Any signalling gurus here?
I’m using cheap 8-core alarm cable (solid core) - 4 data lines and 4 grounds - to drive WS2812B strips. The cable is reasonably long (3m?)
Without a series resistor I’m seeing some quite nasty voltage spikes (and dips below zero), but the LEDs work fine (for now!). However with a resistor the wave is quite sawtooth - shaped and I think some short pulses are failing to signal the chip as their peak is too low.
With my armchair understanding of signalling I assume this is to do with capacitance in my cheap cable.
I have some spare Cat 6A FTP cable and am wondering if using this might help?
I’m also going to have a play with timings and optimising my resistors but wondered about the physics of it!
Is your alarm cable twisted pairs?
If you try out the CAT6 cable, the recommendation is to use one signal and a ground per twisted pair.
Cheers - that’d be the plan, though I’m not relishing resoldering everything if I end up switching! Those strips are a pain. Though I’m gradually convincing myself that it might be worth it - the alarm cable isn’t twisted pair, just a vague coil of 8 cores in a sheath.
For now I’ve dropped down to 75R resistors - some of the short pulses still look very triangular and don’t reach up to 5V but they seem capable of triggering the LEDs, which are working fine (270R was a mess).
I think I’ll increase the duration of a short pulse slightly (I dropped it down to 250ns for the WS2812B timings, but as far as I can see, the only critical point is in the 550-650ns region so long as the total cycle length is correct).
CAT5 & CAT6 cable with 100 ohm resistors works very well.
@PaulStoffregen Thanks - how long a run have you tried, and have you had an oscilloscope on it? Just wondering if the sawtooth effect is any better than with my alarm cable. Certainly I’m finding I need to use smaller resistors than the ~300R often quoted, which is a shame as I’m still seeing overshoot. Hopefully fiddling with the timings will make things a bit more resistant!
On this recent project (scroll to the end for several photos of the art installation) we ran the LED signal about 25 feet.
https://www.pjrc.com/pilot-light-flame-sensor-for-burning-man-art/
Worked fine when the battery charger wasn’t connected to the system. But when the charger was running, it would interfere with the signals running on those 2 long wires. Next week we’re going to replace those with RS422 drivers & receivers for more noise immunity.
To be a little more specific, this board was installed inside one of those art pieces. Only 3 of its 8 outputs were used. The first went to the LEDs in that piece. The others went down ~25 feet cables to the other two.
For anyone digging in here, the Cat 6A unsurprisingly made the signals a lot prettier. Probably mostly the foil around each pair as I was getting a lot of crosstalk (particularly when one line was pulsing 1 while the other 3 pulsed 0). Unfortunately it didn’t solve my problem which was related to one dodgy strip and the fact that clockless_block_esp8266.h only sends the latching signal if it’s been interrupted while sending data. (see another post in here!)
Just adding to the convo here…I was using 18/4 speaker wire for jumpers (data and power) and had to insert repeater LEDs every ~10 feet for a ~35 foot jumper cable, and it was still flaky. I just tested 100’ of Cat 5e using 2 solid colored wires tied together for data and their striped counterparts for ground, and it worked! This was “in the lab” so we’ll see how it performs in real life, but very encouraging.