[norty303]

Ok, so the moment has passed now, but am curious in any debate about what could have been the issue.
We were doing a small festival stage a week or 2 back.

Rear truss had some Robe scans, Atomics, Sunstrips and LED pars.
Just about halfway down the truss, one of the Robes had the display flashing which indicates no DMX signal being received. A number of fixtures downstream (Atomic, a couple of Sunstrips and a couple scans) were dead or showing no DMX indicator, and then beyond them, everything working ok.
We swapped out cables, added terminators, etc with no effect, and finally fixed it by running a separate line from the DMX buffer to the first Robe that was showing the problem.

How can the fixture pass thru data, but not 'see' it? We did think it was an impedance issue, whereby the DMX receiver wasn't getting it, but as the in/out is parallel wired it still went on down the chain for 'some' fixtures to respond to.

We 'think' it may have been an Atomic or a Sunstrip putting out some bad data, but no idea why it only affected some fixtures and not others.

[Gerry]

norty303, on 28 May 2012 - 05:16 PM, said:

How can the fixture pass thru data, but not 'see' it?

It is my understanding that it is common practice for the DMX in and out to be connected together.
Hence the fixture might not "see" the DMX signal but it is passed on to the next fixture.
Cheers
Gerry

[LXbydesign]

Yeah, so the questions then are - WHY didnt the fixture 'see' the data and why were only a few affected before 'normal service resumed'

Not that it probably would of made any difference but thats one reasaon why I prefer to have bits like LED cans and Sunstrips etc etc on a seperate line.

Ive had issues in the past with LED Batterns doing funny things to DMX lines - so have always run those on a seperate line. Can be a bit annoying sometimes, but seems to solve any hicups.

[Brian]

Almost impossible to diagnose after the fact.

My first suspicion would have been lack of terminator but you say you tried that. After that you're into faulty cables and faulty kit. Maybe even two separate faults which mask each other.

[mrboo]

I tend to take out the DMX to the light before the unit not working and if this dont sort it the first unit that works then you know if it the lights or cables. I always like to have a really long DMX lead to test if it cableing by running just once lead from desk to every light. on there own.

[david.elsbury]

Atomics have a bit of a reputation of corrupting the DMX data, though I understand this is more when they are fired rather than all the time.

Likewise, LED Pars are notorious for being picky about their DMX (generally due to being cheap and cheerful)

I always run Atomics on their own, isolated and buffered data line. I try to do the same with LED pars, but there comes a point where you have a dozen fixtures on a truss of 3 or 4 types... it's a bit silly to run 4 data lines out to the truss IMO. (or put a splitter on the truss)

It could have been a bad cable also, but as Brian says, we're all just speculating.

David

[Don Allen]

We know from Brian's maths "As long as the stub's electrical length is less than one quarter of the signal's transition time then the stub will still behave as a lumped load and not a separate transmission line.

For a typical DMX driver (75176B) that means you can have stubs of up to 1m in length. You will however be sitting on the digital cliff edge. A safer figure is a length of 10% so stubs around 400mm will be OK."

How far apart were the fixtures ? is it possible that something downstream of the non responding Robe was reflecting a null upstream ?

The other possibility is a double reversal of DMX cables, I.e. one DMX cable with pins 2 and 3 are reversed will stop correct DMX receival for a few fixtures, then a second DMX cable with pins 2 and 3 are reversed will allow correct DMX receival for all the rest of the fixtures. I did this to myself once when in a hurry and having to fit 5 pin DMX connectors to led pars.

[mrboo]

Yes good point about a double reversal I have had this in a venue after swaping out a lead I ended up opening up the old one to find that it been reversed at one end of the long run, and also at the desk, Think it was done by a hire frim to stop people messing with the rig.

[Modge]

Don Allen, on 29 May 2012 - 12:04 AM, said:

How far apart were the fixtures ? is it possible that something downstream of the non responding Robe was reflecting a null upstream ?

Had that before. Solved by substituting cables until we found the offender, same as you would normally, but downstream of the fixture lacking DMX.

The other time I experienced this it was down to lack of termination. So you're not alone, it does happen.

[Brian]

I started to make a list of all the things that can catch you out with DMX and gave up at item 20.

It's a huge list and sadly contains many unquantifiable items caused by manufacturers not sticking to the spec.

When done properly DMX is very reliable. It'll work day in, day out with no oddities and not a flashing 7-segment display in sight.

One day I will sit down and write the definitive guide to real-world DMX.

[norty303]

Interesting thoughts especially on the Atomic data corruption thing.
We swapped cables out, but quickly resolved it was not a cable issue. Fixtures were close to each other, connected using proper 110ohm DMX cable, mainly 1m and 1.5m lengths. No reverses needed.

Quote

I tend to take out the DMX to the light before the unit not working and if this dont sort it the first unit that works then you know if it the lights or cables. I always like to have a really long DMX lead to test if it cableing by running just once lead from desk to every light. on there own.

Yes, we did this, bypassing the fixture that appeared to be the source of the problems, but the issue manifested itself again.

One item that may be significant, when we added the terminator, we got even more strange results - it was better without. What this might indicate I don't know...?

[Brian]

norty303, on 29 May 2012 - 11:59 AM, said:

One item that may be significant, when we added the terminator, we got even more strange results - it was better without. What this might indicate I don't know...?

What was the desk?

[timsabre]

norty303, on 29 May 2012 - 11:59 AM, said:

One item that may be significant, when we added the terminator, we got even more strange results - it was better without. What this might indicate I don't know...?

In my experience this always means that one leg of the DMX line is broken somewhere.
DMX will often work, unreliably, with either the "hot" or "cold" leg broken, if no terminator is used. But it does weird stuff as it relies on the broken wire floating above ground.
Once you plug in a terminator, both hot and cold legs become identical and then it doesn't work at all.

[Richard CSL]

I have had a similar problem in the past and it was caused by a solder problem shorting pins 2 and earth 3, it only corrupted the DMX for the following 3 or 4 fittings , then everything after that worked ok.

Took ages to find the fault, and of course it was the most inaccessable fittings in the venue, it was also intermittent as the building heated up, the wires touched each other and as it cooled all was ok again, the DMX lead had a long amount of exposed copper on all conductors in the XLR plug, and this was the source of the problem, it had all worked for 2 years prior to me arriving on site. Tricky stuff DMX and it drives you mad at times.

[norty303]

Quote

What was the desk?

Chamsys MQ100 (I think it was) but running ethernet to back of house where a Chamsys ArtNet DMX interface was situated.

Interesting about the hot/cold legs being broken, I'll have to get my cable checker out and test all the cables in use on that truss