Group Actions
Engagement
  • Author Author: Instructorman
  • Date Created: Date Created
  • Views 824 views
  • Likes 0 likes
  • Comments 4 comments
Related
Recommended

Set backs and progress on developing a WS2811 LED string controller

Instructorman

First the progress

 

I've developed solid PIC Basic Pro code that can reliably drive a 4m string of WS2811 RGB LEDs at 60 devices/m without spurious resets.  I'm using PBPX 3.0.7.4 compiler, MPASMX assembler and a Microchip PICKIT 3 programmer to download the code into a PIC18F4520 running at 40 MHz (10 MHz external resonator with 4x HSPLL clock multiplier turned on through configuration bits).

 

The base function that generates the necessary pulse trains is attached as a PDF.  There are still optimizations that I'd like to explore, but this chunk of code works well, so I'm building up pattern generating code on top of this function and adding IR remote control.

 

A video clip showing some of the patterns I've generated is embedded below.

 

 

Next, the set backs

 

I'm a little amazed and simultaneously disappointed at how delicate the WS2811 devices are.  Tonight I destroyed a third WS2811 LED device in a way that took me by surprise.  How did the third WS2811 get destroyed?  I turned on a light.  You can see the evil, chip-killing light in the clip below along with the repair procedure that gets the remaining LEDs working again.

 

I know that fluorescent lights can kick out pretty intense EMI when they turn on.  This light has reset many micro-controllers sitting on the desk that it illuminates.  I have replacement LEDs on order, but I think it is time for a clamping circuit on the input to the string. It seems that these are not robust devices.

 

 

Until next time,

 

Mark

Attachments:
imageWrite Array subroutine.pdf
  • in reply to qmiteam

    Very impressive work.  I agree with your assessment of voltage drop and spikes on these devices, ignoring these issues will result in blown devices and/or annoying brightness problems.

     

    I posted a blog in test & measurement that shows how the drop in supply voltage on a 5m string of WS2811s can be measured with a good benchtop multimeter here.

    Although these strings are finiky to work with, I still think developing drivers and code for visual displays is fun.

     

    Mark

  • in reply to Instructorman

    Like many things that come out of China, the WS2811's are really cool, but they are only really cool in a very narrow set of circumstances.

     

    I have built many things out of large quantities of WS2811 LED's, see below for a flexible display I made that uses 2500 of these.

    https://www.youtube.com/watch?v=pvd434dI840&list=UUwRLo1Lwswtw6YuihTl5tYw

     

    https://www.youtube.com/watch?v=m3EbqGQaGto&list=UUwRLo1Lwswtw6YuihTl5tYw


    Hooking up the LED's is the easy part, having them function reliably is a whole different animal.  By far the biggest issue is voltage drop and voltage spikes due to LED turn on.  The inductance and length of the wiring and ribbon cables that the LED's are mounted on CANNOT be ignored.  I'll give you a quick experiment to try.  If you take one of your 5m aliexpress rolls and power it from one end with 5v (you should never use anything over 5.15v with these) and light it up peak white, measure the voltage at the other end of the roll and observe the brightness from the first LED to the last.  When you have a static display (ie all LED's on white), this is very easy to overcome, but when you have dynamic effects, you have to do a little more planning on your power scheme to avoid voltage drops and voltage spikes due to inductive discharge, both of which have claimed many WS2811s.  My advice to you is to plan your power to the LED strips and data line routing separately, just because it is intended that you can daisy chain serial data indefinitely with these, don't make the assumption you can do the same with power.  Also make sure to use TVS (tranzorb transient voltage supressors) very generously on all data and power lines.

  • in reply to michaelkellett

    Michael,

     

    I just installed replacement LED devices tonight and verified the string is working again.  I'm powering the string from a laboratory bench supply and, as mentioned, I've added a 100 Ohm resistor in series with the data line.  Each device in the string has a small surface mount capacitor physically situated on the output side that feeds into the next device in the chain.  Not sure where this cap is connected just yet.

     

    Do you have any suggestions for clamping the data signal or power supply?  I was thinking of using zeners to clamp the data line, but I wonder if they will operate quickly enough to suppress fast transients.

     

    Mark

  • You may need more than one clamping device - at high frequencies one end  of the 4m string has little control over the other. Does the LED 'tape' have power supply decoupling caps ?

     

    MK