2811 pixel tester wrong timing

[ajp]

I have 2 strings of WS2812B LEDs purchased some time ago. I am using a 2811 pixel tester (it is marked with SP002E). One strip works fine, the other randomly flickers.
When I look at the Data stream on a CRO, the tester '0' timing (T0H, T0L) is OK, but the '1' timing is wrong (T1L is 1.1uS, should be .45, T1H is OK).
Can't see reset pulse.
Is anyone aware of issues with the small/cheap pixel string testers?

 

[Skymaster]

There are two pixel timings, an older one at 400kHz and one at 800kHz. It's possible that this is doing a mish mash of the two timings.

 

[Kent]

So, not answering your question, but incase you haven't considered it, ESP32 (e.g. Quinled Digi Uno) with WLED or ESPixelStick is my preferred tester of choice, and has the added bonus of testing props in xLights in addition to running standalone.

 

[David_AVD]

Also check that you are feeding data into the correct end of the strip.

 

[ajp]

After looking at some cro traces, it seems the tester is putting a 10uS pause between sets of 24bits, and some pixels in the strip are seeing this as a reset. Cro channel 1 shows tester output, input to pixel 1. Channel 2 shows output pixel 1. Tester is putting a 10uS pause after every 48bits.
I will look at a different strip tester, but I guess if the strip doesn't work with the cheap tester, it is unlikely to work with Falcon. I will try the falcon test mode.

 

[Skymaster]

The 2811 spec requires a 50uS reset frame to be sent. Some chips will latch the data at a lower timing interval, but controllers should be putting out the minimum time.
@keithsw1111 can advise the reset timing that's used on the Falcon v4 and v5 boards. I suspect it's in line with the data sheet.

That said, there should not be a 10uS delay between every 3 channels of data. That suspiciously sounds like the pixel tester is not that crash hot.
Each data bits should be set in 2.5uS@400kHz, or 1.25uS@800kHz (with different High/Low timings for 1 and 0); as a continuous stream; with minimal delay between them; there's nothing in the spec that details 8 bit words or 24 bit frames, other than that's what each chip will read and output to its PWM drivers.

 

[ajp]

As a follow up, FPP works great with the 'dodgy' LED string