(This is an old post, but I do plan on posting an update, someday, when I get a digital scope.)
[lighting nerd speak]
The USITT PLASA specification for DMX-512A requires that the last physical device in any DMX link, whether it connects to one device or 31 others, be terminated with a 120Ω resistor. This is to prevent unwanted “reflections” coming down the line and corrupting the DMX data. Here’s the theory: DMX-512 signals travel at around 60% the speed of light. Although fast, this is not instantaneous. When a signal traveling at over half the speed of light hits the end of a cable, a percentage of the signal is reflected back down the signal chain. Each digit of the digital signal is placed on the chain for 4ms. (Milliseconds) The receiving device looks at the value of the digit in the center of the 4ms. That is 2ms after the digit is placed on the line. In 2ms a signal can travel down and back about 590 feet of cable. If a device is sitting on a cable such that it receives the reflected signal as well as the initial signal, it sees two numbers at once. Badness ensues. However, in practice – and I feel pretty confident saying this – people are pretty lax about termination, and termination problems are so rare that I’ve only seen a problem diagnosed to be a termination issue ONCE in my career, with some very old Martin scanners that had been sitting in some guy’s garage in god knows what state of repair for yearsssss. Why is this?
Doug Fleenor DMX might say that it’s because “DMX is incredibly forgiving, it simply works until it doesn’t.” (In fact, he says this.) That’s not really good enough for me. Perhaps manufacturers make their lights so that they have a built-in impedance sufficient to prevent reflections from happening, and so we don’t usually see them. So, armed with a oscilloscope, I set out to find out.
And what did I find? Spoiler alert: I found that my old, analogue oscilloscope has insufficient resolution to allow me to see the effect that termination has. My experimental setup was as follows: my old ProTek oscilloscope with the probes attached to the hot and cold pins of a cable coming from the middle of the fixture chain, both grounding clips on the ground (obviously). I used one of the outputs of a fixture that had 3 and 5-pin in and outs, assuming that the pins were simply “dumb” wired together. (Meaning, no electronics in-line to do anything to the signal, just ganged straight together.) The output of the grandMA was attached to the fixture chain, which consisted of six fixtures of various manufacturers (Martin, Elation, and Chauvet) and a single 120-ohm terminator alternatively added to the end of the chain, or not. Here are some pictures:
I set my scope’s time setting for about as high as it would go, and adjusted the voltage range to about one-quarter the scope’s range, so pretty low. I only listened to one of the inputs at a time, but I did check to see that I was seeing data on the hot and cold pins. On the MA, I made a universe of nothing but 8-bit dimmers and made an effect which was a simple sine wave on the entire universe, so all 512 dimmers went up and down at the same time. I did this so that I could be certain that I was seeing something. With the lights going absolutely bananas (because of the crazy signal they were seeing), this is what I saw with the unterminated chain:
Unterminated DMX signal from Blueshift Design on Vimeo.
And this is what I saw with the terminated chain.
Terminated signal from Blueshift Design on Vimeo.
Whelp, that’s unhelpful. The signal amplitude clearly falls, and the squiggles at the beginning of the signal look like they go away. But, I have no idea what those little sine waves represent – I’m pretty sure I’m seeing some portion of the DMX packet, and I’m pretty sure there shouldn’t be any sine waves on the signal at all. If they were data reflections, I’d imagine that I’d see them all over the signal, not just at the beginning, so their apparent reduction might just be an artifact of the terminator reducing the amplitude of the signal somehow. Perhaps I should not have had the effect running, it might have been making it very difficult to see the data I was looking for. The upshot here is that I lack formal oscilloscope training, so I’m pretty sure that what I was looking for wasn’t going to be seen in the way I was looking to find it. I expected to see something that looked like this:
