i have the A7608 board, which is quite similar to what you have.
I do not know, but it would be best to implement some logic where if the battery is below a threshold you keep deep sleeping.
That battery also costs a lot of money. You might want to first try a standard 18650, e.g. with 2-3 Ah and see how it goes. Connecting to wifi every 15 minutes and using proper deep sleep gives me a month of battery time without any charge.
I think the open circuit voltage of that panel is to high. The board has a CN3065 charging chip (search the datasheet online), and accept up to 6.5V max. Search a "5V" solar panel. Again, I do not know the specifics of your application, but a 1w solar panel can do miracles in many applications.
this will depend on the cpu frequency you set for the esp32. At 240MHz, my board sucks 40 mA, but at 40 or 80 MHz you can go down to 10 mA. For battery powered applications you need to learn to deep sleep as much as possible.
not sure, but I would suspect live streaming 24/7 requires a different thinking for the power system
the antenna can stay in the enclosure you use for the board
i use standard abs or polycarbonate enclosure boxes. There are tons of models available for ~10$
@luki743 LilyGO T-Display S3, easy;
T-Embed not so much.
The T-Embed repository is here: https://github.com/Xinyuan-LilyGO/T-Embed
You only need one GPIO and gnd if the sn65hvd230 is powered by the other device.
CAN takes a single digital output (GPIO) from the ESP and converts it to "differential" + signal, - signal.
The "Grove" connector maybe configured as CAN and maybe a few spare GPIO on that unpopulated development pad on the lower left of the image.
a separate power supply rated for the pump
a small relay or mosfet to switch the pump to help isolate the pump motor from the ESP32 circuitry
all grounds connected to a "common"
no power interconnections, I.e. power isolated between LilyGO and pump power supply