Weird crosstalk/int...
 
Notifications
Clear all

Weird crosstalk/interference problem

13 Posts
5 Users
0 Reactions
3,716 Views
Posts: 6
Topic starter
(@tonydoeslights)
Active Member
Joined: 9 years ago
[#46245]

Hi, I've got a pixel tape problem that I can't seem to figure out, and hopefully you all have seen something like it before. I've got about 5000 WS2812B pixels in aluminum channel affixed to two sides of a large steel plate in sort of an ersatz low-res video screen arrangement. The strips are power injected approximately every 150 pixels. I'm running 6 60 amp 5VDC supplies to power it all, with all grounds tied together. I'm using 4 Sandevices E6804 units to run it all from ELM. The problem occurs when I plug a strip into more than one of the pixel controllers. I can plug 4 strings into one controller and things will be solid, but the second I plug a strip into another of the E6804s, the whole thing begins flickering incessantly. I'm trying to diagnose if it's a crosstalk or a grounding problem.

Firstly, Is crosstalk between data lines a big problem with these lights? I have 12 separate data lines running in parallel. The data lines are also tacked to the steel plate. I fear that the 700 pound steel plate that the whole thing is mounted on is acting as a sort of antenna and making the whole thing ring.

I could keep speculating, but I'm hoping one of you recognizes the problem I'm having and can give a little advice. Thanks!


12 Replies
Barnabybear
Posts: 1583
(@barnabybear)
Noble Member
Joined: 15 years ago

Hi, what length are the data lines? Are they cat5 with data and a ground on a twisted pair?


Reply
Posts: 6
Topic starter
(@tonydoeslights)
Active Member
Joined: 9 years ago

The data lines are 18 AWG stranded right now. I'll try cat 5. Would it be possible to run multiple data lines on the same ethernet cable, or would you suggest running multiple cables for multiple lines?

EDIT: Oh, and the data lines are about 20 feet (6m) long. Thanks!


Reply
Barnabybear
Posts: 1583
(@barnabybear)
Noble Member
Joined: 15 years ago
tonydoeslights wrote:The data lines are 18 AWG stranded right now. I'll try cat 5. Would it be possible to run multiple data lines on the same ethernet cable, or would you suggest running multiple cables for multiple lines?

EDIT: Oh, and the data lines are about 20 feet (6m) long. Thanks!

Hi, 20ft is getting towards a long run for data at that distance it will be better on a twisted pair. There are thoughts that thinner solid data cables work better than thicker stranded ones but that all starts getting very technical. As your setup uses 4 output controllers it’s defiantly worth trying running 4 data / ground pairs in one cable.


Reply
JP90
Posts: 376
 JP90
(@jp90)
Reputable Member
Joined: 19 years ago

I had a problem like that with a 5V pixel run of 22 feet and solved it by replaced the 270 ohm resistor packs with 100 ohm packs. Don't go too low because packs protect the driver chips from shorts on the data lines. If you do replace them get the "non polarized (isolated)" types.

Gary


Reply
Posts: 47
(@mtbehnke)
Trusted Member
Joined: 10 years ago
tonydoeslights wrote:The data lines are 18 AWG stranded right now. I'll try cat 5. Would it be possible to run multiple data lines on the same ethernet cable, or would you suggest running multiple cables for multiple lines?

EDIT: Oh, and the data lines are about 20 feet (6m) long. Thanks!

I had the same problem with my matrix. I ran four pairs of 18 gauge data lines together, with separate power/ground runs, and had lots of crosstalk. I could even disconnect one of the data wires and those pixels would still light up based on the adjacent string. The runs ranged from about 8-14 feet.

I replaced the data lines with Cat 6 cable, with one wire of each pair carrying data and the second one in the pair tied to the controller ground, but not connected at the other end.

Since making this change all my crosstalk issues have gone away.

Mike


Reply
Posts: 6
Topic starter
(@tonydoeslights)
Active Member
Joined: 9 years ago
mtbehnke wrote:I had the same problem with my matrix. I ran four pairs of 18 gauge data lines together, with separate power/ground runs, and had lots of crosstalk. I could even disconnect one of the data wires and those pixels would still light up based on the adjacent string. The runs ranged from about 8-14 feet.

I replaced the data lines with Cat 6 cable, with one wire of each pair carrying data and the second one in the pair tied to the controller ground, but not connected at the other end.

Since making this change all my crosstalk issues have gone away.

Mike

Thanks! I'm going to try that, and will update when I do.

Sent from my Nexus 6P using Tapatalk


Reply
Posts: 6
Topic starter
(@tonydoeslights)
Active Member
Joined: 9 years ago
JP90 wrote:I had a problem like that with a 5V pixel run of 22 feet and solved it by replaced the 270 ohm resistor packs with 100 ohm packs. Don't go too low because packs protect the driver chips from shorts on the data lines. If you do replace them get the "non polarized (isolated)" types.

Gary

Apparently the E6804 doesn't have series resistors on the output, and that makes the data run very susceptible to picking up interference. I rewired everything with Cat 6 cable, but the problem was still occuring. I ended up putting a 150 ohm resistor right on the output of the E6804, and that cleared it all up. Thanks for pointing me in the right direction!


Reply
Posts: 70
(@jchancel)
Trusted Member
Joined: 11 years ago

For future reference, what does the resistor do? Why does that fix the issue?

Sent from my ONEPLUS A5000 using Tapatalk


Reply
Barnabybear
Posts: 1583
(@barnabybear)
Noble Member
Joined: 15 years ago
jchancel wrote:For future reference, what does the resistor do? Why does that fix the issue?

Sent from my ONEPLUS A5000 using Tapatalk

Data is just electrons flying down a cable, when they get to the end if their energy isn’t absorbed they will bounce back and fly the other way back up the cable. If they get to back to the start end and their energy still hasn’t been absorbed they will bounce back down the cable and do it all again – a big problem as we now have two lots of data heading down the cable, a new one and an old one. You can imagen this can get out of hand. The answer is to absorb their energy. The cable naturally does this as it has resistance but we try to keep the resistance as low as possible so we can transmit data over ever longer distances. Yes we create or own problems. So the trick is to introduce just enough resistance to stop the bouncing (called ringing) without reducing the distance the data can travel. This is why you hear of people swapping resistor packs on their controllers to get better distance. In an ideal world we would have about 50 ohms at each end of the cable to provide a little absorption at each end. In reality we end up with just resisters at the controller end, and big ones 220 ohms. There is a very good reason for this it stops the controller output being blown up if you, squirrels or water accidently short the data line to ground or 12V (data is only 5v). You can imagen the manufacturers of $200 controllers don’t need bad reputations because of unfortunate events and it works as they don’t have. Even if you could find a way to get rid of ringing and the risk of blowing up the outputs of your controller without using resistors you would still run into the effects of noise / interference. This is because any length of cable will act as an antenna and pick up small electrical signals from the world we live in; whilst these are mostly quite minor in the scale of things without any resistance to absorb them they too would start to add together and get out of hand.

So, the lower the resistance the better distance you can get but there is a minimum your require dependant on the type of cable you use and the world you live in.

The E6804 is a more commercial product than most generally used. I didn’t realise it doesn’t have output resistors, it must the responsibility of the installer to fit them appropriate to the application.

Hope that helps.


Reply
Page 1 / 2
Share: