by CopperPhil » 14 Apr 2017, 09:30
Many thanks for this feedback, very interesting!
It seems to confirm what I thought. The global system is going down due to some MIDI traffic, in this case it look like it's due to the clocks.
The MIDI clock is generated by the MPC, do you mean the Akai MPC through a AL88 MIDI IN port? You send it to 12 destination: one to port 1 of each AL88, but what are the other destinations? What is the clock rate?
Important: do you also sent MTC alongside the MIDI clock?
Some explanation: in CopperLan, MIDI Clock is transferred as a real-time message, with very low latency and with disabled delivery confirmation sub-system. On the other hand, MTC is a MIDI sysex message, and this kind of message (Sysex) is transmitted through the CopperLan network using fully synchronized data transfer packets, definitely not designed to be used in real-time. What you describe makes me thinking of an overload of this data transfer handshaking subsystem, causing at some point data queuing (just because the sending AL88 must wait for the each receiving AL88 before sending the Sysex to the next one), and finally memory crash if the sending AL88 is not able to send the sysex to all destinations before receiving the next one.
-> So, if you are using MIDI MTC, please disable it just to check in a first time that this is the cause of the problem. Then, use it only where needed (not sure that all your destinations need MTC), and finally try to reduce the amount of MTC per second.
Again, there would be not problem with the MIDI Clock because we put a lot of care to make this working really well, with very low latency. But Sysex is a particular type of message that should never have been used in real-time situation.