midi - thats all standard stuff, comes with the OS.

Right now all is fine and the unit is work as I just cycled power on the Mac, will give it a few days…

@yohst Which OS are you running on? And which midi driver version is installed on the mac?

Data flooding, handshake timeouts, etc. can cause the midi server to shutdown on the mac. Can also get some midi feedback with handshakes negociations in a mixed environemnt…

doh - scratch that was like the first thing in the thread

PEACE
_
_
_

oh, I didn’t realise the CE could be bus-powered (Osmose cannot, it has dedicated PSU).

in that case, Id definitely put it on a powered hub, or even better power from a usb psu.

Ive seen various issues in the past with bus powered devices esp. if they have high current draw (and Id assume given the display, CE will be on the higher end) , even if its below the spec, I think often due to power saving.
I cannot say Ive experienced this with macOS particular, but have a lot with windows/linux.

so yeah, def, check to see if it still occurs when its powered via a PSU.