I am very happy i found this thread with the help of another member in another thread.
First, i have worked with hardware for many years now and i took an especially good look at Li-ion battery charging circuits because i had to work with them a lot in the past.
What i found out is that Li-ion batteries are more dangerous than other types and that they require a very very specific charging technique in order not to damage them or create fire.
What i found out about RTC boards like this one is that the ones with an LIR2032 battery (which is an Li-ion) have one of the poorest charging circuits for a battery of that type that i have EVER seen before. It's just a diode and resistor. This causes the battery to overcharge over time when the board is run at a voltage of about 4.5v or above. In any case, at 5v it definitely over charges as careful measurements have shown. This can cause fire or at least early failure of the back up battery, which results in no back up battery when the power goes out.
One single measurement may not be enough, there has to he several measurements done over time, several days, in order to see the rise in voltage of the battery. When the unit is first powered up at 5v, the battery may seem fine at 4v or even less, but allowed to run for several days in a row the battery voltage climbs higher and higher. My measurements had shown first 4.29 volts, then today (12/05/2015) it was up to 4.37 volts. The absolute max is specified as 4.225v for safe operation and of course battery life.
Solution? Some have been mentioned already, but i decided just to pull the series resistor which is 200 ohms and is surface mount. Using a hot air gun and directing the heat away from the main chip, the resistor was removed. it is easier and safer to remove the resistor than the diode on this board because the diode is much closer to the IC chip than the resistor, and also the connection can be made again in the future if you want to charge the battery again for some reason, for a short time period that is. I doubt you will ever need to charge the battery again however, unless you let it sit for a very long time without powering up.
You will note that the smart designs do not have a back up battery or else have the CR2032 non rechargeable battery with NO charge circuit.
BTW the Due board could run the charge circuit at 3.3v which would not over charge the battery.
Would the CR2032 battery overcharge when running at 3.3v if the board had a charge circuit?
Well the silicon diode used looks like a standard 1N4148 diode and that drops a nominal 0.65v when conducting, so the max charge voltage reaching the battery would be 3.3v minus 0.6v which equals 2.7v, so it may not hurt anything. The diodes voltage drops somewhat though for light currents, so it is possible that it might reach as high as 3v, and as the diode temperature is increased it could reach higher because that decreases the characteristic voltage more, so we could run into some charging of the non rechargeable battery. It'd doubtful this would happen, but we cant fully eliminate the possibility without careful testing. So in the end it is probably best to just desolder the resistor and remove it. After doing that there is NO way it can charge without some very abnormal board or chip failure like an accidental short (drop the board onto something metallic) or perhaps whisker growth. These would be more rare though.
Please pass this information along so other people know about this.