MIDI over long cable runs

[Magnus Danielson]

>>>>> "PP" == Paul Perry <pfperry at melbpc.org.au> writes:

 PP> At 02:39 PM 19/07/98 EDT, Michael Bacich wrote:
 >> I find myself in need of a device that allow me to transmit MIDI data over
 >> very long cable runs (a couple of hundred feet).,,,,,,,,,,,,, 
 >> The MIDI signals will be originating from the keyboard of a modern MIDI'd pipe
 >> organ, mainly just to let the organist play the occasional piano or bell/chime
 >> sound from a MIDI module, so the cable and interface won't have to deal with
 >> any high-bandwidth data such as sysex or multi-channel sequencer data.

 PP> it doesn't matter what kind of midi data is being sent, the performance of
 PP> the link
 PP> is the same (unless you are going to make a whole new protocol).
 PP> I would have thought a 50 or 75 ohm coax with video drivers. With these
 PP> lengths you 
 PP> will have to be sure there is no reflection at the ends. I imagine there are
 PP> video 
 PP> experts on  this list?

Not being exactly a video expert in any sence...

For these fairly moderate lengths (considering the blasting speed of
31250 bps) going coax seems a bit of a overkill but it will work
electrically. I doubt that it will be such a nice thing to use on
stage and nearby areas, but that is a matter of personal taste.
An RG-58 or RG-59 can survive some physical abuse, but they are quite
rigid and doesn't really compare to normal signal/mic cable.
To use these cables with a optimum situation one use driver and
receiver that will have a source and load impedance equal to the cable
it is being connected to. Thus, a video cable of 75 ohm is beeing fed
with an amplifier whos output impedance is 75 ohm and terminated in a
receiver with 75 ohm as the input impedance. Note that you should use
75 ohms BNCs with 75 ohms cables, not that it will matter at these
frequencies, but if you want to be compatible with other 75 ohm cables
you should be aware that 75 ohm and 50 ohm BNCs don't mate very
well... they do but one of them destroys the other, not a way to get a
long time reliance.

Anyway, for similar situations is the DMX light control protocol a
good reference. They use about the same speed (if memory servce me
right, I have the papers at work of some reasone) and they use
standard drivers for twisted pair (remember, 10Base-T reaches 185
meters on twisted pair at 10 Mbps). I beleive that they use RS-485.
There are plenty of drivers and receivers of this around that fits
into 8-pin DIP and require little external components (well, contact
and power decoupling cap is a got start). Some standard input
protection might be a thing to consider. DMX use 5-pin XLR, just to
avoid mixups. Picking the right cable with a good mixture of
rugginess, bendiness, damping properties, characteristic impedance
(usually 100-120 ohm) should ensure a success with this method.

Let me just remind you that things like earth-loops and shielding
practices can haunt these longer wirings. You are in luck that the
MIDI output is hooked to an optocoupler, but a consideration on how to
deal with earth loops as well as other problems could pay out in the
long run with less magic dust problems.

Cheers,
Magnus