RGB Megatree tshooting

[noelb]

I built a megatree last year using WS2801 pixels. I could never get it to work, so I shut it off for the season ( I was busy with family issues, so I just left it standing in the yard during the show). This year I really want to get it to work. But once again due to family, I am slow getting around to it, so I started yet again tonight.

It is made up of :
==1 ECG-P12R and 2 ECG-P12S controllers
==RayWu's WS2801 pixels (87 strands of 50) I am using 3 strands of 50 for a string - power injection every 50 pixels. But I was trying to use Nutcracker last year, so pixel 50 at the top of the tree connects to pixel 51 of the next strand at the bottom of the tree - I have one null pixel in the middle of the data wire coming down to connect to the next strand.
==Power supplies are ATX computer supplies pirated from recycled machines at work (we standardize on Dell, so there are only 2 models of pwr supply involved - I get them all for free)
==Connections are Ray Wu's 2pin connectors
==Wiring is 4 wire security wire from home depot/lowes (suggested in a thread last year). This wire seems like a really small gauge, but I was told it would work, so it is what I went with.

When running a sequence, I get mostly nothing - 95%+ are dark pixels. But there is an occasional flickering pixel in various colors. Sounds like data issues so far?

When I disconnect everything and run only the first controller (ECG-P12R) with the first 3 strands (150 pixels total) in test mode, I get mostly red pixels on, but about 30% are dark.

If I run only the first strand (50 pixels) in test mode, almost all are on as red. But the first pixel is white, and the next 5 or so are dark. Then 90% of the rest of the strand is red (still a few dark ones, but not nearly as many).

The length of the data/power cables from the controller/power supply to the first pixel is just shy of 12 feet (due to the way mega tree was built, with all controllers at the base of the tree pole).

Does this sound like data corruption to you? Is that security wire just too small for the data and/or power signals? I really don't want to re-solder 87 strands...

Since I get red pixels even at the end of the 150 string, I don't think I have power issues (did not test white to verify). This is why I am leaning to data crosstalk, or just too small wire for data to go 12 feet to the first pixel.

Looking for suggestions to tshoot and narrow this down...

Thanks in advance.
Noel

 

[David_AVD]

I think you need to measure voltages at the start and end of each pixel string (run).

Did you ever try a single string of 50 pixels on a controller and power supply by itself? Did it work correctly?

 

[mborg10]

I had something similar last year and was pulling out my hair trying to figure it out. Turns out the cable i was using wasn't thick enough.

It does sound like you have a lot of wiring. I'm not sure how to setup pixels in nutcracker but LSP you can set it up one up one down in a zig zag which would eliminate all those null pixels.

I have heard that recycled computer power supplies have been more trouble than they are worth.

Try plugging a string directly into the controller without any cabling an if it works correctly in test mode I'd almost guarantee you cables not thick enough.

I really hope you get it working cause it sounds like an awesome tree.

 

[noelb]

Did you ever try a single string of 50 pixels on a controller and power supply by itself? Did it work correctly?

Yes, I did (see orig post). It has similar but not exact same behavior. It gets better response from the pixels (more come on) but it is still not right.

I had something similar last year and was pulling out my hair trying to figure it out. Turns out the cable i was using wasn't thick enough.

"thick enough" - you mean too small gauge? This "security wire" is pretty small. I would have to guess it might be 20ga just eyeballing it...maybe even 22ga? It has been over a year since I soldered it all together (last summer), but if I remember correctly, the wires in the "security cable" are smaller than that in Ray Wu's connectors, and also smaller than that in Ray Wu's pixel strings as well. (not much, but enough to notice the difference)

Try plugging a string directly into the controller without any cabling an if it works correctly in test mode I'd almost guarantee you cables not thick enough.

When I tried just the single 50 pixel string last night, it was powered directly by the ECG-P12R power. But the P12R is still being powered by the atx supply. And it still had the long lead on it too.

My boss was pretty generous. I probably have close to 30 power supplies. Of course I only need a few for the tree so I now have plenty of spares. But I get similar behavior regardless of which power supply I use. It is possible there is something wrong with the Dell design, but I don't think there is a problem with an individual supply since changing the supply yields the same type of behavior.

I am debating soldering a new, heavier gauge, shorter lead to one of the strings of 50 and testing again.

When I tested a few of the strings from Ray Wu last year, I had not soldered any leads/plugs on to them yet. It was just the 6 inch wire coming off the first pixel connected directly into the sockets on the P12R. I had also used one of these same power supplies (not sure if it was the exact one from last night's testing though). At that time, the P12R test sequence worked perfectly. This is another reason why I think it is a wiring problem.

I really hope you get it working cause it sounds like an awesome tree.

Me too! Thanks!
Noel

 

[ԆцряєсϮ]

You need to go back to basics, as David said, start measuring voltages.
Start with 1 strand, 1 controller, if you have enough voltage and the right data speeds set and its still not right then you need to start swapping strings or pixels till you get one working, then slowly add from there.
Baby steps.....

 

[davrus]

If your wire is too thin, the most crucial wire is the common one. That one has to be thick enough to satisfy both the positive used to power the pixels, as well as the return for the data signal. You said that you used 4 core cable, but 2811 only uses 3 wires, so what did you do with the 4th one ? If it is still there, try using it to double up the common (ground) wire. You might see an improvement.

 

[noelb]

If your wire is too thin, the most crucial wire is the common one. That one has to be thick enough to satisfy both the positive used to power the pixels, as well as the return for the data signal. You said that you used 4 core cable, but 2811 only uses 3 wires, so what did you do with the 4th one ? If it is still there, try using it to double up the common (ground) wire. You might see an improvement.

Thanks Dave,

I am actually using 2801's - they are 4 wire. Similar to this:
http://www.aliexpress.com/store/product/12mm-WS2801-pixel-module-IP66-DC5V-input-full-color/701799_449802558.htm

But - I think when I get home tonight I will be putting an extra wire on common to see if it is a size issue, and/or shortening the lead for testing purposes (hard to be much shorter once up in the yard tho - so that alone may dictate a larger wire size)

Thanks,
Noel

 

[mborg10]

I have had the security cable running about 4m to the nearest pixel and it seemed thick enough. Although I was using ws2811 pixels and I doubled the common wire.

Just a stab in left field but when I ran the test pattern in the J1sys setup if I had LSP connected it would give me really strange results as well.

 

[mborg10]

Check out this link on power supplies

PC Power supplies, more pain then they're worth http://auschristmaslighting.com/forums/index.php?topic=5680

 

[David_AVD]

4 wire pixels are normally pretty easy to get going, so there's something silly going on here.

What speed did you configure the 2801 outputs to run at? You did configure the controller right? Also make sure you haven't got the data and clock wires mixed up.

When building large items like this, you really need to get a small section working before building the lot. Testing a single string on a single controller output first is essential, otherwise you end up with a large system that's difficult to troubleshoot and may have to be rewired.