Single WS2812b LED are connected over 35 meters (115 feets) UTP Cat. 5e cable and it's not working as expected. Getting wrong colors. Another WS2812b placed in parallel (not chained) on board close to MCU which drives them works as expected. 1st one was working fine also this way (i.e. in parallel) when it was connected via short wire.

Adding 74LVC1G125 buffers on both sides didn't helped.

Other pairs in cable are used to provide +12 Vcc for other devices & +5 Vcc to drive LED.

Wire resistance are only 1.9 Ohm, i.e. 3.8 Ohms two way. So I guess it might be the capacitance part which mess things up? Or could this be some sort of noise coming from +12V power nearby (it drives low power motor, but WS2812b doesn't works normally even when it's not working).

Any suggestions to try? I'm thinking about differential bus transceivers, like SN75176B for example. But probably there is simpler way keeping 1 wire only?

Update 16 July: Here is how oscillogram looks like on separate long 1 wired DATA. This is an output from MCU, at the MCU end:

Output from MCU

This is how it looks like at the wire end on the LED side without resistor to ground (but with in-series 75 Ohm resistor):

Other end

And here how it ends up when resistor to a ground added close to LED (as suggested in comments):

With "termination" resistor

  • 3
    \$\begingroup\$ Do you have a bypass capacitor at the WS2812b LED? Cat5 will have about 50pF per meter of capacitance, 35 meters is 1750 nF of capacitive load on data pin. Also, have you terminated the data pin to prevent reflections back to the cable? \$\endgroup\$ – Justme Jul 3 '20 at 8:26
  • \$\begingroup\$ @Justme do you mean capacitor between Vcc & GND of WS2812b? I have 1uF capacitor there. And I didn't terminated the data pin, not sure how to do this. Resistor in series? But which value it should be? \$\endgroup\$ – NStorm Jul 3 '20 at 8:42
  • 1
    \$\begingroup\$ I think you need the buffer to drive the cable. And make sure you drive a pair with data&ground twisted together. The cable has 100 ohms characteristic impedance, so it will look like a 100 ohm load for a step waveform. And at the LED end, it needs 100R termination from the data pin to the GND return pair to prevent reflections. Also make sure that any power supply and it's return are also a twisted pair, do not mix pairs. \$\endgroup\$ – Justme Jul 3 '20 at 9:36
  • 1
    \$\begingroup\$ OK. That describes a completely different scenario, where you don't have a 35 meter transmission line between the MCU and LED. Won't help you here. In fact the series resistor does not prevent reflections or terminate anything because it is in series with the data input. It only protects the LED input from getting destroyed by over/undervoltage transients that happen when the transmission line is not correctly driven or terminated. It will also protect from destroying the LED or MCU if LED is unpowered. Buffers didn't help likely because the transmission line was unterminated. \$\endgroup\$ – Justme Jul 3 '20 at 10:04
  • 1
    \$\begingroup\$ Yes, you must terminate the transmission line. It will help if you have an oscilloscope to see how the waveforms look like, to pinpoint the problem, whether it is driving the 35m cable at source, or terminating it at the far end. \$\endgroup\$ – Justme Jul 3 '20 at 10:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.