WS2811 on Teensy 3.0 using FastSPI_LED library

Hey Paul,

Thanks for the quick reply. Once I receive my batch I can surely take these measurements. I am not certain yet, but perhaps I will order just a meter of those 144/m to review your concerns. Do you happen to have a link to those similar, densly packed once you mentioned?

I was also wondering if the voltage drop couldn't be avoided by adding 5V every meter or would this be overkill? You were saying that you lost 0.04V with a #22 for such short distance? Any recommendations how to avoid this?
 
Do you happen to have a link to those similar, densly packed once you mentioned?

It was email only. I've already deleted it, sorry. Even when I delete messages quickly, my inbox still gets badly backlogged....

You were saying that you lost 0.04V with a #22 for such short distance? Any recommendations how to avoid this?

Bigger wire!
 
this is what happens sometimes when i reboot or upload new code: (this is running the plasma example)
photo.jpg
I guess they LEDs are not actually white, just quite bright.
 
Nice photo Giorgio!

If you have access to an oscilloscope, it might be good to look at the data signals and also the power lines for noise, oscillations, etc. Or if you don't have a scope, at least measure with a multimeter on both AC and DC.

At least one person had quite a lot of trouble from a PC power supply that would create massive noise and spikes on the 5V line output when it wasn't driving any power on the 12V output. I've seen small power supply too have quite a bit of trouble running over a wide range of currents, like these LED project do when the LEDs are all off to suddenly all bright. Clean power is pretty much essential for these projects. Sadly, there are a lot of poorly/cheaply designed power supplies in the world which probably work fine with their originally intended product that uses them in a specific way, but they can act badly when used in a much more demanding application like a big LED project.


any thoughts regarding the '5V for every meter' thought I had?

I think I'll be quite interested to hear how it goes, specifically the voltage measurements.
 
http://www.youtube.com/watch?v=-Ey0ihu0b3o

is a video of the strip glitching during the rainbow example - I am squeezing the data and ground extension cable coming from the teensy to get this glitching. It happens on its own during the plasma example but I couldn't really get it on video - this is hard enough to tell what's going on. it is back and forth btwn rainbow and glitching
 
This looks like a power supply issue. Seems like it's overcurrenting the power supply to me. Have tried swapping power supplies? Almost seems like there is a point where the leds may all turn on and then overwhelm the power supply. May adding some big storage caps would help/prevent a brownout type condition.

Quick Update: I am using a 60A power supply with this setup. It's a kids table with 6 panels of 16x16 leds in a 32x48 configuration. It has 1536 leds. It's pretty awesome, but as Paul says current and heat can quickly become an issue.

KidTable.jpg

https://vines.s3.amazonaws.com/v/videos/AE7AE58E-6C40-4A70-8507-E678D0348615-379-0000002E8F5E346A_1.1.2.mp4?versionId=Lq6xgawc_Hv_hv2sRapYW__2HJQqBXzX
 
Last edited:
This looks like a power supply issue. Seems like it's overcurrenting the power supply to me. Have tried swapping power supplies? Almost seems like there is a point where the leds may all turn on and then overwhelm the power supply. May adding some big storage caps would help/prevent a brownout type condition.

Quick Update: I am using a 60A power supply with this setup. It's a kids table with 6 panels of 16x16 leds in a 32x48 configuration. It has 1536 leds. It's pretty awesome, but as Paul says current and heat can quickly become an issue.

View attachment 611

https://vines.s3.amazonaws.com/v/videos/AE7AE58E-6C40-4A70-8507-E678D0348615-379-0000002E8F5E346A_1.1.2.mp4?versionId=Lq6xgawc_Hv_hv2sRapYW__2HJQqBXzX

Automata ftw! Nice setup indeed. How long do you have this running at a time and how warm/hot does it get? Also where did you get your power supply? Would you recommend it?

Cheers.
 
i build a second panel identical to the first and even though half of one of the strips is burned out right now, this whole panel is working a lot better than the first one - no glitching, all strips come on when I load new code, etc.

to answer your question, i did swap the power supply and the issue stayed with the strip and did not follow the supply so I think it is something to do with the data line, especially since squeezing the data line exacerbates the issue. gonna swap out witha new data extension cable, re-heat the solder joints, and see where that gets me.

haven't had time to pull out the scope yet...
 
When testing out new boards I have noticed that some of the leds are just wonky. Usually I have to replace 1-2 for every 256 leds. That or you may want to touch up each of the led solder pads. A cold joint could definitely cause this problem.

---------------------------

The power supply I used is very similar to these on alibaba. I am planning to have these for sale very soon at www.rgb-123.com. They work really well. You have three circuits of 20A to get the full 60A output.

If you run the leds at full brightness the pcbs get pretty warm/hot. Production boards will have a min of 2 oz copper, right now these are for prototyping at 1 oz copper. Really though running the display beyond a brightness of 128 counts and more practically 64 counts. For the kids table I typically run them at 32 counts since the kids are pretty close to the lights. It really can't be underestimated how stinking bright these leds are.

image.png
Based on this curve there is a diminishing return in the light perceived by your eye beyond 125ish counts. I did a pretty big writeup at www.ohararp.com/blog on the light/color output of these leds.
 
When testing out new boards I have noticed that some of the leds are just wonky.

I've noticed that they are quite sensitive to voltage drop-off. Both in their behavior, and in the hue at 255,255,255.
The Greeled 68/m strips need multiple power taps for a 5 meter length or they may misbehave, and not all the chips appear pure white.
 
These are boards that I have designed and have had fabbed. I buy the WS2812s in 1000 led reels and assemble the boards with my pick and place machine. The testing is during the functional check I do after the boards have been assembled.

The Greeled 68/m strips need multiple power taps for a 5 meter length or they may misbehave, and not all the chips appear pure white.
Woz, I had a simlar problem. The difference in color is from the voltage drop from one end of the strip to the other. I am running 12AWG silicone wire between boards to avoid this. That and running 5V power in "parallel" to the boards/strips versus a series input is the way to go.
 
Last edited:
That photo sure doesn't look like LEDs packed so tightly to get 144 per meter....
If anyone does get one of these 144/meter strips, please turn the LEDs on fully white and make some quick voltage measurements?

If you scroll down to nearly the bottom of the page, there is an image of a very densely packed ws2811 strip. It's interesting, because it looks like there are solder pads on the edge all along the sides...perhaps to allow additional power input.
According to the spec. a 5m reel will draw nearly 40 amps!
 
The power supply I used is very similar to these on alibaba.

I'm using 10 of these power supplies. So far they seem to be working perfectly.


**update
I should also note that I'm powering 160 LEDs per teensy pin from both sides. And I'm the pumping 60A of DC thru a 3"x2" prototype board I had printed at OSH Park. Here is a photo of it:
20130511_iphone_0285.jpg


please note that the #18 wire shown the pic quickly got changed to #10 :) Other then that no heat issues with the board.
 
Last edited:
Hunter, cool board! A breakout board like this is next on my list. Good call on switching to the 10AWG. Got any pics of your led display?
 
ok so that glitching was from the resistors i used. i thought they were 220s but maybe not. i replaced a strip after it seemed the strip was the issue, but then 1,2, 7 and 8 were all not working properly. bypassing the resistors on 1 and 8 got all 4 back up and now i have 48 2m sections playing live video using my firewire camera as a capture card for the LZX video synth. pics and videos soon - off to install now. thanks for all y'alls help!!!
 

Based on those 72s dense setup 144s seem hard to imagine.

I'm using 10 of these power supplies. So far they seem to be working perfectly.


**update
I should also note that I'm powering 160 LEDs per teensy pin from both sides. And I'm the pumping 60A of DC thru a 3"x2" prototype board I had printed at OSH Park. Here is a photo of it:
20130511_iphone_0285.jpg


please note that the #18 wire shown the pic quickly got changed to #10 :) Other then that no heat issues with the board.

Those are quite nice. Do you sell these?
 
Back
Top