Finding that bad Pixel

[Little_b]

Wondering what your process is to find the "bad pixel."  I have 4 matrix panels that I re-wired this past week.  I had issues all winter of flickering, so I decided to do a little more power injection and regrouping of channel counts to eliminate some wire lengths.  3 out of 4 panels work just fine. The 1 panel flickers throughout the panel of 4 outputs.  The 2 outputs more than the others.  The problem is that it is the entire strand and not just part of it, so I can't just find where the flickering starts.  I'm guessing it's a bad pixel since the panels are all wired the same and I used the same 4 outputs on each one when I tested them.  I hope this makes sense.  To confuse you even more when I unplug the really bad flickering strand it causes the others to get worse, and the strand I unhook starts to do the exact same thing as the output next to it.  It should just not have anything going on with it.

[k6ccc]

The normal failure mode of pixels (at least the WS2811s failures I've had) is a portion of a string is either dead or solid white. The bad pixel is either the last one lit or the first one that's wrong. If the whole string is flickering, you likely have something else going on. Power would be my first guess.

What kind of controller and pixels?

[EmmienLightFan]

It is probably either grounding or a bad controller.

Are you using a BBB and octoscrolla, or another, more DIY method of controlling the panels?

[Little_b]

I am using WS2811 12v pixels running off a Pixcon16 controller.  The reason I believe it's not the controller is that I have hooked each panel individually to the same power supplies and output ports on the controller when testing.  Only the one panel flickers.  So if it is a possible grounding problem, could there be a short somewhere in the strand connection points?  Meaning, could there be a cross connection where I have soldiered them together on accident?

[Darryl Lambert]

Many places to check.

When you hook up the flickering panel to a known working port does it still flicker?

If it does I would start with the looking to one of those strings.

Double check that the data direction is correct on the strings. Easy thing to overlook.

Verify the correct wiring on the pixels, That the +, -, & data are right. It wouldn't surprise me if one string is different. I've had strings with different color schemes on the same string. 

Check the wiring on your pigtails. Making sure that the correct pins go to there corresponding wire. I once had 1 pigtail in a batch of 20 that I had ordered where the data & ground were reversed from all the rest. Causing the string to flicker & because of this the first pixel in the string was toast.

You can bypass your pigtails & hook the string directly to the euroconnectors on the pixcon to see if the problem goes away

[Little_b]

Ok so I've checked each string to make sure "in" and "out" are wired correctly, tried different power supplies, different controllers, and all have the same issue.  So now I'm wondering if it's the way I constructed the panels.  After going back and checking the other panels, they all do it now.  So here are some questions I have, and maybe I just don't understand fully how the pixels work.  Here is what I'm working with:

12v WS 2811 Pixel Nodes

360w 30amp Power supplies

Pixcon 16 Controller (Tested issues with Alphpix4 and Sandevices e6804)

16/4 Burial Speaker wire (Tried with SPT2 16awg wire as well)

Matrix Panels are 63 tall X 15 wide. Total of 8 panels making the entire matrix 63 tall X 126 wide.  Strands run vertically.

All power and Data connections are at the top.

Power to each strand is connect directly to the power supply, and not through the controller outputs.  Power injection every 126 pixels at the top.  The V+ and V- from the pixel string stay soldiered together with the power injection wire.  After 10 runs I don't connect the V+ to the next run as the new power supply takes over.  All the V- stay connected together through out the entire matrix.  This makes 630 pixels per power supply.  Thinking it was drawing too much I've tested closer to 500 getting the same flickering results.  Tested for voltage drop and didn't find enough to be an issue.  (Less than .08v)

Data runs are 252 pixels.  (4 runs so the data "in" is on the top.)  So each panel runs to 4 outputs on the controller.  the 4th output has a little carry over to the next panel. Data Cable length for all the testing has only been 6' long. (Trying to avoid having to add null pixels.)

I haven't connected the entire matrix together yet as I am still waiting on some more Pixel Nodes to arrive.  I started with 4 panels and want to expand to 8 for this year.

Here is what I may not be understanding, when I hook up the Data cables and run the "Bars" effect in PE it works fine, except for some flickering.  When I only connect the first Output and disconnect the other 3, all three sections do what section 1 does. Is this normal?  I was thinking if I don't hook up an output, that section of lights shouldn't be doing anything at all since there are no commands being sent. Is this correct?

[wbaker4]

Flickering is typically a sign of data corruption.  The 2811's run at 800kbps.  Any kind of noise on the data line can lead to flickering.

Try using a run of twisted pair wires from the controller to each pixel string.  Run ground and the data signal over the twisted pair and see if flickering continues.  (You still need to supply power, but that will be a different gauge wire).

All of your grounds are connected together right?  (All power supplies and all of your controller boards - common ground?)

You should not have any pixel activity when the data signals are disconnected.

Edited March 23, 2016 by wbaker4

[wbaker4]

...also - since you are using the PixCon16 and WS2811's, I have observed flickering.  I was given a test version of code from LOR, and once that was installed on the PixCon16, the flickering went away.