Understanding Specifics of the DMX512 Protocol [Newbie Nate's Question of the Day]

@Chris Pflieger - you want to explain the OSI model for Nate? He and others may not understand the "layers" to network communications.
"OSI" Oh Shitt Insurance?
EDIT: Interesting; the site wouldn't let me post s h i t, it kept changing it to crap.
I've edited and learned the site accepts shitt.

Toodleoo!
Ron Hebbard
 
To pick up one point that DrewE and other commenters alluded to:

The reason that newer generation fixtures aren't limited to 31 devices on a link is that newer generation *silicon* isn't: this is equivalent to the reason why it's possible to go longer distances than the original spec says on Ethernet *on switched networks*. Before switches became ubiquitous, the distance limitations on Ethernet had to do with being able to detect collisions, the length of the packets, the speed of light, velocity factor, and like that; if you made the network too long, you couldn't detect collisions, and all was chaos.

Since we went to switches, that's not an issue, and simultaneously, Ethernet PHYs got *much* more impressive; you can go half a mile on copper now, as long as you don't get too close to fluorescent lights. :)

A similar thing has happened with RS-485 transceivers over time; the distance/noise issues have gotten better, as has the amount of load the receiver places on the line. Different transmitters can be at different ends of the specification tolerance, as well...
 
To pick up one point that DrewE and other commenters alluded to:

The reason that newer generation fixtures aren't limited to 31 devices on a link is that newer generation *silicon* isn't: this is equivalent to the reason why it's possible to go longer distances than the original spec says on Ethernet *on switched networks*. Before switches became ubiquitous, the distance limitations on Ethernet had to do with being able to detect collisions, the length of the packets, the speed of light, velocity factor, and like that; if you made the network too long, you couldn't detect collisions, and all was chaos.

Since we went to switches, that's not an issue, and simultaneously, Ethernet PHYs got *much* more impressive; you can go half a mile on copper now, as long as you don't get too close to fluorescent lights. :)

A similar thing has happened with RS-485 transceivers over time; the distance/noise issues have gotten better, as has the amount of load the receiver places on the line. Different transmitters can be at different ends of the specification tolerance, as well...

Well, yes and no. You still need to follow the 300 ft. rule and don’t want to design an install that violates that with long cables, assuming gear is better to accommodate errors. I can think of bigger buildings that have to either do fiber or added switches.
 
I've got at least 2 different clients who have switch to switch runs over 500ft, one pushing 600.

In both cases, they're between managed switches in the $3-500 class, which have higher-class PHYs.

I tested them, at the customer's insistance, having myself thought they'd need to pull in fiber, and in fact, we never dropped a packet. 1024byte packets, flood mode, 60 minutes continuous.

From what I've gathered, current generation PHYs, in the Gig-E class, aren't electrical, they're more ... damnit, I *cannot* put my finger tonight on the digital technology that applies here. The receiver is digitizing the incoming line, and applying digital processing to extract the signal, rather than the traditional approach.

[ ETA: "DSP"; they use Digital Signal Processing to read the analog encoded signals off the cable. ]
 
Last edited:
I’m almost afraid to ask, because it will confirm that I’ve been wrong all this time, but isn’t one use of pins 4 and 5 RDM?

also, what is the official length limit of DMX on XLR5 Cable? Is it less than 300’? I have a 250’ data snake that Christie Lites made and I’ve never had any issues with it.
Thanks.
 
.... using pins 1&2.
Err, 2&3.

also, what is the official length limit of DMX on XLR5 Cable?
DMX over copper, theoretically 1640' with perfect equipment and circumstances. That doesn't mean 1641' won't work, however. Or that 1639' will always be trouble-free. Some say 1000'. It's really a moot point, as likely anything over 500' will have a different transport mechanism than raw DMX512 over copper or even fiber-optic.
 
When I used to part time for ITV Sport I was chatting to one of the OB techs and he said that they had been surprised once to find that they had been running a couple of kilometres of cable between the Cypher keyboard (on the gantry) and the crate (in the truck) problem free. The comms for that was RS-422, not that different to DMX. Often you'll get away with more than you ought, but I wouldn't rely on it.
 
I would have thought they would have done that but, no, RDM used a break in the DMX stream to poll and transmit back to consoles, computers, etc.... using pins 1&2.

My understanding is that this is related to the reality that in the time between when DMX became a standard and RDM became a standard, many cables were made with only three conductors given that pins 4 and 5 went unused. Rather than adding the additional headache of needing "rdm compatible cables" RDM was squeezed into the pins 2 and 3. It makes it grosser, but more likely to work.
 
Your Honda Civic may accelerate to 120mph... DMX may transmit beyond the specified cable length recommendations. Your Honda Civic might catch on fire... DMX might go squirrely.... If I follow recommended specifications I have a starting point of it shouldn't be the cable length. Anything after that is a question and unreliable variable. Will it work? maybe shoot probably. I honestly care less about when it works right, but want to be able to accurately troubleshoot it when it doesn't work.
 
Per the RS-485 spec, a data rate of 250K, you should be OK with something just longer than 1000'.
Assuming your cable meets spec (which Cat5 is close to meeting)

proxy.php
 

Users who are viewing this thread

Back