yeah, the sequence is consistent: audio in/out OK, combined with midi out (eg: Mac sends level LED updates on Z2) - everything fine, then add some midi in (eg. z2 fader data to Mac), and gone. single packets, such as the browse encoder take longer to reproduce, so I suspect it is just about getting the timing between the packets right. The midi in and midi out are on different endpoints. I have to say the packets/sec for in data is much higher than the packets/sec for out data. IOHIDLibUserClient:0x10000087b setReport failed: 0xe00002d6 (the timeout) repeats at 5sec intervals while the audio continues. Two possibilities: 1. noting that in/out data is on different endpoints, the Z2 could be losing the request (in a buffer or the USB controller?) from the Mac to send data on the USB line? or 2. IOHIDLibUserClient (or close to it) is for some reason not sending out the midi data (known as the "complete") from the Mac to Z2 after sending a "submit", resulting in the timeout. (or IOKit is getting in the way?) (<-- NI seem be supporting possibility #2 by passing the issue on to apple to solve). I am also wondering about Driverkit taking over IOkit in future releases?