Crosstalk between RS485 Networks

[brianfox]

I am attempting to do multiple RS485 LOR networks this year. I have learned my lesson in past years about having the comm lines too close to power.

What about distance between comm lines on different networks? In comparison to power lines, how serious of a problem could this be?

I am using solid core unshielded CAT5E, and will have 7 networks from a central USB hub to the networks of boxes. Since the PC will be in my garage, all of these network lines need to feed to the outside. Ideally, I'd like to bundle them, but realistically, I plan on routing them along the wall or floor with about 2" spacing between them. The cable runs will be around 100' each from CPU to first box.

If you guys think they should be no less than a foot apart or by no means run in parallel, I'm going to have a problem...

[-klb-]

I don't think you will have any issues..

The data is much lower voltage and current than the power cords that induce problems in the com cables

[gizmomkr]

7 networks - how many controllers do you have ?

[brianfox]

gizmomkr wrote:

7 networks - how many controllers do you have ?

Last year I had 42 boxes and a neverending series of problems. The worst part was that when I saw wonky channels on box X, often times it was because of a cable between boxes A and B.

This year, I will have no more than 5 boxes on any network; if there's a problem it will be isolated to that network and I should be able to quickly fix it.

Yeah, it's overkill, but I was ready to box it all up and sell it after last year. Instead of buying $500 worth of additional boxes this year, I spent $200 on extra RS485 converters.

Now if this doesn't work, I'm definitely going to either sell it or get a gallon of gas and make the best 30-minute light show ever.

[Max-Paul]

Brian,

Step back from the LOR computer!! Easy my friend, First I have to ask, you have 7 USB ports on a computer? I saw something about a USB hub. Dont fully understand how that comes into play.

But I have to agree with klb's statement. Both the voltage and CURRENT are much lower than your A.C. lines. Thus the magnetic line will be fewer and weaker. Always a chance of cross talk. But note the twist of the wires in a cat5 cable. They are there to help reduce the possibility of cross talk. And the pairs will spin around the axis. Again this is to minimize the chances of cross talk.

[brianfox]

Max-Paul wrote:

Brian,

Step back from the LOR computer!! Easy my friend, First I have to ask, you have 7 USB ports on a computer? I saw something about a USB hub. Dont fully understand how that comes into play.

I have a 10-port (powered) USB 2.0 hub that plugs into one of my laptop's ports. It's giving me the 7 USB ports I need for the networks. All you do is hook up the RS485's one by one and they are assigned unique COM ports. In this case, my converters are assigned COM5-COM11. Each of these adapters will run an LOR network.

I did a test plugging all of the adapters into the hub with a single LOR box on each network, and I hooked up a channel of lights to each box. They all happily blinked away.

This is the same sort of thing mentioned in another thread (but that guy was using USB extenders, and I'm not):
http://forums.lightorama.com/view_topic.php?id=29031&forum_id=76&highlight=hub
You made a comment last week that his hub should have no problem with 5 networks.

[KStatefan]

Are you using line terminations? I do not see it mentioned here very often.

[-klb-]

KStatefan wrote:

Are you using line terminations? I do not see it mentioned here very often.

I believe I have also seen Dan post that in early testing, they saw more issues in tests with terminators, than without.

I believe he has also stated that in some cases they do help, and may be worth trying. But I think bad cables, or dirty connectors are a far more common source of issues. I normally have one network that is likely over 1,000 feet and 40 controllers, with zero issues, that grew to that size over 4 years. I'm dividing it up this year, not because I need to, or because I expect issues, but because I am adverse to loosing that much of the show, if we have our fist com issue.

[Max-Paul]

True, but I was trying to address the question if there was enough bandwidth. But I lost track and my answer was not right. I was going to say that it would NOT starve for data due to the LOR network operates at a much lower data rate.

I'll admit, I would not have thought that you could have multi adapters on a single USB port. I have to admit, that I live and learn something new in this respect.

Thanks

brianfox wrote:

Max-Paul wrote:

Brian,

Step back from the LOR computer!! Easy my friend, First I have to ask, you have 7 USB ports on a computer? I saw something about a USB hub. Dont fully understand how that comes into play.

I have a 10-port (powered) USB 2.0 hub that plugs into one of my laptop's ports. It's giving me the 7 USB ports I need for the networks. All you do is hook up the RS485's one by one and they are assigned unique COM ports. In this case, my converters are assigned COM5-COM11. Each of these adapters will run an LOR network.

I did a test plugging all of the adapters into the hub with a single LOR box on each network, and I hooked up a channel of lights to each box. They all happily blinked away.

This is the same sort of thing mentioned in another thread (but that guy was using USB extenders, and I'm not):
http://forums.lightorama.com/view_topic.php?id=29031&forum_id=76&highlight=hub
You made a comment last week that his hub should have no problem with 5 networks.

[Steven]

Cat5 cables are designed to run together in a conduit. Cat5e cables can easily handle 1000BaseT while touching each other, so you will have no problems with the LOR network, which is a much slower data rate.

[brianfox]

Steven wrote:

Cat5 cables are designed to run together in a conduit.  Cat5e cables can easily handle 1000BaseT while touching each other, so you will have no problems with the LOR network, which is a much slower data rate.

I hear you, Steven, but then why are we supposed to not coil excess Comm cable because it tends to act like an antenna, picking up noise?

[Steven]

brianfox wrote:

but then why are we supposed to not coil excess Comm cable because it tends to act like an antenna, picking up noise?

Good question! I assume because such cable is designed to run mainly in straight lines and gentle curves, but I don't know the whole answer.

[Greg Young]

brianfox wrote:

Steven wrote:

Cat5 cables are designed to run together in a conduit. Cat5e cables can easily handle 1000BaseT while touching each other, so you will have no problems with the LOR network, which is a much slower data rate.

I hear you, Steven, but then why are we supposed to not coil excess Comm cable because it tends to act like an antenna, picking up noise?

Actually coiling it, depending on the number of turns, acts like a common mode choke, and will serve to decrease signal radiation/pickup. This however really applies to nontwisted pairs, hence there is little impact on today's CT5 twisted pair cables, so it should not be an issue either way.

I prefer to keep things neat, and cut back any extra length, putting a new connector on, and then using the extra length as a patch cable by putting an connector on it as well!

[Greg Young]

I should mention if you are worried about nosie, odd effects (phantom signals), etc. your best insurance is to simply terminate the last controllers in each line with a 120 ohm resistor.

Since I use ELLs, I wound up with 3 terminators for my display. I believe it was Steven that came out with a simple how-to use 1/8 watt 120 ohm resistors, installed directly in the CT 5 cable.

Greg

[brianfox]

Just a quick update:
The multiple network approach really works nice. I have my RS485 boxes strapped together in a block, and the comm cables bundled together until they get outside of the garage.
All of my comm runs to the first controller are between 10 and 150ft. I have all 7 of the networks set to "shorter" in Network Preferences, and have zero sync issues between networks in the show. One of the 7 networks is solely doing DMX; the other 6 are all LOR.
It's like the good old days when I only had 5 boxes - only now I have 7 networks with up to 5 boxes each.
This is a heck of a lot better than trying to use a binary search to find a bad cable among a daisy chain of 42 boxes.

[jeffl]

Is this sticking issue mainly a LOR issue because of how the protocol works? I'm starting to notice this happening in my show again this year. Last year I split my 30 controllers in two networks and I still had issues. In the end for me it was a USB issue. This year I thought I had it fixed with a different machine but now I'm not sure.

I'm questioning if the DMX protocol would be less of an issue.

[jeffl]

I forgot to mention another experience I had this year. I purchased a 100' cat5 cable off eBay. Ran the cable one day and a rabbit chewed it up the next.

When I cut the cable to fix it I noticed none of the wires were twisted. I think for me this was another lesson in you get what you pay for.

[-klb-]

I would think there would be a bit of a toss up. Other factors probably make bigger differences.

DMX runs at 4 times the normal LOR speed. Higher speed may make it more susceptible to interference, and noise.

LOR has intelligence in the controllers, that DMX does not. All a DMX device can do is display the intensity it was last sent, and ever controller gets its value sent many times a second, up to 44 times a second. LOR commands have a duration. It gets a command to fade from 10% to 100% over 15 seconds, and it will do that without any further updates. You tell it to turn on, and it will stay on, until it gets a command to do something else.

So, a brief com error in DMX might mess up more fixtures, but the next update is the next frame, and much faster. I'm not really sure that either one is going to be more reliable that the other.

Remember, my estimated 1,000 foot long run, with 40 controllers is under a 500KV transmission line. I can't even use a non contact voltage probe in the area, because it is always going off because of the lines over head. Yet I have not seen any sort of issues that go back to communication.

Maybe part of my good luck so far, is that I treat my cat 5 cables as the most valuable component in the show. I don't let the ends lay on the ground. They are stored separate from all the other components. I pay attention to not kinking them. I have a tester that identifies not just correct pinout, but also that 1-2, 3-6, 4-5, & 7-8 are all pairs. I use high quality cable ends for all of the cables I make. I think for either protocol, paying attention to detail, and proper care of your data cables is important to it all working right.

[-klb-]

jeffl wrote:

I forgot to mention another experience I had this year. I purchased a 100' cat5 cable off eBay. Ran the cable one day and a rabbit chewed it up the next.

When I cut the cable to fix it I noticed none of the wires were twisted. I think for me this was another lesson in you get what you pay for.

I'm only using stranded for my patch cables, so all the batches below are stranded for head to head comparison. The box I purchased for the show about three years ago, I really liked and it was about $100 locally for a 1,000 foot box. This year, I ordered a 1,000 foot box for about the same price, and I am really disappointed with it. I can't imagine what I would think of the stuff that was $60 so many places. Of course I may be a bit spoiled too. The last batch of cat 5e I used at work was $350 per 1,000 foot spool. But you put cat6 ends on it, and it would nearly always certify as a cat6 patch cord for any length under about 50 feet.

[jeffl]

Any recommendations on bulk cable to purchase? I do remember the green cables from LOR were much thicker than any cables I have purchased since. My guess is the quality of some of my cables might be poor but I don't know.

I know how to make cables if I could get known quality cable and ends. Or are there patch cables that are known high quality for a reasonable price?