DMX Polarity on the Roboscan 918

After a decade or more of ownership, not to mention wiring adaptation at every gig they went to, I finally dug in and "un-flipped" the DMX data line polarity on my Martin 918s. They had originally come in "Martin RS485" mode, where pin 2 of the XLR connector is data (+) and pin 3 is (-), but the standard DMX spec is the reverse of that [on both 3 and 5 pin connectors]. With only the four of these lights generally used in local pairs, a small fistful of data-line adaptors got me through all the shows to date, but now that someone else was about to work the units into his own design it was time to make them more compatible with modern setups.

Jumper documentation

[Small pix are linked to larger versions]
Opening the top of the light The jumpers in question are on the main logic board, located under the end cover at the "top" [output-mirror end] of the light. They're right behind the circled connector, removal of which [and possibly the other white one to its left] makes tipping the board out much easier as the harnesses aren't of particularly generous length.

Flipped DMX (old Martin style) Straight DMX (compatible w/ modern systems)  
Flipped Straight  
The jumpers are right behind the connector as indicated. Fixing the polarity requires removing both jumpers and restoring them to the new position shown. An easy enough fix, just a bit cumbersome to get to. It looks a lot easier as presented in the manual (pdf, 932k), but having a few real-world pictures and discussion can't hurt.

If you think about what this needs to do electrically, it becomes obvious that the square layout of jumper-block pins has to have its circuitry routed something like this:

       XLR pin 3 in -------o   o------ data (+) to board

  data (-) to board -------o   o------ pin 2 in from XLR
and placing the jumpers straight across or 90ed acts like a typical double-pole reversing switch, similar to four-way types used in household electrical systems where more than two different wall switches can all control one light. That's all there is to it.

It's also ironically telling that many more modern units can automatically sense which polarity they're receiving and simply adapt on the fly without human intervention. Why Martin ever gained enough market influence to get away with making gear in violation of the spec is beyond me, but then again the "spec" was still rather new back when they were ramping up their product lines and maybe they just didn't know which way would ultimately be considered the right way. At least they made it user-configurable in some of the later units; the Roboscan 518, for example, is hardwired on the "wrong" Martin pin-2-hot setup and since I had been using some of those in rigs I worked with, that had been one prime motivator for leaving the 918s set "wrong" as well all those years.

Be careful with threading! Because the case screws thread into partially-open extruded channels in the unit frame, lit "bluer" from below here, one has to be super-paranoid about NOT stripping those with the self-tapping screws used. A touch of anti-seize on the screw threads makes this all go back together much more easily, and one must be absolutely sure to thread gently back into the original cuts and avoid chewing up the channel with new thread paths, and never overtighten.
Despite these minor caveats, the 918 possibly represents a pinnacle of solidly-built moving-light frames and most gear since their heyday has gotten quite a bit more flimsy and underbuilt. The burgeoning Asian price-competitive market hasn't helped this, of course; but for gear that's going to routinely go on the road it's worth the premium to avoid the cheapass "DJ-grade" gear for the sake of longterm reliability. Other than having to replace one slightly out-of-spec thermal sensor early on, these units have been rock-solid for me and excellent bang for the buck, with the minor penalty of more weight to haul around. I generally prefer scanning-mirror types over moving-heads just because of the speed, especially useful in a fast paced dance-floor environment where you really want to be able to slash beams around.

_H*   140106