Teensy 3.0 boards dying on me!

Status
Not open for further replies.
L

luigo

Member
Hi Paul,

Apologies for the long post - I'm hoping to provide as much detail as possible on my issue.

We recently started a small project based on the OctoWS2811 library and tutorial you kindly shared on your website. The project is relatively simple compared to the tutorial's large LED array displaying video. Ours is a simple Christmas tree with LED 8 strips going across it horizontally like 'belts' at different heights. Each belt is a single LED strip and because the tree is effectively a cone, every strip has a different length. We are using a Teensy 3.0 and the OctoWS2811 library to control individual pixels in our array. Each strip is connected to a PIN in the Teensy board. In total, we're controlling approximately 320 individual pixels; the shortest strip having just 8 pixels and the longest just over a hundred.

We're using a 5V, 60A power supply which suppressing, has a fan! All the previous ones I've bought don't have a fan and can get really hot after a while.

Anyhow, our first attempt was successful (for a while!). After making all the connections and uploading our code, we had all 8 belts working and displaying the rather simple animation we wrote which basically just alternates between an upward/downward motion and fixed colours.

After about 4 minutes, the LED array suddenly stopped showing the animation. I found that odd and unplugged the PSU feeding both the Teensy board and the strips. When plugging it back in, I had no lights or animation being shown at all. I then tried connecting to the board via USB and found that the PC was no longer able to 'see' the device. I tried the process suggested on the troubleshooting guide unsuccessfully - the PC just wouldn't detect a device being plugged in. No messages at all in my /var/log/syslog file. It is literally like the board died!

I first assumed I had a faulty connection somewhere (though I couldn't see any on visual inspection) and decided to start all over again with a brand new board, new cables, new resistors, new PSU, etc. To my surprise, I had exactly the same problem! After 5 minutes this time the board died just like the previous one.

I now have three dead (or apparently dead) boards and I was hoping to pick your brains before I carry on burning more! :p Surely, I'm doing something wrong but after going through the tutorial step-by-step several times and checking my connections, I just can't see what's wrong!

I tried with two different PSUs but both are the same model and from the same vendor. Would it be possible that these are somehow killing my boards? How could I tell if that is the case? Finally, is there anything at all I can do to verify if my boards are truly dead?

Many thanks, in anticipation, for your advice,

-Luis
 
I read through the tutorial and I wonder if you implemented the series resistors as well as the 74HCT245 buffer / voltage translator? The buffer may also provide some nominal protection from PSU spikes, which is perhaps how your teensy got fried.

With a large array, it could be worthwhile to introduce a crowbar circuit to trip a polyfuse whenever the 5V voltage supply goes off the reservation. That is, combine a beefy 5.1 zener diode with a large polyfuse, which would safely divert excess voltage away from the Teensy's outputs. Then, as the voltage levels come back to normal, the zener stops conducting and the polyfuse cools and resets.
 
Last edited:
Hi Constantin,

Thanks for the reply. I did add the 220 Ohm resistors but not the buffer chip. I will order a few right away and test again. Quick question though - I can see that the 74HCT245 chip needs to sit between the Teensy signal pins and the LED strips. If I add the buffer chip, do I still need the resistors? Also, is it ok to power all three elements (the Teensy, the 74HCT245, and the strips themselves) from the same PSU?

Sorry for the newbie questions and thanks again for your help.

-Luis
 
I hope I'm helping. (!!!)

My understanding is that the resistors help attenuate any reflections in the signal traveling from the teensy, hitting the end of the wire and bouncing back. This is called source termination, IIRC. I'd counsel you to take an oscilloscope to have a look at the quality of the signal, i.e. are you getting nice square waves at the LED strip input / buffer chip or not. Short wires apparently do not benefit as much from a source terminator as much as longer ones do.

As for the PSU, it might be worthwhile looking into a dedicated small power supply for the teensy and a separate, beefy one for the light strips. They will still have to share GND, however. If the problem persists despite the use of the buffer chip, I suggest you go with an opto-isolator instead of a buffer. That would give you complete isolation of the Teensy from the LEDs (including ground) and hence minimize chances of issues arising. The only downside is figuring out which opto-isolator would be good enough for your application. I'd prefer experts like Paul to chime in - I could only take stab in the dark and suggest a digital isolator like the SI8440BB, which has a rise and fall time of less than 500ns (0.5us) since that is what the 74HCT245 features.
 
Last edited:
The big LED arrays and power cabling might have large impulses due to large changes in current flow and the inductance of LED the power wiring. The LED power of course needs to be on an independent power supply, with a proper common ground with the microprocessor. If you must use a single power supply, there may be a need to use decoupling of the current-induced spikes. A resistor will only slightly do so. Best is a proper choke in series with the power to the microprocessor. That would be fairly large in size - maybe 1-2 in. diameter on a torroid form.

But best is to use a 2nd power supply for the low current demand of the microprocessor. Make the common ground between the two power supplies be at the actual terminal of the big power supply, not after several ft. of wire going to the LEDs. And the inductive caused pulses on the LED power wire could still get into the microprocessor, I suppose, without the proper inductor-filter/choke.
 
Last edited:
Gents,

A late but nonetheless big thank you! Having implemented separate power supplies for the LED strips and the Teensies, things seem to be working fine. At least I stopped frying boards like there's no tomorrow.

Many thanks,
 
luigo said:
Gents,

A late but nonetheless big thank you! Having implemented separate power supplies for the LED strips and the Teensies, things seem to be working fine. At least I stopped frying boards like there's no tomorrow.

Many thanks,
Click to expand...
You should go back to the prior flawed method. So that Paul can sell more boards. :p
 
That's definitely not the way I'd like to be selling more boards!

But I would love to see a photo or video of it working. :)
 
Cheers guys.

@Paul, I'll definitely post a video of the Christmas tree once it's completed. I should add a 'Powered by Teensy 3.0' legend somewhere!
 
Status
Not open for further replies.
Back
Top