[sdiy] Caution with DIY MIDI cabling

[rsdio at audiobanshee.com]

You are correct that the cable ground and shield can be connected at both ends of a MIDI cable with no problems, but this assumes that each separate cable is kept separate from all other cables, as well as individual jacks being kept separate from each other.

In your patch panel, connections between shield and the metal plate would effectively connect all jacks together, meaning that In and Out could be grounded together. You have a point that each MIDI device in your rack would still have isolation on its own MIDI In port, but your physical construction could end up connecting the MIDI Out grounds together for potentially all of the MIDI Devices in your rack.

I’ve seen this happen when I accidentally connected MIDI Out from one device to MIDI Out of another device. Hum started immediately. There probably was no damage to the MIDI Out circuitry, but the devices were certainly not designed to deal with the mistake of connecting output MIDI grounds together.

So, I’ll revise my original suggestion. Instead of isolating just the MIDI In jacks, you’d actually have to isolate each of the MIDI Out jacks. At that point, you might as well just make sure that every MIDI jack is isolated, and then you wouldn’t have to designate them as “In” or “Out"

Ove has a very good point. Grounding errors should not normally cause MIDI data errors. However, at some point all digital signals have a ground reference, before the conversion to the current loop of MIDI is involved, and a significant ground loop in a rack of MIDI gear could actually cause bit errors.

I’ve also had problems with digital audio data glitches when power spikes causes two or more digital audio devices to see different ground references. In this case, there’s no optical isolation, but I still think that enough of a ground loop between two or more MIDI devices could actually affect the non-isolated digital signals. In other words, the optical isolation of MIDI is only effective if you maintain the ground isolation between units. As soon as you have a ground loop, there’s no longer any isolation. It’s just that the problem isn’t coming through the MIDI current loop - it’s coming through the MIDI shield ground.

Anyway, sorry if this is way too much analysis. I happen to be always on the lookout for potential problems with designing musical electronics, so this is a particularly interesting scenario to consider. Thanks for sharing, M.E.C.

Brian Willoughby


On Dec 26, 2018, at 6:58 PM, Michael E Caloroso wrote:
> Only true for the ports in the devices, not for the cabling.  If you
> think about it, it makes sense.
> 
> I am not using a generic OTS MIDI panel, I built my own MIDI panel
> using raw materials.
> 
> MC
> 
> On 12/26/18, rsdio wrote:
>> Hmm, I think the reason that this is happening is that MIDI In and MIDI Out
>> jacks should be wired differently. MIDI Out should connect the shield
>> ground, while MIDI In should not. It’s easy to manage in a synth, where
>> there is only one of each (well, two MIDI “out” if you count MIDI Thru).
>> 
>> I assume that your DIY MIDI I/O panel was perhaps generic? … so that the
>> panel does not distinguish between In and Out, but depends upon which port
>> you connect internally to the rack gear?
>> 
>> It might work if you had one row dedicated to MIDI In, with isolated ground,
>> and another row dedicated to MIDI Out, where you wouldn’t need to worry
>> about grounding issues. Seems like that would be the only way for a generic
>> panel to work, although you would be limited by the dedicate Input versus
>> Output/Thru designations.
>> 
>> Brian
>> 
>> 
>> On Dec 26, 2018, at 6:14 PM, Michael E Caloroso wrote:
>>> I build my own I/O panels for my racks, and my own MIDI cables.  I
>>> recently uncovered a nasty bug that has perplexed me for years:
>>> 
>>> If you use metal shell MIDI plugs on your DIY MIDI cables, DO NOT
>>> ALLOW MIDI GROUND TO COME IN CONTACT WITH THE SHELL.
>>> 
>>> I was gigging my road system and was experiencing intermittent MIDI
>>> errors.  After finally being able to reproduce it at home, I figured
>>> out that the metal shell of the MIDI plug shorted to earth ground via
>>> the metal DIN jacks I was using on my DIY I/O panels.  Stray shield
>>> wires from the cable had contacted the cable clamp of the plug
>>> (despite my best attempts at trimming the stray wires), completing the
>>> short circuit of MIDI ground to earth ground.  MIDI gets very upset
>>> under this condition.
>>> 
>>> I use an EBTech cable tester after I build all my DIY cables, and it
>>> does not have the capacity to detect a short from the shell to any of
>>> the pins - it's not supposed to.  Even with my diligent testing after
>>> completing a cable, this bug had eluded me for YEARS in my studio
>>> system.  Yes I was well aware that the MIDI specification grounding
>>> system is designed that way FOR A REASON.
>>> 
>>> The only way to guarantee complete isolation of MIDI ground from earth
>>> on the metal shell of the plug was to disassemble the plug and add
>>> shrink tubing over the trimmed end of the cable.  This separates any
>>> stray shield wires from the metal clamp of the plug.  Once I fixed my
>>> cabling for the stage rig, it behaved MUCH better.
>>> 
>>> Nasty bug, this.
>>> 
>>> For years I had wondered why my SYSEX dumps were never 100% reliable,
>>> now I know why.