Well, back to the drawing board
The production project as I mentioned is supposed to be a 16 row, 300 columns, where each row of 300 LEDs is driven by a separate channel output, data send to each of the 16 rows at the same time.
Signaling worked with a 330 ohm resistor in series with the channel data lines.
However the production display will be a distance away from the controller so I obtained a 10ft, 20 wire 24Ga cable where 16 of those wires would be the data for each of the 16 channels/rows, the remaining 4 wires for power/ground from the display's power to the controller package.
The cable harness has the 330 ohm resistors integrated on the display side of the cable.
Well, that brings back instability, especially on channels 4,5,12,13. On random shows, some rows will blink, sometimes show bad data across the entire row. Error occurs every few seconds. It's almost tolerable.
What I really found strange was the problem seeming more pronounced on the wires for channels 4,5,12,13.
I checked resistance and even capacitance down the cable for those wires, no different than the others.
I did find that WS2812B's input impedance is quite high, where if I don't have a channel connected to the level shifter, it can pick up signaling from one of the other channels.
Scoping the signal, there is ringing during the '1' (~5v) preamble, but doesn't seem different between the more stable but still problematic channels and the more problematic ones.
But then since the problem occurs randomly, it is hard to catch the signal during a glitch.
Seems the WS2812B input can trigger on signals as low as 2v and with the high impedance, doesn't take much current to produce it.
It is a shame too, as the cable is a sturdy quality, 24Ga stranded tinned copper. It was easy to work with during soldering assembly.
But it is straight through wires, so what can I expect from it in handling what comes to the equivalent of 2.5-3.5Mhz, depending on the pulse widths?
I did try reducing the effective input impedance by adding a 1K resistor on the display side from data line to ground. Didn't really help with stability response to the data signals, but did help reject accepting stray signals from neighboring data lines.
Any ideas?
I'll have to see if using a 10ft ribbon cable is any better,
The production project as I mentioned is supposed to be a 16 row, 300 columns, where each row of 300 LEDs is driven by a separate channel output, data send to each of the 16 rows at the same time.
Signaling worked with a 330 ohm resistor in series with the channel data lines.
However the production display will be a distance away from the controller so I obtained a 10ft, 20 wire 24Ga cable where 16 of those wires would be the data for each of the 16 channels/rows, the remaining 4 wires for power/ground from the display's power to the controller package.
The cable harness has the 330 ohm resistors integrated on the display side of the cable.
Well, that brings back instability, especially on channels 4,5,12,13. On random shows, some rows will blink, sometimes show bad data across the entire row. Error occurs every few seconds. It's almost tolerable.
What I really found strange was the problem seeming more pronounced on the wires for channels 4,5,12,13.
I checked resistance and even capacitance down the cable for those wires, no different than the others.
I did find that WS2812B's input impedance is quite high, where if I don't have a channel connected to the level shifter, it can pick up signaling from one of the other channels.
Scoping the signal, there is ringing during the '1' (~5v) preamble, but doesn't seem different between the more stable but still problematic channels and the more problematic ones.
But then since the problem occurs randomly, it is hard to catch the signal during a glitch.
Seems the WS2812B input can trigger on signals as low as 2v and with the high impedance, doesn't take much current to produce it.
It is a shame too, as the cable is a sturdy quality, 24Ga stranded tinned copper. It was easy to work with during soldering assembly.
But it is straight through wires, so what can I expect from it in handling what comes to the equivalent of 2.5-3.5Mhz, depending on the pulse widths?
I did try reducing the effective input impedance by adding a 1K resistor on the display side from data line to ground. Didn't really help with stability response to the data signals, but did help reject accepting stray signals from neighboring data lines.
Any ideas?
I'll have to see if using a 10ft ribbon cable is any better,
Statistics: Posted by DanMan32 — Sun Jan 28, 2024 10:22 pm
