Oh No! *Pop*

A project log for FLED

An LED display showing visualizations and rendering data from a variety of TCP sources over the Open Pixel Control protocol

written by Ben Delarre 02/28/2014 at 18:341 Comment

So just after presidents day we came back into the office and found that the FLED was dead! Turned it on and nothing happened. After eliminating the ATX PSU from the equation it became clear we would need to pull the display down and open it up to find out what was wrong.

Once we got it open we discovered that the PSU was tripping out, evidently from a short in the LED chain. We disconnected the LEDs and everything started up again. So it was time to find the short. Since we'd constructed this string of LEDs ourselves it seemed likely that a connection had broken or shorted. We carefully went over every connection and found that none of them were obviously broken. Therefore it must have been the LED units themselves.

Because we're using WS2811 driven LEDs it didn't seem like there was any easy way to debug the LEDs individually. So we had to resort to a binary search, this meant cutting the strip at various points to find where in the chain the issue was found. After 7 cuts we managed to isolate the one LED that was causing the problem. After tearing it off the backplane we were treated to this lovely example of what happens when you have lots of heat sealed inside some silicone!

Once we replaced the LED and reconnected the cuts we made FLED was restored to all its blinky glory. If anyone has any better ideas about how to debug WS2811 strings we would love to hear your thoughts.


Nippey wrote 04/04/2014 at 05:35 point
Well, if the LED does only short the power when it is turned on, but not when it is *just supplied*, you could try to initialise only one LED per run. So with each iteration you have n+1 LEDs running until everything gets dark again. There it is, your little sucker. Alternatively, measure the voltage between VCC and GND. It will get lower the more you get near the dead LED, but behind the LED it wont change anymore. This requires, of course, that you have a measureable voltage drop.

Are you sure? yes | no