Jump to content

Artnet drop out


numberwrong

Recommended Posts

Had some Artnet dropout problems the other day. The rig froze for a second, once every hour or so (sometimes more)

 

When it restored movement effects snapped to where they would have been if it never froze (opposed to stopping then starting again) suggesting data dropout not the desk software freezing, although it could be either.

 

Desk was an Avo Quartz on titan V10, node was a Chauvet net X.

 

I was only running 5 universes

 

It was a fairly small job so I originally decided to do it on sACN just for the hell of it.

 

With sACN I had issues where rasing some intensity cues for moving lights was causing the dimmers to dip for a second. Dimmer was a power cube running DMX straight from the node.

 

I switched to broadcast Artnet and nothing worked so I started a new show from scratch. Broadcast Artnet didn't work at all but unicast artnet (patching each output) worked but with the dropout issues mentioned.

 

The original show file was imported and messed around with from an old show.

 

When I started a new show I imported the fixtures only from the fist version of the show, maybe I shouldn't have done this.

 

Any idea what this could have been?

Link to comment
Share on other sites

Not necessarily the same thing, but I had problems a few years ago when using a Chamsys node (I can't remember which version- it was a 4-port jobbie that could be either a node or a buffer) with an EOS system. I can't remember if we ever exactly got to the bottom of it, but I'm pretty sure when things were re-routed around the node, I recall the problem went away.
Link to comment
Share on other sites

The network path from desk to node was set up was like this:

 

Desk > 100m CAT6 shielded cable > ethercon coupler > standard non shielded CAT 5 patch cable > 5 port Netgear switch (the blue ones) > from the switch to the node using a cat 5 patch cable.

 

I swapped over to a spare line on the multicore and used a spare Chauvet node but the problem persisted. I tried new patch cables.

 

The only things that were not changed were: the Desk, the Switch, the Ethercon coupler.

 

As the switch is a 12v appliance it was not earthed. I did not have any shielded patch leads to maintain the shielding all the way to the node but the shielding would have been grounded the desk end. power for FoH was coming from the racks on the same power supply.

Link to comment
Share on other sites

The network path from desk to node was set up was like this:

 

Desk > 100m CAT6 shielded cable > ethercon coupler > standard non shielded CAT 5 patch cable > 5 port Netgear switch (the blue ones) > from the switch to the node using a cat 5 patch cable.

 

Too much cable. 100Mbit and Gigabit Ethernet is designed around a maximum segment length of 100m. Your first leg is > 100m long (by how much depends on the patch cable) and so you are over. (If the Netgear for any reason is a hub not a switch you are even more over, but that's unlikely).

 

The 100m length limit is not an "analogue" limit to do with signal loss / attenuation (which are always fuzzy limits and can be exceeded with good quality cabling), but based on timing and propagation delays in the cable. Being a digital system, once you exceed the system error budget, it stops working, in a very binary way. Basically, with less than 100m of cable both ends never transmit at once, because there is a defined time wait on starting to transmit. If you exceed the time budget then both ends start, thinking the line is clear, then discover the other had started, but the data just hasn't made it to their end yet due to the excess delay. My suspicion is that each time you saw the rig freeze, you would also have seen the link light on the first leg go out, then the two devices re-negotiate link speed and bring the link back up, so that packets could start to flow again. With only two devices running one-way data collisions will be rare - which is probably why it worked at all!

 

If you are using a 100m cable, it needs to go into equipment both ends directly. More practically, you want less than 100m in one piece to allow for patch leads at the ends. You can use two ports of a switch (not a hub!) as a repeater to re-start the distance clock counting.

Link to comment
Share on other sites

I would try just plugging desk straight into node, unless you need the extra connections for another device. As well as helping with length as described above, that will also remove a lot of connectors and potential for the switch to do weird stuff which artnet seems to cause sometimes.
Link to comment
Share on other sites

If you are using a 100m cable, it needs to go into equipment both ends directly. More practically, you want less than 100m in one piece to allow for patch leads at the ends. You can use two ports of a switch (not a hub!) as a repeater to re-start the distance clock counting.

 

I understand the standard suggests 90m with 'some patching', In this case the patch cable was less then 0.5m before going in to a switch which as mentioned should buffer the signal so I was only a little over. Most (all?) of the long multicores I've ever used have been 100m and I've never had a problem. I always assumed that this was one of those things that in practise doesn't matter that much (a bit like DMX terminating). This is confirmed by my experiences and the fact that the companies I work for stock and use 100m cables.

 

In this particular case I was working for a company that is prominently an audio company (they got the lights as well on this stage). The multi I was using was theirs and spece'd and designed for audio networking (they mainly use Midas Pro series desks) so maybe there was something odd about that cable? It was the Van damme tour CAT 6 stuff. The Audio guys obviously use 100m cables with no problems for their networks.

 

PS the Switch was definitely a dumb basic switch not a hub.

 

 

I would try just plugging desk straight into node, unless you need the extra connections for another device. As well as helping with length as described above, that will also remove a lot of connectors and potential for the switch to do weird stuff which artnet seems to cause sometimes.

 

Yes I should have tried this In hindsight this would have been a good shout. , The show was running and I didn't have much time to mess about as the racks were all patched up internally and it was easy for me to access the switch on the front. There were 2x Chauvet Net X's in the rack and I tried both (which had their own cables running from the switch) as mentioned I swapped the only patch cable which was effectively converting ethercon to standard RJ45. I suppose the port on the switch or the ethercon coupler could have been dodgy.

 

The network status on the desk showed a ping of 0ms and the event log was not logging the quantity of errors (freezing) I was seeing

Link to comment
Share on other sites

Just my 2p's worth, I would be slightly sceptical about the length explanation as I often use a setup like this with 100m cat5 (Van Damme tourgrade or whatever it's called) and I haven't experienced this problem.

Normally we go straight from the Avo desk into the first node, the nodes have link out on the ethernet so if there is more than one node I just link them like that. Also sometimes we use a short patch lead and ethercon coupler at the desk end as some of the Avo desks just have normal ethernet connectors not ethercons, again seems to work fine.

 

I would guess most likely it was a cabling or connector problem. Bummer trying to find it in a live show situation though.

 

Link to comment
Share on other sites

Another afterthought, but you could have run artnet into the power cube, and used the 2 universes you get out of that as an a/b test to see if it was likely to be the nodes, or the desk. If the universes coming from the power cube didn’t suffer the glitch then it implies the nodes were at fault, if both the universes from the nodes and the powercube glitch then look at the desk / cable infrastructure. Either way a laptop with luminet monitor / artnetnominator / DMX workshop / your software of choice is highly recommend to be sat on the network at dimmers to give you some idea of what is going on with the network.
Link to comment
Share on other sites

No answer for you, but I also own the netX and have this exact problem with a TT2. (I also have a problem with Output 1 that stops outputting after some time which I believe it is a heat issue)

 

So maybe just a budget node issue.

Link to comment
Share on other sites

Another afterthought, but you could have run artnet into the power cube, and used the 2 universes you get out of that as an a/b test to see if it was likely to be the nodes, or the desk. If the universes coming from the power cube didn't suffer the glitch then it implies the nodes were at fault, if both the universes from the nodes and the powercube glitch then look at the desk / cable infrastructure. Either way a laptop with luminet monitor / artnetnominator / DMX workshop / your software of choice is highly recommend to be sat on the network at dimmers to give you some idea of what is going on with the network.

 

I thought about this however I'm pretty sure you can only get 1x universe out of a Powercube via artnet (the same universe as the dimming), so I'd have to change the addressing on the dimmers and the desk which was beyond what I wanted to do live given the problem wasn't that bad. Good point about using a PC to monitor the network, I've always used the desk to check the Ping and tell me all is good but a live monitor might be handy in future.

 

No answer for you, but I also own the netX and have this exact problem with a TT2. (I also have a problem with Output 1 that stops outputting after some time which I believe it is a heat issue)

 

So maybe just a budget node issue.

 

I've actually not heard of any issues with the Net X's until this and I know quite a few people using them. I had the same problem with both the Net X's in the rack so I'm pretty sure it wasn't them.

 

I've come to the conclusion that my issue was probably a network problem, maybe a connector in the switch, or even the desk itself. The mulitcore was a muddy festival veteran so it could have added just the right about of mud-dust in to the mix for a connection to work but not all the time.

 

 

Link to comment
Share on other sites

Powercube gives you two universes, the first being the one with the dimmers on (A output on back), the second being the one after that (B output on back)

 

I’ve had a few issues with the net-x’s, on both chamsys and grandMa’s, enough that I wouldn’t put my money into them. I stock luminex, chamsys and DMX king nodes.

Link to comment
Share on other sites

I got no answers at all, but just saying that I have used that Chauvet box several times (albeit with ChamSys) and it seemed rock solid.I agree with Richard about cable length though. On the MA training course we were told that they spec maximum of 75 metres for data runs.

Of course you can use more and many people do, and it usually works. I think the point is that things work until they dont, and to guarantee stable performance stick to the guidelines.

Link to comment
Share on other sites

I got no answers at all, but just saying that I have used that Chauvet box several times (albeit with ChamSys) and it seemed rock solid.I agree with Richard about cable length though. On the MA training course we were told that they spec maximum of 75 metres for data runs.

Of course you can use more and many people do, and it usually works. I think the point is that things work until they dont, and to guarantee stable performance stick to the guidelines.

 

I agree that the system should be designed within the limits of the protocols its carrying. Comparing this misdemeanour with DMX, you can easily add a DMX terminator if you've got glitches but you cant easily make your multi 25m shorter.

It is however a fact that the industry regularly use 100m multicores which until now I have had no issue with (although it remain unknown whether it was the length or not)

 

Should we be pushing our clients to switch to fibre on the long multi runs? I can see this as a hard sell for many companies.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.