Project help

Status
Not open for further replies.
P

play2liv

Member
Hi All


I am working on a project with the Teensy 3.1 and the Octows2811. I have made a cylinder withe the 4 strips of leds with 74 led in each strip. I have a 1000uf 35v compacitor on the power in lines. it is 18guage wire. almost all strips are power fed from both ends. i have a 12v to 5v ubec providing power, to strips and to +5v and GRD on the on the OCTOws2811. i take the Teensy off the Octows2811 header when i program it
i have followed the guide. As best i can. I have loaded the Teensy with the basic test sketch. Also tried the rainbow sketch. it will turn on with green and blue solid on the strips then after a second start blinking that is all it does Any ideas what it might be?
 

Attachments

  • 20160319_114329.jpg
    20160319_114329.jpg
    118.5 KB · Views: 175
  • 20160319_114152.jpg
    20160319_114152.jpg
    182.6 KB · Views: 155
  • 20160319_114140.jpg
    20160319_114140.jpg
    103.5 KB · Views: 123
Last edited:
What is the current capacity of your PSU would be the first question with that many LEDS. Suggested first step is to drop the ledsPerStripValue to 8 since that will cut out a number of potential troubles.

You might also add a routine to blink the onboard LED while it's cycling the colours, since it's possible that the Teensy is being hung up.

You can also try the Adafruit neopixel library and point it at one strand at a time. This allows you to verify the strand wiring power etc in isolation.
 
I used adafruit.com neopixel Uber guide in planning. I have 4 19 led strips on the first 4 outputs. 16 19 led stripes total 292 leds.
I have tried to use the USB to power it. That lit up the first 19 of 74 leds. In the first 3 strips. Last strip was out. That was also only power feed from the beginning and end of the strip. Then put more power leads in the middle of the strips. Was able to light the first 8 strips red from a 26650 lipo 3.7v then the second half the same way. I am currently using a 12v 8Amp lead acid battery for bench testing with a 12v to 5V 4 amp UBEC to power the led matrix. see attached diagrams. Left one is how the martix is wired. Right is how i plan to power it later. bottom is how i am powering it now. also see this video https://www.youtube.com/watch?v=ZuHYyJ77zeM
 

Attachments

  • IMG_20160320_0001.jpg
    IMG_20160320_0001.jpg
    73.6 KB · Views: 153
  • IMG_20160320_0003.jpg
    IMG_20160320_0003.jpg
    63.9 KB · Views: 140
  • IMG_20160320_0002.jpg
    IMG_20160320_0002.jpg
    34.7 KB · Views: 111
Last edited:
When a RGB LED is lit red it's normally a sign that the voltage is low. Target voltage is 4.5-5.5, and any ripple there will tend to show up as jitter in the colours and/or corrupted data.

USB power is only good for about ~60 LEDs, and if you got 19*3 to work that's 57 so pretty much on the mark. your full rig will need 5.8 Amps at 5 Volts run. And even if you tame the brightness back there will still be times when it's drawing that full 6amps to run because of the way the PWM in these work.

powering LEDs is hard.
 
Do you have a voltmeter? Any chance you can try measuring the 5V power at various places on the strips while the problem is happening?

Also, if you meter has a low voltage range, try measuring between ground to ground at places, like from the ground at Teensy to the ground at the far end of the strips or the ground at the power supply. Since they're all wired together, you should see zero volts. Grounding issues are common problems, so if you see more than about 0.2 volts between any 2 ground locations, that's cause for concern.
 
https://youtu.be/aOox-nuDXHE I was wondering if i can put all the leads from the matrix together. all positive together on one wire, all ground together on one wire. then take multiple step down converters and hook up to one set of wires going to matrix.
 
Last edited:
Looking at the video it looks like the strand you have are working fine, and the Teensy is happily ticking over. So testing is very much as paul noted above making sure you really do have 5V everywhere, and also making sure your data is propogating correctly wherever you have a light out where it's neighbour is working. If doubt really exists if you have a dead LED (not at all unknown from Ebay) and you don't have ane oscilliscope you can splice on a short length of strip in parallel at the dead pixel to confirm data is coming out of the own upstream since the pixel outputs can drive more than one input. Scope is easier if you've got one though.

This also allows you to check that it's not something code related that means you just have pixels se to black, though the example code should be running more than 19.

Re power, it's not clear what you mean with multiple boost converters. Simplicity would have the minimal number of power supplies involved, but if you already have batteries and converters in some arrangement you can use them in parallel as follows:

Under no circumstances connect +ve to +ve, double check to avoid that
All the -ves must connect
You will need seperate power caps for each power supply
Avoid plugging or unplugging while everything else is running
Where segments on a pixel have different power supplies (teensy-pixel 1 or pixel n to pixel n+1) you must have +ve voltages within 0.5V of each other or your risk data corruption since one chips 1 may be the next chips 0.
 
Hi again thank you for your help so far. I have decided to go with one bigger 12amp buck converter. Ordered it should be here in a month. Ugg. WISH foreign eBayer's would use a priority shipping service. I am able to light 35 leds per strip before they start to flicker. The ubec are preforming at their specs. The 35 per strip are drawing 3.1 Amps on white and yellow. Just at the limit of the ubec. Have some 3d printing to do in the mean time. A battery case and some support to keep it round. Will post when I get the parts in.
 
Hi again thank you for your help so far. I have decided to go with one bigger 12amp buck converter. Ordered it should be here in a month. Ugg. WISH foreign eBayer's would use a priority shipping service. I am able to light 35 leds per strip before they start to flicker. The ubec are preforming at their specs. The 35 per strip are drawing 3.1 Amps on white and yellow. Just at the limit of the ubec. Have some 3d printing to do in the mean time. A battery case and some support to keep it round. Will post when I get the parts in.
 
Hi All


I was doing so well. I got the power supply. i hooked it up once i adjusted it. All 74 leds per strip and all 4 strips were lit up. no flickering, i was running your rainbow sketch from the library. the color change was a little off. the color change was a little off. i checked the voltage it was vering a bit. between 3.4 and 5.1 volts. the current was about 10.7 amps. i was going to take some video and i turned it off. whan i plugged it back in the fancy smoke came out. Dammmm!! some how when i hooked it up to the power the out put voltage got turned up to 12.4 volts. ssame as imput. Toasted the Teensy.

can it be repaired?
 
Re repair, the answer 'not really' since it's likely the only re-usable part is the PCB, and even that may have burnt tracks so the repair would be something along the lines of 'hot air gun all the components off and replace each one.

Most likely you will need to replace every 5V rated part in the rig since both the LEDs and the Teensy are only good to 5.5V so got hammered by over 100% overload. It's not impossible that some parts survived if something fused short circuit but it'll be slow careful work getting everything apart and then testing them one at a time.

Teensy - see if you can get blink onto it
LED strips, see if you can supply 5V with no data in and a low current. If you can get stable 5V then it's worth seeing if they respond to digital data

re octoboard 74HCT245 is rated to 7 volts, so it ONLY got a 80% overload. You can test that in isolation if it seems worthwhile by soldering some LEDS on the end of a RJ45 cable and seeing if the blink moving the inputs from 0 to 3.3V. see layour https://www.pjrc.com/store/octo28_adaptor.html
 
Last edited:
Thats what i figured, just had to ask. The pain of learning. Do have a nother question can i power the teensy with a diffrent power supply? or does it have to be the same source as the leds?
 
You can have different power supplies, but you need to ensure that

the negatives are connected
The positives are NOT connected in anyway
Neither connect to ground - This is not normally a problem and they 'float' but with larger capacity supplies you may find one that grounds the negative, and another that sits -2.5V - +2.5V. This sort of things results in noisy signals and potentially smoke, so can be worth the checking as things get more serious. Simple step is to measure negative to negative and see if there is a difference in potential.
 
Hindsight is obviously a wondrous thing, but may I interject and say this is where good planning from the start can be very useful?

Those WS2812's can pull up to 60mA at full white brightness (obviously depending on voltage applied). 272 LEDs at 60mA = full load current of 16.32A. You always want to avoid overloading a power supply, so taking into account a bit of overhead, you'd probably be looking at something like an 18A supply to be well on the safe side.
 
See my original post above. Certainly don't just plug them in to see, since reasonable chance something will have failed short and cause more destruction. If you are in a hurry just replace everything, if you have the time then you'll need to try each element separately to see what state they are in as described above. And yes this will take a while since you can't trust that just becuse one element works that all elements will. Every individual part will need testing in isolation with a degree of care.
 
I had an incident with ws2812s once. I thought I blew out the strips. As it turned out, only the individual LEDs closest to the power connection were blown. However the blown ones would not pass the signal, so it appeared that the entire strip had failed.
I was able to cut the bad ones off and join a new ones in their place, and all was well again.
 
Last edited:
Status
Not open for further replies.
Back
Top