ESPixelstick 4.0 ESP32 data corruption from AF Feather at higher brightness

[perigalacticon]

I am running ESPixelstick 4.0 for the first time and I'm using an Adafruit Feather ESP32 module. I have found that the data on channels 3 and higher get corrupted with random patterns when I increase the brightness beyond about 25% when controlling directly from xLights. I am unable to use channel 1 because that is GPIO 2 which is not broken out on the Feather board. Channel 2 works correctly, tested up to 50 pixels. Higher channels do not work if the brightness is increased on any of them beyond about 25%. Has anyone ever had this problem and knows how to fix it?

Thanks.

PS the strands work fine on other controllers.

 

[merryoncherry]

5V or 12V? How are you powering it? Do you have a way of checking voltages?

The data issues are often related to voltages not maintaining proper proportion.

 

[perigalacticon]

5V or 12V? How are you powering it? Do you have a way of checking voltages?

The data issues are often related to voltages not maintaining proper proportion.

Thanks. I have tested both 5V and 12V strands with different power supplies, it doesn't depend on what the lights are or how they're powered. The supply voltages don't dip more than a tenth of a volt. The Feather is powered from a USB port on the PC, a powered USB hub I have a wire connecting the grounds.. I'll try different supplies for the Feather.

 

[merryoncherry]

It is strange that it is related to brightness if it isn't higher current causing more voltage drop causing data problems.

Data amps can be a good friend but this experiment is so small with so many things varied maybe not to reach for them just yet.

 

[perigalacticon]

Yes I agree, I also tried using a SN74AHCT125 level shifter which I have good success with.

 

[TerryK]

It does sound as if power injection or lack thereof might be a problem here. However, I suspect that data line logic levels may be the or part of the problem instead. AdaFruit has several Feathers available and although I haven't worked with them I suspect the Feather's logiv outputs are 3.3 Volt and not the 5.0 Volt that just about all pixels expect.
The increasing brightness levels may not show a drop in voltage but it will increase the pulsation on the V+/V- wires as the RGB LEDs switch On and Off. More of a hunch than anything but I am inclined to think that the increased electrical noise is causing the 1st pixel in the string to not be able to correctly see the 3.3 Volt level data coming from the controller.

I would suggest inserting a level shifter between the controler and pixel strings. Too, there are several Forum Threads discussing logic level shifters. Some ACL members appear to have good success driving pixels with 3.3 Volt logic devices, other struggle.

 

[TerryK]

Threads crossed. If a level shifter hasn't helped I'd revert back to power injection. Beyond that configuration or firmware/software. Can you post a wiring diagram sketch.

By the way, which Feather?

 

[perigalacticon]

Yes I just mentioned the level shifter. So Far I have tried 3 different types of leds. 1 - a 5V WS2812b strip of 7 nodes. 2 - a 12V WS2811 pixel-node type strand of 50 nodes. 3 - a single 5V 3W high powered WS2811 node. I have tried 3 different power supplies, these are 5V/4A, 12V/8A, 12V 15A. No matter which leds I connect to the data output of channels 3 or higher, the display is corrupted if the brightness on any of the channels is increased above ~25%. I tried all combinations of leds and power supplies and with and without the SN74AHCT125 level shifter (which I used extensively with ESP8266 boards). It's possible I have a bad ESP32, just wondering if anyone else has had problems with the Feather ESP32 running ESPixelstick 4.0.

 

[perigalacticon]

Adafruit HUZZAH32 – ESP32 Feather Board:

Adafruit HUZZAH32 – ESP32 Feather Board

Aww yeah, it's the Feather you have been waiting for! The HUZZAH32 is our ESP32-based Feather, made with the official WROOM32 module. We packed everything you love about Feathers: built ...

www.adafruit.com

 

[perigalacticon]

BTW, the corruption does not depend on how bright the channel 2 leds are, I can put the the 50 node strand on ch2 at full brightness (no glitching) or no brightness and this doesn't affect the corruption on the higher channels, so I'm pretty sure it's not the power.

 

[Notenoughlights]

Do you have any form of resistor on the data line leading to the first pixel?

 

[TerryK]

I've read through this thread several times; think I'm going to return to the configuration or firmware/software as the cause. If the problem existed in xLights, due to it being so popular if a problem existed there it should have been found by now. So back to ESPixelStick4.0 and the Feather.

Feathers aren't discussed much on ACL so the 'knowledgebase' for them will be rather small. If you are using EXPixelStick4.0 on a Feather for the first time and from your posts it seems you have a number of ESP8266 devices; what happens if you run EXPixelStick4.0 on one of them? Also, I noticed that ESPixelStick 4.0.0 was in Beta not too long past. Are you running a stable release?

@Martin Mueller is another ACL member that appears to use a number of ESP8266 devices. Since I 'tagged' his username hopefully he'll notice the thread.