Difference in accuracy between light sleep and deep sleep modes

[daniel1704]

I'm working on a project that will require to keep the ESP32 on an idle state for about 5 mins and then start again. As the project will be running with a battery, I decided I would like to use power saving options. When I run the code using light sleep, the accuracy of the sleep time is pretty good, and everything works fine. However, when I tried to use deep sleep, keeping every other configuration the same, the sleep time started to lag for about 10 s. Do any of you know why is this happening?

[WiFive]

Light sleep can resume but deep sleep has to boot the whole app again. Are you taking that into account?

[daniel1704]

Light sleep can resume but deep sleep has to boot the whole app again. Are you taking that into account?

I did, I measured the time from sleep start to reboot when I was using deep sleep. Even if I didn't, boot time with the configurations I'm using is about 1 s, and the difference between both sleeps modes is around 10 s, so it wouldn't be justified

[sargun1]

IMO if you are putting the ESP32 idle for 5 mins, it will be better to use Deep sleep mode, because there is significant current consumption difference while in Light sleep mode and deep sleep mode. according to the table below:
table source: https://iotdesignpro.com/projects/vario ... sleep-mode

It will take approx 300ms for bootloader to reload when ESP32 is restarted, so 300ms in nothing compared to 5min idle state, so better use Deep sleep mode.