[Solved] How to drive the two inputs (GPIO23 and GPIO22) on the ESP32 with an external signal?

I need to drive an external signal to two inputs on the ESP32 (specifically, I assumed the use of GPIO23 and GPIO22), and as these are two floats drowned in tanks of water, I found that I would use galvanic separation using a two-channel module with PC817 optoisolators. Generally on the 'short' it seemingly works. For testing purposes, I apply voltage (3.7V) from the 18650 battery to the input of the optoisolator module (the target is to have 5V through these floats - they act as a jumper depending on the position) and the output connected directly to the GPIO on the ESP32 and of course to ESP ground. Generally works.... until I put it together on a prototype board, wired, added connectors etc.... Then it turned out that one GPIO works correctly (we power up the optocoupler - the state changes) oh the other one triggers both.... That is, applying power to the second optocoupler causes both GPIOs to change state.... The same happens without the optocoupler if I short-circuit the GPIO directly to ground (briefly if I short-circuit only the pins on the ESP itself this does not happen). I added 10k pull-up resistors between GPIO and 3.3V and GPIOs shorted to ground work fine, the state only changes when the GPIO data is shorted to ground. The problem is that they have stopped responding if we feed a signal through the optocoupler. Applying power to the optocoupler input changes absolutely nothing about the state of the GPIO input. What have I done wrong?

Tailor

Hello,
One problem is the lack of any schematic.

krawietz wrote:

The same thing happens without an optocoupler if I short the GPIO directly to ground (briefly if I short only the pins on the ESP itself this does not happen).

This sentence is also a problem at first glance.

Greetings.

Target diagram below. I will actually check the power supply, as the target is supposed to be 5V.
.
Of course the yellow jumpers removed so that the input side ground is separate from the ESP ground.

I also have 24V AC available in the circuit, supposedly some documentation for the module says that 3.3 - 24V can be applied to the input - there is a 3k resistor in series and a signalling LED. There is also a 3k resistor in series between outputs V1 and V2 and the collector.
Another thing is that the possibility of using 24V AC would simplify the whole circuit and eliminate an additional power supply.

Krawietz

Added after 10 [minutes]:

mpier wrote:

krawietz wrote:

The same thing happens without the optocoupler if I short the GPIO directly to ground (briefly if I short only the pins on the ESP itself this doesn't happen).

The same thing happens without the optocoupler if I short the GPIO directly to ground (briefly if I short only the pins on the ESP itself this doesn't happen).
This sentence is also a problem at first glance.