I'm working on a battery-powered DSLR camera controller with a MCU (ATMEGA328P) that spends most of the time in deep sleep and a RTC (DS3231) that wakes up the MCU a few times a day to take a picture. The system currently has two Lithium-Ion 18650 in series as power source (because I need 7.4v for the camera). The DS3231 is embedded in a Chronodot 2.0 module, which includes a coin cell battery holder (for a CR1225 battery). My sleep current consumption now is 120uA, of which 20uA is due to the MCU, and 100uA due to the RTC. This low standby current end up using about half of the battery capacity.
Therefore, I want to reduce the RTC power consumption to extend battery life. For this, I know I can supply power to the DS3231 through the BAT pin, which significantly reduces power consumption (from 100uA to a few uA). My system has a regulated 3.3v rail (with a MCP1702-3302E), which I can connect to the RTC BAT pin (which can take between 2.3 and 5.5V).
However, if I keep the coin cell battery in place, I fear that keeping it in parallel with the 3.3v rail can damage it.
On the other hand, if I remove the battery, the system will lose time every time the main battery is replaced, which is not desirable.
Therefore I've considered a few possible paths to a solution:
1.- Find a coin cell battery that can tolerate being in parallel with a 3.3v source. However, I haven't found such information in the datasheets of non-rechargable coin cells.
2.- Use a rechargeable coin cell such as ML1220. However, it requires charging at a maximum of 3.1+-0.15V, so 3.3V would be out of the range. To reduce the voltage, I thought of a voltage divider, but to have one capable of providing the 200uA the RTC needs, it would introduce a constant current loss that would be worst than my actual setup. Another option is to add a diode with a voltage drop of 0.2V, but after looking in the largest parts suppliers, I can't find a through-hole diode with less than ~0.3V. This could be acceptable, but the battery would never be fully charged (not a big deal). I've looked into other chemistries like Li-ion and LiFePo4 coin cells, but they all seem to come in larger packages (20mm diameter and more), and I need a small one (12mm diameter, like a CR1220 or CR1225).
3.- Remove the coin cell battery and add a 1000uF capacitor to keep the RTC running for a few minutes when the main battery is removed. The downside of this option is the large size of such capacitor. This would require increasing the size of the case. I would rather keep the device smaller.
4.- Remove the battery holder from the RTC module and put a standard CR1225 battery outside of the module and use a diode-OR to get the DS3231 powered from the 3.3V rail when powered up and from the battery otherwise. But in this case, a 0.35V drop from the diode would make the battery usable only down to 2.65V (as the RTC needs 2.3V minimum), which would waste about 20% of the battery capacity (which is not such a big deal either). If taking this approach, I could also use ideal diodes or an IC such as ICL7673. But I would rather avoid such extra components due to price and space.
So, my questions are:
Is there a better solution I haven't considered yet?
Am I overlooking some problems/advantages of these options?
Or am I just worrying too much about minor voltage differences, and I can use standard coin cells (3V) or common rechargeable ones (3.1V), and they will be just fine?
