Low CIO Quality - Display Unresponsive - S100102

[Syvion]

Hey All,

I've been trying to get this fantastic project to work for my Hollywood LayZSpa.
I've explored the WEMOS D1 Mini approach and ran into low quality signal issues, so I ended up trying and going the recommended approach and even got a full kit that I assembled from metaljay.

The main problem seems to be a low CIO message quality between the ESP and the CIO.
No matter what I do, or which chip (D1 mini or ESP8266) I use, I can't get the touch inputs on the Display to work, or make reliable changes through the web portal.

Lay-Z-Spa.Hollywood.Videocapture.mp4

Tub: LayZSpa Hollywood (6 Wire, Airjets, 2022)

So far I have tried:
Code:

• The main branch code, and the development branch code.
• Tried each with and without the filter spikes flag. (Turning filter spikes off is definitely a lot worse and causes a flickering display)

Hardware:

• D1 Mini & ESP8266
• With and without Capacitors
• With and without resistors
• With and without LLC. I also tested the input/outputs of all LLC channels with a multimeter, and checked all resistors and all channels correctly convert 5V to 3.3V.
• I even have 2 of the S100102 pumps, (I got a new one under warranty after the water backflow seal failed on the air pump) I cannot get the module to work with either pump.
• Using the default wire length for the connectors, and then cut them drastically down to see if it improves quality.
• Ran the HWTest, passes without errors

Debug info here: in zip file is a log capture and an Inputs capture
Lay-Z-Spa_Hollywood_DebugInfo.zip

Let me know if anyone has any recommendations or other things to try.
I don't mind making code changes if needed.

[visualapproach]

The inputlog seems fine, but it will only see what the CIO is sending. The display doesn't send anything, it just responds to "requests".
The outputlog says that only 60% of the messages from CIO is legit. My guess is that it should work anyway, but obviously it's not.
Can you turn on/off bubbles from Web UI reliably, even if it takes up to 5 seconds for it to react? (The ESP simulates the button press until it sees the requested change, OR 5 seconds has gone by.

[Syvion]

Hey, sorry for the late reply. For some reason I didn't get a notification that you got back to me.
I've (just) now installed the board again into the pump I'm actively using.

Based on the log, the quality is the same (+-60%), but on this pump some of the buttons on the display work kinda ok.
Toggling the "Pump/Heat/Bubbles" works although it usually with about a 3 to 4 second delay. (Using the dashboard for those functions is slightly more responsive)
The temperature however is almost impossible to reliably change on the display, but when using the web dashboard I can "set" the temperature and it'll eventually adjust to the desired temp.
It does the whole "massively overshooting/undershooting" then slowly coming down/up to the requested target temperature. As long as I let it do it's thing and don't mess with any other buttons it "works".

Since I mainly want to control the temperature via a schedule through the dashboard, it's not the worst thing that those buttons don't work on the display. As long as I can toggle the bubbles when I'm in it, that's kinda enough.

I'll keep the board running and attached and monitor how it goes.
I'm puzzled how these mass produced pumps seem to so widely differ in how well they work with this ESP solution, and how the display chip itself has no issues in getting a good signal in/out.