Why would sections of an RGB LED string freeze animating after 30 mins?

[Pretzelzzz]

Hi all,

I have a t1000s RGB controller:
http://www.kutop.com/program/T_1000S_SD_card_program_manual.pdf

and 300x WS2811 LED pixels end to end - with 5V 20 Amp power inserted at either end:
http://www.ebay.co.uk/itm/like/DC-5...18372&clkid=891682632273770700&_qi=RTM2063741

The RGB controller (when it doesn't contain an SD card with presets) shows one demo pattern on the RGB string plugged in. This lasts 2 minutes, cycles through R, G, B and fades through some colours. Then repeats.

When turned on the first time, the preset loops continually for about 30 minutes. Then about 75% of the lights (starting at the far end of the string) freeze on their current colour, and the remainder continue to animate. Then some minutes after, another section of lights freeze on their colour, and a few minutes after that, more sections freeze... until there are only about 40 LEDs near the beginning of string which continue to animate. If the controller is then unplugged at the wall for 10 seconds, and replugged in, only those 40 LEDs power on, the rest stay off. And sometimes only 1 LED powers on, animating.

Then if I leave the controller unplugged overnight and turn on the next evening, it all starts again. (Guess the controller lost its memory by then). So once again, everything works perfectly for about 30 mins, then sections stop animating until only a small section near the beginning of the string animates. I've tried this three times nights, with consistent results. -- See screenshot attached of how it looks after about 40 minutes. All lights should be on the same colour, but you can see the sections which have gradually stopped animating but stay lit.

I'm guessing this is an issue with the controller - but why would it work perfectly for the first 30 mins, then quickly degrade?

Many thanks if you have any info that might help solve this mystery...

 

[(*steve*)]

Try injecting power at the far end and see if it improves things.

 

[duke37]

Possibly due to excess temperature.

 

[Harald Kapp]

A thermal defect is one possibility as noted by Duke.

Another issue could be with the firmware in the controller. The program used to cycle through the pattern may run amok, hog to much of memory or whatever can lead to the symptoms observed.
You could try putting 2 different programs on an SD card and once the controller starts showing these symptoms switch manually to the other program. I'd expect the controller tor eset when switching between programs, but of course there's no guarantee.
If it resets and the new programs starts correctly and shows these symptoms after some time, it is very likely a fault in the controller's firmware.
If it resets but the new program does show these symptoms from the beginning, I'd expect a thermal issue.

Does the controller become hot? If so, does the behaviour improve if you put it in a very cold environment (e.g. a cooler)?

 

[Pretzelzzz]

Thanks for your input, I look forward to trying the suggestions when I return home in a week. I can confirm that power is injected into the end. And the controller is in the living room, which is often cool at the time I've switched on the lights. With the lights themselves obviously outside, which is about 1 to 6 degrees C this time of year.

I'll definitely try running some presets instead as Harald suggested, to see if this has some affect. I am disappointed that the controller itself couldn't do something as obviously as running its own default preset for a continuous period of time without some major hickup... how was that not address in quality control I don't know.

To be continued.

Merry Christmas to all!

 

[(*steve*)]

Just to be clear, my suggestion was to also inject power into the far end, i.e., connect both ends to power.

 

[cjdelphi]

The fact it takes a while suggests heat..

I'd put money on the controller being faulty than the addressable strip

 

[(*steve*)]

True, but the temperature may vary either the output voltage from the power supply or the voltage sensitivity of the assessable LEDs. If either of these only happen at the end of the string then increasing the voltage at the end of the string may be sufficient to keep them working.

These devices draw significant current and the tracks on the flex PCB they're on are hardly "beefy". You'll find lots of references to the need to inject power regularly on long strings.

It may not be the solution but it's really easy to test before moving on to more complex possibilities.

 

[Pretzelzzz]

Just to be clear, my suggestion was to also inject power into the far end, i.e., connect both ends to power.

Hi Steve, thanks for your reply, just confirming that yes, power is injected at the beginning and end of the string.

I'm home now, so over the next few days will experiment with running different presets from an sd card, rather that 1 looping direct from the controller's stock memory. If that fails, I guess I can buy a new controller with the hope it doesn't exhibit the same problem again.

 

[Pretzelzzz]

Well, it's been a month since I wrote, but I can conclude that this very strange problem happens even with a replacement controller! It doesn't matter whether its the controllers preset that's looping, or some of mine from an SD CARD... after the same time length (about 35 minutes) of running perfectly, lights from the end of the string in clumps stop working... with more ceasing to work over the next few minutes, until only a few work near the beginning of the string. And if resetting the power at this point, only the first few lights on the string light up (and animate), and the other 270+ lights remain off. What the heck is the deal? (And that's with power injected at both ends of the string.) Extremely frustrating as I've paid over £100 for the whole setup, and have nailed the lights around the garden. And that they work perfectly for first 35 minutes of runtime!

 

[(*steve*)]

Hie long is the cable run from the controller to the first LED?

Edit: I've reread the whole thread again. You note that "sections" fail. Are the sections which fail consistent? If there something significant about the first LED in these sections (e.g. They are connected by wires to the end of the previous section)?

 

[Pretzelzzz]

Thanks for your reply Steve, I'll look and find out, those sound like very good questions...

 

[(*steve*)]

Some background on those questions: there have been issued reported when the connections between the chips are too long.

To combat this, some people insert another module/chip to break up excessively long cable runs. If these are not wanted they are covered in heatsink of otherwise blacked out.

I'm not sure what "too long" is.

 

[Pretzelzzz]

Would these be 'issues' that taken some time before they manifest?

Also, I know that the pixel strings were 50 lights x 6 strings... but what I'm seeing from the last picture I posted (and here it is again with some different markings) the sections of lights which stop working are uneven in length. For example, the first 75% of lights which stop working, start at the far end of the string working their way forward. That's about 200 lights in red, on the photo. (they go off screen, it's a rectangular patio fence). Then there's about 20 lights frozen in white, and about 20 lights still animating (green).

The original sequence takes all the lights (300) and cycles them all through uniform colours, so you can clearly see where sections of the light string have stopped animating because of their different colour to other sections.

Yet, all lights remain with power. So it seems it's the timing on the data cable which has messed up?

But, I have still have yet to confirm with you your questions, consistency of the sections, etc. I hope to also take a video showing the behaviour, that might need to wait a few evenings. Thanks again for your input

 

[(*steve*)]

Yes, it is almost certainly timing related. However it is probably also related to temperature. Various factors can reduce the tolerance to out of spec signals.

We've covered reduced voltage, now we're investigating the possibility that long cable are causing a problem. Whilst cable length might appear innocuous, you can get reflections and distortion of digital signals. The chips get warmer as they operate and this can also affect the tolerance to noise etc. This is almost certainly a contributing factor. Whilst you can't do anything about that, you can keep the signals in spec.

 

[BobK]

I suspect it's a timing problem in your code. I was going nuts trying to get WS2812 strings to work. The spec said that it would reset at 150us in the low state. With a bit of debugging I found that they actually they reset with somewhere between 5 and 6 us pulse of low! An unexpected reset will do exactly what you are seeing. The LEDs early in the string are working and later in the string are frozen because it is resetting before getting to the data for them. You have to make sure there is never more than about 4us between pixel output for the entire string.

I was also under the impression that the data was shifted along the string as you input it. This is not the case. The first 24-bit data goes to the first LED, it then switches to a state that passes the next 24-bit data to the next one, and so one. So, if you have a string of 300 LEDs, you output the data for the first one (nearest the input) first, and not last as I originally thought.

Really neat stuff though, after you get it working!

Bob