MIDI recording problem with PT HD w/HDX

[Christopher Dobrian]

My system is:
Mac Pro 2 x 2.4 GHz Quad-Core Intel Xeon, 32 GB RAM
Mac OS 10.8.5
D-Command 24-channel control surface
Avid Sync HD, synching two HD I/O interfaces, and one Apogee Symphony I/O for a total of 24 channels of I/O (we use the analog inputs from three different 8-channel mic preamps, one of which is a Grace m802)
XMON audio monitor
HDX card recently upgraded from our original Accel 3 system
MOTU MIDI Express XT MIDI-USB interface
JL Cooper Surround Panner
Pro Tools 11.2.1 and Pro Tools 10.3.9 (we upgraded from PT 9 HD at the same time as we upgraded to the HDX card)

Audio recording by itself is working fine, but MIDI recording (since upgrading to the HDX card and PT HD 10/11) sometimes causes crashes, and almost always has timing errors, spurious data, and dropped data. This does NOT happen when recording MIDI on the same system with Logic or Max.

Additional bit of information: Pro Tools often reports loss of communication with the Grace m802 mic preamp, which normally works fine in emulation of a PRE controlled by the D-command. Since that's a MIDI communication, between the m802 and the D-command. I attribute those lost communication messages to the same MIDI data loss problems we have been having when recording, i.e., to dropped ping messages between Pro Tools and the m802.

In short, MIDI recording is simply not working right. Sympotms are apparently: dropped MIDI data (resulting in stuck notes when a note-off message gets dropped), rhythmic timing errors due to delayed data, occassional appearance of recorded data not present in the input. Problem does not occur on the same system with other software.

In the attached screenshot graphic you can see the results. The input was only short notes, repeated triads. Wherever you see long notes, those were not in the input. They seem to be the result of dropped note-off messages, which then affects the duration of every subsequent note on that pitch. At 12 seconds you can see a spurious note-on for D that was not even present in the input (which then causes subsequent elongation of other notes of the same pitch).

Any help anyone can offer in debugging or addressing known issues will be much appreciated!