- Tue May 12, 2020 4:17 pm
#87069
Do the temperature and humidity go missing in pairs? If so then that would tend to indicate a glitch in getting wifi connectivity on that cycle. If only one goes missing then it would tend to indicate a glitch in just transmitting one reading.
Also if you are interested in driving battery usage as low as possible then you could consider using a slightly smarter reporting strategy. The overall consumption is being largely determined by the active current even during the few seconds every 10 minutes as the active wifi current (say 80 mA) is much higher than the sleep current. When you have slowly varying readings like temperature and humidity then one can measure and compare with the last reported values (stored say in EEPROM). Only if they differ by more than a significant amount do you then activate wifi, report the measure, and store the measure. These type of sensors have a fair bit of noise in the readings so could for example say that only if the temperature had moved by 0.5C was that meaningful. When I use this strategy I also back this up by forcing a report if one hasn't been done for several cycles (also needs a counter stored). So, for example I might only report if either a significant change has happened or if an hour has elapsed since last report. As turning on the wifi increases the current a lot and increases the duration of the active period, this can have a significant impact on battery drain.
Also if you are interested in driving battery usage as low as possible then you could consider using a slightly smarter reporting strategy. The overall consumption is being largely determined by the active current even during the few seconds every 10 minutes as the active wifi current (say 80 mA) is much higher than the sleep current. When you have slowly varying readings like temperature and humidity then one can measure and compare with the last reported values (stored say in EEPROM). Only if they differ by more than a significant amount do you then activate wifi, report the measure, and store the measure. These type of sensors have a fair bit of noise in the readings so could for example say that only if the temperature had moved by 0.5C was that meaningful. When I use this strategy I also back this up by forcing a report if one hasn't been done for several cycles (also needs a counter stored). So, for example I might only report if either a significant change has happened or if an hour has elapsed since last report. As turning on the wifi increases the current a lot and increases the duration of the active period, this can have a significant impact on battery drain.
