Passively splitting DMX.

[sunray]

2 hours ago, kgallen said:

The resulting stub length on the transmission line within the fixture is a couple of centimetres, not long lengths of cable. Electrically, at DMX frequencies, this is a very different situation to making a passive split then hanging tens-of-meters of cable off each leg. Electrically, a DMX network is a transmission line, which has very specific electrical characteristic and behaviour. The tolerance to stubs and impedance mismatches causing reflections is a function of frequency. At DMX frequencies, short unterminated stubs like those inside the fixture can be tolerated whereas such stubs would be completely unacceptable at RF frequencies.

I'll go further than that and say stubs of those sorts of length are deliberately used for RF work for very specific purposes.

As an example an unterminated ¼λ length of cable at a specific frequency will actually transform the impedance from ∞ at the open circuit end to a very effective short circuit at the other. This is frequently used as a low cost/compact filter against a specific unwanted source of interference.

If a spur on a system just happened to be the perfect length for the frequency in use (that includes the data rate in DMX) it wiil place an effective short on the main line at the point of the Tee, just as effective as a solder bridge in a plug.

[timsabre]

A quarter wavelength at DMX speed of 250khz is about 300m. So as Kevin says the short length of "T" cable inside a fixture is not significant, and indeed very few DMX installations will get anywhere near 300m of cable.

As the others have said passive splitting DMX will nearly always work but can come back to bite you later at an inconvenient moment, often in combination with a different fault. If you are planning a rig - don't do it.  In an emergency - maybe but be prepared for strange things to happen.

 

I've been on a few jobs with Robe fixtures having both 3 pin and 5 pin DMX connectors, where these connectors had been used as if they were an active splitter (they aren't, they're just wired in parallel). All worked fine until a faulty DMX cable with broken "+" conductor came into play, then extremely weird random movements started happening.

[alistermorton]

I see lots of postings on Facebook groups where people say things like I never terminate and I use whatever cable comes to hand from the audio box  and in 30 years of using DMX I've never had a problem so stop being a pussy and just get on with the show. I just (virtually) shake my head and ignore them. Others who will tell you that terminators cause problems. It never occurs to them that the terminator could simply be making a problem that they were just about getting away with manifest itself.

Saw one guy say that when he added a new fixture in the middle of a run, suddenly the others started flashing and pan/tilt was jittering. I asked if he'd terminated the line - he said, no, but he'd found a terminator, plugged it into the last fixture and the flashing and jittering had stopped. "Do you think that could have been the cause?" he asked.

[sunray]

2 hours ago, timsabre said:

A quarter wavelength at DMX speed of 250khz is about 300m. So as Kevin says the short length of "T" cable inside a fixture is not significant, and indeed very few DMX installations will get anywhere near 300m of cable.

As the others have said passive splitting DMX will nearly always work but can come back to bite you later at an inconvenient moment, often in combination with a different fault. If you are planning a rig - don't do it.  In an emergency - maybe but be prepared for strange things to happen.

 

I've been on a few jobs with Robe fixtures having both 3 pin and 5 pin DMX connectors, where these connectors had been used as if they were an active splitter (they aren't, they're just wired in parallel). All worked fine until a faulty DMX cable with broken "+" conductor came into play, then extremely weird random movements started happening.

Everything there is dead right, I haven't double checked the length you quote but it feels right. The branch in a fitting (provided the interface chip presents a very high impedance!) is something like 1/5000 λ and I agree is far too short to create any issue.

I have some of these:

Which are terminated prior to the interface chip and the output follows a second interface chip and they tend to be correctly described - something liked 'buffered' etc, however the RJ45 are simply parallelled to the appropriate XLR and I've seen people trying to connect as you describe with the Robe's.

I do fit terminators but confess I don't panic to look for one however it's rare for my networks to reach say 50m,

Invariably there is a fault in the system if a terminator stops it working properly

[timsabre]

The trouble with the "buffered" type device is that it blocks RDM, though not many people seem to use that.

Terminators are very good at showing up other problems e.g. if you are missing the + or - side of the signal it will probably work without a terminator but not with one.
It seems to me from a not very statistically significant survey of temporary event rigs and installs I have seen, that hardly anyone routinely uses terminators.

[Lampy512]

9 hours ago, dje said:

Interestingly when you look inside a lot of fixtures, the male and female DMX connectors are just soldered onto one another in parallel.

Thus in essence if you are daisy chaining these fixtures, then you are passive splitting on the rear of the fixture. One leg is going into the electronics and one leg is going out to the next fixture.

Not saying it is/isn't OK... just saying, devil's advocate, maybe more people are passive splitting than realise it 😀

This is one of the arguments my colleague is making but when you look at some of the PCBs them leg going to the fixture has some circuitry I assume to act as an active split to the fixture but I'm no expert. 

[timsabre]

It's quite rare to have active split circuitry in a fixture, mainly because if the fixture fails or loses power you don't want it to break the DMX line, so then you have to put failover relays in and stuff which just adds complication and expense. Most fixtures have surge protection on the DMX input which may be what you can see.

The real answer to your "internal passive split" question is given above, it is all about how much cable is connected to each "leg" of the split, and the very short length inside the fixture is insignificant, so it doesn't cause the problems that you can get with a passive split with long wires on it.

[kgallen]

8 hours ago, timsabre said:

A quarter wavelength at DMX speed of 250khz is about 300m. So as Kevin says the short length of "T" cable inside a fixture is not significant, and indeed very few DMX installations will get anywhere near 300m of cable.

300m: You had the conviction there Tim that I wasn’t brave enough for on this forum of clever folk! When posting I was thinking, ‘Hmmm square wave at 250kHz: should I go to at least the 5th or 7th harmonic to consider the ‘max’ frequency to calculate the quarter wavelength?’. 😊

[timsabre]

Well you might be right, it's a square wave so who knows. The fact is I've spent a lot of time working with DMX as a fixture developer and in some weird and wonderful configurations and only ever seen problems when combining a passive split with long cable lengths and another fault like one-legged DMX. I agree with whoever said above that DMX will run over spaghetti. 

Edited February 17, 2022 by timsabre

[sunray]

1 hour ago, kgallen said:

300m: You had the conviction there Tim that I wasn’t brave enough for on this forum of clever folk! When posting I was thinking, ‘Hmmm square wave at 250kHz: should I go to at least the 5th or 7th harmonic to consider the ‘max’ frequency to calculate the quarter wavelength?’. 😊

That's not an unrealistic consideration, however the real world of hanging a few fitting on the line will soon add the mismatches & start adding reactance and the square wave gets rounded quite quickly. Using the 3rd is fair but I don't think going further is likely to be required.

Edited February 17, 2022 by sunray

[sunray]

28 minutes ago, timsabre said:

Well you might be right, it's a square wave so who knows. The fact is I've spent a lot of time working with DMX as a fixture developer and in some weird and wonderful configurations and only ever seen problems when combining a passive split with long cable lengths and another fault like one-legged DMX. I agree with whoever said above that DMX will run over spaghetti. 

Indeed, I've had DMX working on 200m of telephone dropwire and TV 75Ω coax. Sometimes one just has to try if running a cable is tricky and something else exists...

[gareth]

Terminators? Well, on the whole I don't use them. That's to say, when I cable a rig, they don't get put in. But whenever I'm proddy LX on a show or a tour, there are always a few terminators in the spares box just in case. If I encounter any issues which might be DMX-related, the first thing that happens is that a terminator gets put on the end of the line to see if that solves the problem. Most times it doesn't, and the issue lies elsewhere - but it's a really quick thing to try, and it might just work. Not exactly a 100% 'by the book' approach - but it's served me well for the last three decades, so I won't be changing now!

 

Edited February 17, 2022 by gareth

[kgallen]

I’m a bit bemused. Why wouldn’t one just put in a terminator as part of the install in the first place? They cost what, a couple of quid each?

I’m not sure I understand the general belligerence towards just doing the right thing in the first place. When I’m rigging I end with a terminator on each (actively split) branch, there and then. Done once. Why wouldn’t I? One problem I don’t have to deal with. 
 

Slightly rhetorical as I know we’ve discussed this topic ‘several’ times on the Forum over the years!

Edited February 18, 2022 by kgallen

[ianknight]

50 minutes ago, kgallen said:

They cost what, a couple of quid each?

*gives a knowing wink*

To some people that couple of quid is their "much needed profit" - certainly from personal experience when I was an employee that used to be the case :P

[kgallen]

1 hour ago, ianknight said:

*gives a knowing wink*

To some people that couple of quid is their "much needed profit" - certainly from personal experience when I was an employee that used to be the case 😛

Would that be a slightly portly gentleman with a beard? 🤣