search

pvvx / ATC_MiThermometer

Custom firmware for the Xiaomi Thermometers and Telink Flasher
https://github.com/pvvx/pvvx.github.io/tree/master/ATC_MiThermometer
Other
2.98k stars 207 forks source link

Different ADC values for advertisements #38

Closed MartMet closed 3 years ago

MartMet commented 3 years ago

Hi,

thanks for this great project. I really appreciate your work.

I use the LYWSD03MMC with firmware 1.8 and advertising typ "Custom".

I try to power it from a very small solar panel in combination with a super cap. So far it works reliable down to ~2V. I expierenced that the ADC values from the advertisments and the ADC values from connection mode differ. For example:

Connection Mode (via TeLinkMiFlasher.html): 2296mV (this one seems close to the real value) Advert. Mode: 2058mV

Is there any difference in the ADC measurement for this two modes?

I also noticed (running with CR2032 battery) that in advertising mode the battery level indicator byte is sometimes over 100.

pvvx commented 3 years ago

I also noticed (running with CR2032 battery) that in advertising mode the battery level indicator byte is sometimes over 100.

? ''' if(battery_level>100)battery_level=100; ''' https://github.com/pvvx/ATC_MiThermometer/blob/master/src/battery.c#L82-L83

Connection Mode (via TeLinkMiFlasher.html): 2296mV (this one seems close to the real value) Advert. Mode: 2058mV

The device operates in pulse mode. When connected, a pulse with one TX-RF - transmission is carried out via one channel. With advertising uses of 3 pulses TX-RF for 3 channels. More costs. If the internal resistance of the voltage source is large, then during the measurement, the voltage source sags.

pvvx commented 3 years ago

I try to power it from a very small solar panel in combination with a super cap. So far it works reliable down to ~2V. I expierenced that the ADC values from the advertisments and the ADC values from connection mode differ.

LCD contrast drops dramatically below 2.5 V. CJMCU-2557 BQ25570 ? image

MartMet commented 3 years ago

Thanks for your fast reply.

? ''' if(battery_level>100)battery_level=100; ''' https://github.com/pvvx/ATC_MiThermometer/blob/master/src/battery.c#L82-L83

Yes, thank you, ive already seen that.

Device (new): a4:c1:38:93:22:66 (public), -81 dBm Complete Local Name: 'ATC_932266' 16b Service Data: <1a1866229338c1a4c3068015770b51340d> --> 1a1866229338c1a4c3068015770b51340d So 0x77 should be the battery indicator, or am i addressing the wrong byte?

Connection Mode (via TeLinkMiFlasher.html): 2296mV (this one seems close to the real value) Advert. Mode: 2058mV

The device operates in pulse mode. When connected, a pulse with one TX-RF - transmission is carried out via one channel. With advertising uses of 3 pulses TX-RF for 3 channels. More costs. If the internal resistance of the voltage source is large, then during the measurement, the voltage source sags.

Ah ok i see. Would it help to delay the ADC Measurement?

MartMet commented 3 years ago

LCD contrast drops dramatically below 2.5 V. CJMCU-2557 BQ25570 ? image

Cool converter. :) Actually i still can read the display, what matters most to me is, that BLE is still running.

MartMet commented 3 years ago

--> 1a1866229338c1a4c3068015770b51340d So 0x77 should be the battery indicator, or am i addressing the wrong byte?

In fact i was addressing the wrong byte, the correct byte should be 0x51. And the voltage bytes are 0x0b77. (As the values looked reasonable i even was wondering why the voltage endianess was reversed :D)

Thanks for your fast response. :) I created a PR to fix the doc.