Need Advice - Pixel Installation from He11

Good day! I have been struggling with a very temperamental outdoor pixel installation for 2 years now. It's just one problem after another. I would tend to think of myself as a decent engineer and this is really hurting my pride that I can't get it right, not to mention a pissed off girlfriend. I'll go into the full backstory below, but here's the synopsis: There are a total of 32 12-volt bar-style pixels, in pairs, 11 feet apart between each pair. My current problem is that between pixels 11 and 12 (starting at zero) the data stream is getting corrupted. Pixel 11 works great. #12 and #13 flicker and obviously are getting corrupt data. #14-31 don't even come on. If I hook up a string of spare pixels right after # 11, they work exactly as they are supposed to. If I hook up my string of spares just before #12 I get the same corruption problem. Power is fed with a piece of 16 gauge zip cord (landscape wire) and the data was fed along another 16 gauge piece of wire. The wire is run thru 1/2" gray PVC conduit run just under the surface. This worked for over a year (albeit with other problems). Before doing the installation, I ran some experiments to see how far I could run WS281x data along a single piece of regular wire, and in agreement with the WS2811 datasheet, I found I could do about 10 meters or 33 feet. By using twisted pair for the data, I found I could go at least 150 feet. (Reminder, as the signal passes thru each pixel, the WS2811 chip cleans up and reshapes it.) My current theory for my current problem is that I think the underground power cable for the house passes under the pixel wiring at about this point. But I have some holes in this theory. 1) Why did it work for over a year? 2) According to the building/fire code, the power wire should be buried at least 18" deep and thus is at least 18" from my pixel wiring. 3) The power wire is typical American 220 volts, split phase, with a neutral and ground. Since the power is making a round trip thru these wires, shouldn't the magnetic flux cancel itself out? 4) If I run a temporary data wire across the ground (only 1-1/2" higher than the normal one) it works fine. I do NOT have a decent oscilloscope. I DO have one of those really cheap Chinese ones that you assemble from a kit. I admit that I have not yet tried the oscilloscope, but I'm not sure what to look for. I thought I might hook it the data wire, but not transmit any data, and look for 60 cycle interference and take note of how many millivolts or volts it peaks at. I have tried replacing the data wires with twisted pair. (one side data, the other side ground). That did not solve the problem. I've tried connecting the negative of the power supply to earth ground, and tried connecting the positive to ground, and tried leaving them floating. I've tried connecting the other side of the twisted pair to the +12v instead of ground. None of these solved the problem. However [as mentioned above] If I run a piece of plain 16 ga. wire above ground it works. The pixels just before and after the problem have been replaced (numerous times) to eliminate the possibility of it being the pixel itself. I do intend, on my next visit to the installation, to hook up my cheap scope and check for 60 Hz noise, but if my hunch on that is correct, I'm not sure how to resolve it. My next thought was to try a piece of coax (RG-59 Cable TV) for the data wire. Picture of a Light-Up-Paver Picture of Bar Pixels Paver Pixels Wiring Diagram Paver Lights Layout

What have I overlooked? Can I not see the forest for the trees? All advice is appreciated!

The Backstory: Two years ago my girlfriend had her driveway redone with pavers. A big part of her wanting to do that is that she found some light-up pavers on Amazon, and she is even more of a lighting freak than I am. The pavers included 12volt wedge base incandescent bulbs as are typically used in landscape lighting. But my girlfriend lives less than a block from the beach, and the light-up pavers are not completely sealed underneath, and corrosion quickly became a problem. I ended up replacing most of the sockets, and filling them with NOLOX electrical grease before installing the bulb, but the corrosion problem continued. My girlfriend is O.C.D., and if just one light is not working, she freaks out and she can focus on nothing else, and I have to come fix it, pronto! Arrrrrg! I tried replacing the incandescent bulbs with LED bulbs but not only did that not solve the corrosion problem but the LED bulbs themselves corroded and so the problem was even worse. Furthermore, as part of the original installation, each and every paver light was connected to its own wire, all of which went to an Arduino which would flash them in patterns. Alas, I used cheap Chinese Copper-Clad-Aluminum wire. I did not know at the time what a fiasco that was! The wire and its connections kept corroding out, even when soldered and sealed in heat-shrink and silicone. After fixing countless connection problems and replacing half the wires I had to give up. A word of advice: Stay away from Copper-Clad-Aluminum wire!! It is awful stuff! That was a tough lesson learned for me. So in a total reboot, I switched from dumb LEDs to Pixels. I pulled up all of the pavers along both sides of the driveway (including non-lit ones) and replaced the wiring with #16 gauge copper landscape wire (SP-2 Zip Cord) and a 16 ga. data wire, and ran it thru PVC conduit just below the pavers. All connections soldered, heat-shrunk, and further sealed in silicone. That took me 4 days under the hot Florida sun. Built a new Arduino controller accordingly. This, generally speaking, worked, for a while. I programmed all sorts of cool animated patterns and colors into the Arduino for different holidays and when working properly, it looks awesome! (Flickering orange and purple for Halloween, Red-White-and-Blue flag for July 4th, Pink and Red pulsing "heartbeats" for Valentines...) The next problem was that the blue and green portions of the pixels kept failing at a very high rate. I was constantly replacing pixels because one of the LEDs had failed. I dropped the power supply voltage from 12 volts to 10, then to 9. I added 47uf filter/decoupling caps to the power wires at every other paver. Because the WS2811 includes its own constant-current regulating circuitry, this did not really reduce the power to the LEDs and the blue and greens kept failing. (Note: I've found that you can operate 12v pixels on as low as 4v without significant change in the brightness.) Then the latest problem I mentioned above started. I went thru and replaced every single pixel with some newer (and hopefully better) ones and replaced the data wire all the way with cat5. Initially I tried using all 4 pairs, using the solid colors for data and the whites for ground, but that did not work at all! I had to revert to using just one pair. Every connection has been redone with tremendous care to keep them all sealed and waterproof. I changed the Arduino code so that the blue and green never gets more than a 70% duty cycle. But none of that matters when the data won't make it all the way down the line.

Thanks in advance, Dr. Wizard

You assume #11 has not failed. Temporarily wire #10 to #12 so #11 is removed from the system.

ALL of the pixels have recently been replaced, and then more of them have been replaced again and again on both sides of the problem point. I can take a string of spare pixels and connect them anywhere along the line. If I connect them just after #11, or anywhere before, they all work. If I connect them just before #12, or anywhere after, they don't. If I bypass the normal data wire and run a scrap piece of wire across the ground, #12 on works as expected. The normal data wire runs thru a pvc pipe just below the surface of the pavers. I have pulled the data and 12v power wires out, carefully inspected them, and replaced them several times. I also built a spare, portable controller in a small plastic box that sends out several test patterns. If I inject a fresh signal just after #11, then #12 and #13 act wonky, and #14 on do not work. If I inject the test signal just before #12, they work as expected. After countless hours of replacing, rewiring, and running all sorts of tests, I have conclusively proven that the problem exists BETWEEN #11 and 12. But I don't understand why, and so I'm not sure what to try and do to fix it. My only guess is the 220v power cable that feeds the house, which I think runs 18+ inches below the paver wiring, approximately at the point where the corruption starts. And I'm pretty sure the power cable is at least 18" deep, or close to it. It was installed by the power company itself (not an electrical contractor) with a giant trenching machine, and it also has to dive under a drainage ditch nearby, so if anything, it's probably 24-30" deep. There is also a 120v power wire also about 18" away that feeds a post lamp. But I completely disconnected it and for the time being anyway, left it disconnected.

I have a master's degree in electrical engineering, although I earned it 40 years ago and spent my career doing software engineering, not electrical. Still, I consider myself to be a reasonably good electrical engineer and technician. I've done lots of work with pixels (including a gigantic musical Christmas light show at my house using Light-O-Rama) and I've never run into a problem like this. That's why this problem is driving me absolutely nuts!! I wonder if its not a situation where I can't see the forest because of all the trees in the way. Or if I'm somehow overlooking the giant pink elephant in the room. If it was a snake, would it have bit me?

Voltage drop over the cable? After the 11th, the voltage might be too low; use a 'fatter' cable for 12V and GND.

The wire is run thru 1/2" gray PVC conduit run just under the surface.

If you run identical wires along the ground between #11 and #12 does the problem occur?

sterretje: Voltage drop over the cable?

1) I foresaw that and it's why I used such heavy gauge (#16) wire, even though max total current is under an amp. After the initial installation I measured at the first and last pixels and had only about a 0.4 volt drop despite almost 200 ft total of wire. I've since rechecked multiple times with the same results. 2) After having such a high failure rate on the blues and greens, I dropped the power supply voltage from 12 to 10, and then to 9. That had no effect whatsoever on their operation [at the time] or brightness (due to the constant current driver in the WS2811 chip). Thinking that this might be causing the new problem though, I bumped it back up to 12v. And, doing some testing on the bench, I accidentally discovered that they work just fine on as low as 4 volts, and with only a minor reduction in brightness (again, due to the constant-current driver).