Big problem 8000 pixels show

[Natalia]

Hi ! I am Natalia, sorry for my English, I am writing from Spain. I am writing here because I don't find where is the problem with my light show and how to solve it.

This is my first light show, it has 8000 pixels 5v so I got a Kulp k8 , I was testing and that was perfect but right now I am creating the props and putting them in the place where they will be in the show. I will place the kulp and the beaglebone in the center of the show, so more or less all the first props will be at the same distance from its own port. I have connected to the port the first prop (100 pixels) that is to 6 meters (distance) and I have seen that lights didn't work properly, they were shaking, the colors were not as the color selected, there were different colors and it was even worst when I connected the second prop (100 pixels more) and an injection power, the pixels got even more crazy making stranger things.

I realized that the output volts was ok 5v at the end of the 6 meters wire but data output fell to 0.3 (when I verified that when I have a 0.40 centimeters wire the data output from my kulp is 1.08). How is it possible that only data fell and such a big fall in that short distance? I don't understand where is the problem. I see in YouTube layouts with a big distance and only 1 fpp and pixels work correctly. It is impossible to have all the first props in each port to 0.40 centimeters. I consider that 6 meters is not a big distance from the port. Please, I need help, I need to know what I am doing wrong, because I cannot run my show when I have ready my 8000 pixels. Thank you in advance.

Have a nice day.

 

[MikeKrebs]

Data at six meters from a K8 may work provided good wiring. It might be at the edge of its distance. See here for another opinion: https://shop.allnetchina.cn/collections/quinled/products/quinled-data-booster

However, power is a different situation. I do not understand what size wire you are using. .4 cm wire would be huge and should maintain voltage very well. I wonder if you mean .4 mm wire and that is way too small. Go here and do some playing around. http://spikerlights.com/calcpower.aspx It is in American measurements but you can translate the results.

So, a couple of things.
What is the wire type? Pure copper wire is recommended for both data and power.
What size is the wire? Wire must be adequate for power transmission - especially with 5V as it doesn't leave a lot of headroom to lose a volt.
Measuring data with a volt meter is very prone to error as it doesn't tell anything about the transitions that cause the zero and ones to be counted. However a loss of .78v seems a bit excessive and is likely a cause for concern.

A couple of experiments to try.
Attach a V+ and V- at the end of the pixel string running back to the same psu you are using. Does it improve? If so, power injection might be required in your situation.
Use a 3m cable connected to both sides of a single pixel then continuing on to the string of 100. Does that improve the shaking? If so, you might need to implement a "null pixel".
Reduce the intensity of lights to 25%. Does that help? If so, then probably power related. Make sure the lights are getting good power all the way through. Measure voltage at the end of the string. You can see from the Pixel Light Power Calculator when the pixels will likely get flaky.

"Shaking"...I wonder if you unintentionally sending multiple data sources when you were doing your test. Are you using FPP testing page or testing from Vixen/xLights?

 

[Skymaster]

One other consideration is the frame rate with that many ipxels.
With 8000 pixels on 8 ports, you've got more pixels (1000) than a single port can do at 40fps (800).
You may need to reduce the frame rate of your sequences to 20fps.

This wont affect the issues you're seing.

Measuring data with a volt meter is very prone to error as it doesn't tell anything about the transitions that cause the zero and ones to be counted. However a loss of .78v seems a bit excessive and is likely a cause for concern.

I'd hazard a guess that this is a red-herring. The loss is probably not loss; just a difference in data that is being transmitted at a point in time.
A multimeter is next to useless on the data line to tell if anything good/bad is happening.

 

[Natalia]

Hi! Thank you very much for your help. The problem was the wire, I was using not the correct one. I have bought awg 18 just to test with 10 meters and it works perfectly. I was using a current wire and not the correct one.

Thank you again. I am happy to have joined that amazing comunity.

 

[Indigogyre]

Glad you were able to figure it out with proper cable sizing.