[sdiy] USB and the art of MIDI

[ASSI]

On Samstag 19 April 2008, Batz Goodfortune wrote:
> Y-ellow Humans.
>          I'm wondering if anyone here has tried those little, Cheap,
> Chinese USB to MIDI adaptors? More specifically, has anyone tried
> using 2 or more out of a hub at the same time?

That should work in principle, but I'd rather use a USB2.0 high-speed 
hub if more than four MIDI interfaces are in use at full bore.

> Even more specifically, I'm wondering if it might be possible to make
> an ad-hock, expandable, multi-MIDI interface with say 10 or more
> I/Os. I'm not worried about the electrical limitations. IE: Power
> consumption and so fourth. I can take care of that with suitable
> hacks. All I'm interested in is if the computer would see multiple
> individual MIDI interfaces or get thoroughly confused and see them
> all as just one single MIDI interface. Or something in between?

Depends on what you mean by "computer".  With a modern Linux I don't see 
any problem, and all interfaces should show up as seperate devices 
(they may have multiple ports).  Getting them to always use the same 
device name and MIDI port number in ALSA will be a bit tricky however, 
normally they will show up in the order they are found during device 
discovery.  MIDI interfaces which need a driver downloaded at each 
plug-in also often require manual configuration, but class-compliant 
interfaces just work.  If the USB descriptors are halfway 
spec-compliant each interface is listed with the name given by the 
manufacturer so it is relatively easy to tell them apart.

WinXP has an arbitrary limit of 10 USB MIDI devices and it counts the 
same device on a different USB port as a seperate device to boot; also 
it will show class compliant interfaces as "USB loudspeaker" with no 
indication as to which interface you're looking at and hideously breaks 
sysex messages that are longer than the send buffer.  USB MIDI does not 
work at all in Win98/SE or earlier (just plugging in such an interface 
will blue-screen the computer instantly) and is probably even buggier 
than WinXP on WinME.  The sysex bug is supposedly fixed in Vista.  
Sysex with interfaces bringing their own drivers is hit-and-miss: if 
they used the Microsoft demo code it won't work, if they actually 
checked the result they may have fixed it.  Sometimes it works and then 
they break it again with a new driver version...

Dunno how the Mac is stacking up in that department, but at least it 
should show the proper device name if the interface provides a suitable 
string descriptor.

> Anyway, If there's anyone out there who has experience with these
> things I'd appreciate at word from the wise.

If timing is important to you, then this setup is probably too 
troublesome all-around.  What may work with one hub may not work with 
another or even when the order of interfaces somehow changes.  With the 
prices I can get here it is still more cost effective to have 4x4 
interfaces directly hanging off the computer's USB ports.  I also find 
the 1x1 "cable interfaces" appealing for their lack of another box to 
have around, but I haven't had a chance to try them yet.


Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk]>+

Waldorf MIDI Implementation & additional documentation:
http://Synth.Stromeko.net/Downloads.html#WaldorfDocs