[SML] Fwd: DMX hardware

[Dale Farmer]

Hi everyone.
   Friend of mine who likes to understand things in excruciating detail 
was doing a lighting gig and found one of those weird DMX control 
failures.  Then he chased it down to root cause, because he enjoys that 
sort of thing.   (You should see his pages on the Toyota Prius)
I'll let him explain, as he does so much better than I.

  --Dale



-------- Forwarded Message --------
Subject: [tf at techno-fandom.org] DMX hardware
Resent-Date: Tue, 5 May 2015 15:26:45 +0200 (CEST)
Resent-From: hobbit at avian.org
Date: Tue,  5 May 2015 13:26:00 +0000 (GMT)
From: *Hobbit* <hobbit at avian.org>
To: tf at techno-fandom.org

On the Armory fashion-show gig last weekend I had a bit of dimmer trouble.
One of my grey LD360 packs stopped receiving signal when the stuff
*downstream* of it on the DMX chain was plugged in, but everything else
downstream was still receiving okay.  This was likely one of those
profoundly annoying marginal situations where the signal is just barely
spec somehow and some devices are happy with it and some aren't.

Fortunately the site had a similar pack of their own, and I swapped it in
for the show and that worked okay.  But I wanted to try and reproduce the
problem with my pack once I got it home.  Long story shorter, I found that
the RS485 receiver had developed a partial short from its (+) side to
ground, dragging down the whole DMX signal level in an asymmetric fashion.
Not in a way that prevented the pack from receiving in a test setup back at
home, but driven by the Hog widget instead of the Armory's Chauvet board.

First off, the DMX interface chip in the Leprecon pack is a 75176B, a 
typical
but older-vintage RS485 *transceiver*.  Commonly used in countless widgets;
refer to  http://techno-fandom.org/~hobbit/lighting/dmx-signals.txt  for
some earlier research notes on this.  They generally get hardwired to one
function or the other in these applications.  So my first thought was that
perhaps the chip had gotten confused and was trying to enable its 
transmitter
at the same time, but it turned out to simply be a passive, resistive short
of about 240 ohms from the "A" lead to ground.  In receive mode *or* in a
powered-off state the chip is supposed to tri-state itself at high
impedance, and the one in this pack apparently wasn't.

But it would still receive without the downstream stuff, about 200 more
feet of cable and the Armory's CD80 pack and my terminator, plugged in.
It wasn't a dead enough internal short to kill everything, just cause
major headaches.  In fact it wouldn't receive with just the next leg of
adapters and 100 feet or so of *my* wire plugged in, so I marked all that
and kept it together at strike to try and make sure none of *those* cables
were bad.  But then the same setup worked at home driven by the Hog,
although I could now scope it out and see how this pack was dragging the
line down.  [The other pack is fine, causes zero signal degradation when
plugged in.]  At home, termination lowered the bus signal even more but
not even to a point where either pack couldn't see it.

It is posslble that this pack has had the problem for a while, and we just
happened to get away with it in other setups like Arisia.  It is also
interesting that 100 feet farther of *unterminated* wire would cause the
signal loss at the Armory, placing things firmly in the Land of Subtle.
What I suspect was happening is that the slew-rate limited Hog output
greatly reduces edge ringing and makes the terminator less important,
but the Armory's Chauvet probably also has a 75176 chip driving the line
which is *not* slew-rate-limited and was firing harmonic-rich square waves
into this degraded-impedance environment and that got things *just* sub-
marginal enough to cause signal detection loss at a receiver that
wasn't quite right on its own.

Fortunately the 75176 is a socketed 8-pin DIP, and trivial to replace once
the pack is opened up.  As they're used receive-only in the dimmers their
output slew rate doesn't matter, so I'll probably just get a pair of new
ones [possibly from a different manufacturer] and update both packs.  But
one thing I learned about testing for DMX issues, albeit on the bench back
home instead of on site, is that a simple ohm test across pins 2 and 3
and then from each of those to ground, taken anywhere on a failing chain
with the driving device unplugged, could possibly reveal a failed receiver
somewhere along it [or even bad cabling] which could then quickly be found
and bypassed.

_H*