Another thing about the wiring. Have you investigated how power is fed into the LEDs? With 24VDC, no more than 5 can be connected in series. Yet I do not see series/parallel connections in the wiring.
You do not discuss the wiring. On my set of 400, 2 wires run to feed 200. At the first LED, there are 4 wires. Thereafter, there are 4 wires, it appearing that 2 run the length of the 200, while 2 go in to an LED, & 2 come out. Have you looked further into the light string?
man I would love to make a home made controller replacement for these, I have several strands that died from water intrusion and it seems a waste to throw them out
For those following along I have made some progress with help from some people. We suspect these are based on the QED3110 LEDs. Documentation is poor but the one thing we do know is that the LEDs have an address burned into them when in the factory ... and the data packets I have now decoded by writing a Saleae plugin which turned the waveform into my suspects stream of 1s and 0s. From this I can see each "WORD" contains some start bits, an address and 24 bits of colour. With a sample of 1 it is possible the start bits are just common address bits but i suspect not. I do see address 0 is odd as the data stream is one bit shorter than all the others and there does not seem to be any data for LED 1. Other than that it looks pretty straightforward. That means a part 2 is looking more likely. For those curious here is the first few words in a frame: (spaces added by me to the first few to make the suspected sub-parts more obvious) 11 0000000 00000000 00000000 10100100 11 01000000 00000000 00000000 10100100 11 11000000 00000000 00000000 10100100 11 00100000 00000000 00000000 10100100 1110100000000000000000000010100100 1101100000000000000000000010100100 1111100000000000000000000010100100 1100010000000000000000000010100100 1110010000000000000000000010100100 1101010000000000000000000010100100
The address is confirmed. Some people have attempted to repair broken strings and replacing a bad bulb requires replacing it with the exact same bulb from another string or it will light at the wrong time. Thanks for the shout out btw. Any chance of a link to my page in the description?
also i'm incredibly interested in finding out if the pro controller works the same way. If we could put any string we wanted on them that would be amazing.
Despite hearing/seeing the name mentioned a number of times on FB pages today was the first time I came across someone using these lights. This video has been my first dive into whether there is any boards been developed to get full control over these lights.
You should be able to feed the 10.8V directly to the output via the D1 and remove Q4. D1 will prevent the pulsed 24V from feeding back to the 10.8V supply and you won't have any potential gap in the 24V / 10.2V transition on the output. It can probably be simplified even further with a bit more thought.
I believe I may have a generic version of the EXACT same pixels you have! They're all over Amazon now as RGB Christmas light sets, and unlike those, they're suuuper cheap! But the controllers for them don't have WiFi at this price point and really suck, so I'd love to hack them. They have only two wires and run at 29V, and on an oscilloscope the signal goes between that and 10V just like yours.
Great work, looking forward for a wired solution to be run from ex a Rasperry Pi HAT or simular. Already tested out my Twinkly through wifi and Xlight.
I am really not looking to make a product. The best you will get is a proof of concept code for a Pico in a Github repo ... turning it into a controller will be left as an exercise for others.
This is good investigating. I'm not sure what your motivation is though. You already have an integration into xlights and really you might have a better result flashing new firmware. It's an 8266 chip. I can also tell you that the strip lights they sell aren't encoded strangely like this. They use 12v and use a standard strip in them. I've done a lot of "hacking" of their stuff myself. I had an integration for xlights before you had an official one, and as far as i'm aware, i'm the only person who has managed to reverse engineer their web api and decode the mapping. My program now has an in-app editor as well as an import to pull in xlights models.
First and foremost the motivation is curiosity and a desire to see if they can run without wifi. Wifi does not scale or requires pre-positioning content which then only works with their effects ecosystem which limits the ability to integrate it further with other display elements. Sending ArtNet data to "many" devices over Wifi either requires an advanced wifi setup or it will suffer lag (sometimes unpredictably ... ie it works until it doesnt).
Another thing about the wiring. Have you investigated how power is fed into the LEDs? With 24VDC, no more than 5 can be connected in series. Yet I do not see series/parallel connections in the wiring.
You do not discuss the wiring. On my set of 400, 2 wires run to feed 200. At the first LED, there are 4 wires. Thereafter, there are 4 wires, it appearing that 2 run the length of the 200, while 2 go in to an LED, & 2 come out.
Have you looked further into the light string?
man I would love to make a home made controller replacement for these, I have several strands that died from water intrusion and it seems a waste to throw them out
After Christmas I hope to get back to this.
For those following along I have made some progress with help from some people. We suspect these are based on the QED3110 LEDs. Documentation is poor but the one thing we do know is that the LEDs have an address burned into them when in the factory ... and the data packets I have now decoded by writing a Saleae plugin which turned the waveform into my suspects stream of 1s and 0s. From this I can see each "WORD" contains some start bits, an address and 24 bits of colour. With a sample of 1 it is possible the start bits are just common address bits but i suspect not. I do see address 0 is odd as the data stream is one bit shorter than all the others and there does not seem to be any data for LED 1. Other than that it looks pretty straightforward. That means a part 2 is looking more likely. For those curious here is the first few words in a frame: (spaces added by me to the first few to make the suspected sub-parts more obvious)
11 0000000 00000000 00000000 10100100
11 01000000 00000000 00000000 10100100
11 11000000 00000000 00000000 10100100
11 00100000 00000000 00000000 10100100
1110100000000000000000000010100100
1101100000000000000000000010100100
1111100000000000000000000010100100
1100010000000000000000000010100100
1110010000000000000000000010100100
1101010000000000000000000010100100
The address is confirmed. Some people have attempted to repair broken strings and replacing a bad bulb requires replacing it with the exact same bulb from another string or it will light at the wrong time. Thanks for the shout out btw. Any chance of a link to my page in the description?
also i'm incredibly interested in finding out if the pro controller works the same way. If we could put any string we wanted on them that would be amazing.
Despite hearing/seeing the name mentioned a number of times on FB pages today was the first time I came across someone using these lights. This video has been my first dive into whether there is any boards been developed to get full control over these lights.
great work. fingers crossed to crack these. Id love to try to get these working with WLED.
You should be able to feed the 10.8V directly to the output via the D1 and remove Q4. D1 will prevent the pulsed 24V from feeding back to the 10.8V supply and you won't have any potential gap in the 24V / 10.2V transition on the output. It can probably be simplified even further with a bit more thought.
I will try that. I am sure there is lots of opportunity to do better. My electronics is pretty basic.
I believe I may have a generic version of the EXACT same pixels you have! They're all over Amazon now as RGB Christmas light sets, and unlike those, they're suuuper cheap! But the controllers for them don't have WiFi at this price point and really suck, so I'd love to hack them.
They have only two wires and run at 29V, and on an oscilloscope the signal goes between that and 10V just like yours.
Can you identify those Amazon-sourced light sets that use this same protocol? Thx
Great work, looking forward for a wired solution to be run from ex a Rasperry Pi HAT or simular. Already tested out my Twinkly through wifi and Xlight.
I am really not looking to make a product. The best you will get is a proof of concept code for a Pico in a Github repo ... turning it into a controller will be left as an exercise for others.
Many Thanks for a great video.
This is good investigating. I'm not sure what your motivation is though. You already have an integration into xlights and really you might have a better result flashing new firmware. It's an 8266 chip. I can also tell you that the strip lights they sell aren't encoded strangely like this. They use 12v and use a standard strip in them. I've done a lot of "hacking" of their stuff myself. I had an integration for xlights before you had an official one, and as far as i'm aware, i'm the only person who has managed to reverse engineer their web api and decode the mapping. My program now has an in-app editor as well as an import to pull in xlights models.
First and foremost the motivation is curiosity and a desire to see if they can run without wifi. Wifi does not scale or requires pre-positioning content which then only works with their effects ecosystem which limits the ability to integrate it further with other display elements. Sending ArtNet data to "many" devices over Wifi either requires an advanced wifi setup or it will suffer lag (sometimes unpredictably ... ie it works until it doesnt).
@@led_man You're correct. I must have been thinking of gen1.
Nerd