Issue #5, October 1997
A True Story. This past summer I had the opportunity to visit Las Vegas. The rock band U2 was in rehearsal for their forthcoming tour. A tour rehearsal, I've learned, is not only a rehearsal for the band but is also a rehearsal for the sound, lighting, video, etc. A Doug Fleenor Design DMX interface was "rehearsing" the roll of xenon searchlight control, and having trouble (or so it seemed).
A two channel relay card was being used to enable the searchlight and to strike (ignite) the fixture. Unfortunately the light would turn itself off several minutes after it was turned on. My over-the-phone diagnosis was that the relay card was glitching momentarily and turning off the light. I faxed the technician in Las Vegas a hardware fix which would prevent the relay from dropping out for about 1/2 second. The fix worked but I felt it was a band-aid solution; I wanted to know what was causing the glitch.
The next day I arrived in Vegas, oscilloscope in hand. The first check was to see if the microprocessor on my card was failing due to noise (arc lamps can generate electrical noise), power supply problems, or some other cause. The card seemed to be doing fine. Next I drove the DMX line at the worst offending searchlight with a Goddard Design lil DMXter... no problems. We hooked the offending searchlight back up to the DMX network and I watched the scope for a glitch. Within five minutes the relay dropped out for about 1/40 of a second. My first clue! A common DMX512 data rate is 40 updates a second; the DMX input likely told the relay to drop out. I confirmed this by looking at the DMX input during the next glitch (with a digital storage 'scope you can capture an event like a glitch and what was happening before, during, and after the glitch). So the relay card was doing exactly what it was being told to do, but why was the DMX input failing?
The DMX network was fairly elaborate. The lighting console sent analog control to the stage. At the stage a (competitor's) analog to DMX512 converter generated the DMX which was sent to a (competitor's) five way splitter. The five isolated outputs fed searchlights near the stage. A feed-through connector on the splitter was used to go back to the front-of-house (FOH) position where another five way splitter fed searchlights around the audience. The feed-through connector on the FOH splitter was used to feed a sixth audience searchlight. The problem was finally traced to a faulty connector on the cable from the FOH feed-through to the sixth audience searchlight. How could one intermittent connection cause so many problems?
REFLECTIONS! The problems were caused by signal reflections that occurred when the bad connector caused an open circuit. This removed the termination resistor from the end of the line. When a long DMX512 control cable is not terminated, the digital signal is reflected back up the control cable. The reflected signal can combine with the direct signal and make the signal unreliable. Since the feed-through connector had been used on the first splitter, and the failure point was on the feed-through connector of the second splitter, the entire network was affected. If the network had not used feed-through connectors (using isolated splits instead) only the searchlight on the bad cable would have been affected. If only one searchlight had been affected the crew probably could have found the bad cable instead of suspecting the searchlight control cards. But then I wouldn't have had an excuse to fly to Las Vegas.
Doug Fleenor Design splitters do not have non-isolated outputs (except the model 1211, designed for fixed installations). There is no feed-through to be used improperly and no termination switch to be in the wrong position. Inputs are automatically terminated and outputs are individually buffered.
DMX tip of the month. Avoid using feed-through connectors except to go a short distance to the next device. This will improve reliability, assist in troubleshooting and help us to sell more splitters.