I am working on a project with 6 total 24v WS2814 RGBW COB strips, each 3 meters long. Power supply is 24v 300 watt. Controller is Quinled Dig Uno. Power wires are 16 AWG. Data is 22 AWG. There are two groups of 3 strips with a 3.5 meter gap between the two groups. The group closest to the controller works perfectly. The group after the gap lights up, seemingly completely unresponsive to data input. Each time power is cycled, a different random pattern is on the three strips, with some parts of the strips not lighting. All strips were individually tested before install and confirmed working.
See attached image for wiring diagram. The data signal runs on its own, far away from any ground signal. I have tried the gap with 16 AWG and no difference.
I am guessing there is corruption as a result of this “long” gap? Would something like the Quinled data booster help me? What about using something like Cat5e cable (shielded, twisted pairs)?
Check the power connections and measure the actual voltage at strip 2 (far one) compared to strip 1 (close one). You could also try setting up a segment that starts at the position of strip 2 and only displays 20 LEDs so none of strip1 is lit, only strip 2.
If you still have no control, I would try to disconnect strip 2 from strip 1 and move the controller to drive strip 2 directly. This is to prove you can get proper display from strip 2 when supplied with proper data.
If not, fix strip 2. If it does work, then you’ll have to try and see if the output of strip 1 will actually drive something closer than 3.5m (which is not a huge distance).
Yes I agree I need to try and isolate the second side to troubleshoot. I did check the voltage on all three strips on the second side and it is measuring 23.7 volts, so practically no voltage drop.
If I use the segments, the first strips still are supplying the data signal to the second side, so I am dependent on the chip’s ability to drive the line, right? Does something like the Quinled data booster provide more output current? On their site they say the Uno should be able to drive 10m with no problem, so I might just run the second output from the Uno straight to the second set and see if it works that way
Well I am 99% sure I have solved this issue. When the LEDs were first installed, I plugged them in to test the power connections. It is very possible that one of the loose tails shorted 24 volts to the data wire. This would take out the first IC and pixel, and cause the symptoms I am seeing. I removed the suspected strip and now the remaining two are working perfectly. I will cut out the first pixel of the suspected strip and replace.
Sadly, I am still struggling with this. I replaced the damaged strip entirely with a new one, and still have a corrupt data signal making it to the second side. I have tried shielded coaxial cable (with the shield properly grounded) and no difference. I also tried using the second output from the DigUno and still no difference. I will try again using the second output on the DigUno with the shielded cable, then putting the data booster halfway and see if that helps.
Coax cable is exactly the wrong thing to try here!
Without proper circuitry (much, much more complicated) to drive the cable correctly, coax will only add more problems to your setup.
I normally will use 22/4 wiring for a jump of that size, but it is borderline without “help”. I would strongly consider using a TxRx pair if your setup is very noisy. Checkout Long Data Lines in the KB. They work very well with all kinds of cable, including Cat5e (that’s what Cat5e was designed for).
Once you go to RS485, you eliminate all kinds of distance issues for your data lines, it just works.
I soldered the R0 pads together which puts a 120 ohm terminating resistor in circuit on both boards.
My connections are as follows:
Vcc → 5V (provided by buck converter)
Tx → Output from ESP32; Rx → Data (strip)
GND → Ground
A → A
B → B
Earth is left open.
The Tx and Rx LEDs on the boards are lit solid, respectively.
Changing the program on WLED causes flashing and random colors on the strip, but nothing sensical.
I also went back and tried using the DigUno output directly with a Data Booster halfway across with the same results.
More information:
I tried connecting two strips before the jump with a long data wire, simulating a gap. When the wire was as long as the gap I am trying to bridge, the downstream strip began to flicker. Unlike the other lights, this still maintained most of the same color, with glitches here and there. Cutting the long wire in half and splicing in a Data Booster solved the flickering and glitching.
Now, replicating the exact same thing with the strips that have not been working did NOT work!!! Still, I got complete nonsense on the other side. I shortened the data cable as much as possible, with the Data Booster exactly in the middle of the gap, and still the exact same thing. Switching the switch on the booster does nothing.
Starting to think I might have a bad strip??? I have a ton of left over pixels. I originally ordered some of these from Amazon and some from BTF direct. Maybe I can try and wire up a small strip either in the middle of the gap or after the gap to rule out that the strip after the gap (which is already a replacement) is faulty.
I may also try with a different MCU wired directly to the “bad” strip, and see if it even works then
Try the TxRx pair on your long cable with good strips.
It sounds like you wired the pair correctly the 1st time.
Those are the same pairs I use and a large number of setups.
They just work and will definitely outdistance the data boosters, work in much noisier environments.
sorry i dont understand your schematic. do you connect the data output from the controller to multiple strips in parallel?
some theoretical background;
the data is a 800 Mega bit / second stream of “1” and "0 " where the difference is the timing of the “5volt” period - with only 0,4 microseconds difference. this means that any capacitive load on the line can garble a “short” into a “long” pulse.
this type of very simple data is not helped by thicker wire. you would need to know the impedance of the output and match that - wich is not for beginners.
i think the best is to run the data connection over a twisted pair together with ground.
If you indeed load the controller output with multiple strips, you may need a “booster” - but a fast one. RS485 is mostly only run at rates like 4800Bps or maybe 32800, with is nowhere near the 800000000 bps of these LEDS!! this is serious radio-frequency stuff.
Reliable RS485 speeds are largely dependent on the cabling and driver chips used.
With the popular MAX485 driver IC’s, you can easily get into 10Mbps over <50m distances (more with better cabling).
That’s good for LED data, where clock speeds are at 800Kbps, but critical pulse widths are at 150ns (or less).
Correct. The complexity is: RS485 is a balanced signal designed on 120 Ohms impedance symmetric cables, where both transmitter and receiver have correction on “common mode” interference and alien voltages.
The 2812b system has none of that. It is unbalanced, non specified impedance. Wich makes it much more prone to garble the 400ns “1” and 800ns “0”'s
Conclusion: use twisted pair, dont go beyond 10meters.
