Super capacitor to replace RTC coin battery?

I am considering replacing the coin battery in my shield design with a super capacitor. I wonder if someone else is doing this already. Any comments/suggestions?

I'm using DS1307. It has a maximal I_battery of 500nA, maximal V_batt 3.5V. I intend to use the Arduino 3.3V supply to charge it to 3.3V. The minimal V_batt is 2V.

I guess using some math I'm looking at some time t that is acceptable for temporarily removing the power for no more than a day or so and then reattaching power.

I found that with a 0.22F supercap, it lasts 6.6 days. Am I too optimistic?

Supercaps work great. The only drawbacks are a) slightly higher cost, and b) they take some time to first charge up before they can provide backup power. Whatever resistance is in the path of the cap (including internal self-resistance) forms a time constant that can be non-negligible before the cap is fully charged.

Also, in computing your 6.6 days be sure to take into account capacitor self-discharge. See this Panasonic app note for example:

http://industrial.panasonic.com/www-data/pdf/ABC0000/ABC0000TE5.pdf

--
The Rugged Audio Shield: Line In, Mic In, Headphone Out, microSD socket, potentiometer, play/record WAV files

6+ days to remove & reattach power seems like plenty.
skyjumper has been using a supercap with boost regulator to provide time to complete data writes to an SD card. Designed his card to separate out the parts that really needed power for the writing, let the rest go unpowered.
Has some diodes to keep it from backfeeding into the charging circuit, and a resistor is needed to current surge at initial power-on. Resistor also slows charge up time, your tradeoff to make there.
Make sure you get low ESR super-caps.

Thanks guys! I see that arduino UNO's 3.3V supply has 50mA maximal current rating. So If I use a 75 ohm resistor to limit the current then it takes about 50 seconds to charge up to 95% (3.135V). Then I will leave it connected to the capacitor as long as the arduino power is on. I'm not sure about feeding current back to the supply when the arduino is powered off. Should I add a diode? But then I take away 0.6V from the supply voltage and can only charge up to 2.7V. Then the capacitor will also last half as long (3 days, still long enough for me).

CrossRoads, Is there some diodes with less turn on voltage for this type of work?

RuggedCircuits, is the self-discharging due to air?

So about ESR, is that some sort of resistance like internal resistance of a battery? I see this one for $1 a piece with 75 ohm ESR, maybe good enough to directly hook up to the 3.3V supply with initial current consumption still under control?

http://www.ebay.com/itm/10Pcs-5-5V-0-22F-30-H-Type-Coin-Button-Super-Capacitor-/270973653764?pt=LH_DefaultDomain_0&hash=item3f1749d304

Using this will double my cost but it seems small and reusable.

There are schottky diodes with a drop of 0.3 V or so.
You can also regulate to something like 4V using an adjustable LDO regulator, and use a regular diodes for the 0.5V drop.

Thanks jwatt. I'll consider a simple schottky diode then.

That high ESR will also limit the current flow out of the cap.

Diodes like this
DFLS130LDICT-ND

0.31V drop.

RuggedCircuits, is the self-discharging due to air?

No, it's due to internal leakage.

The diode thing is interesting -- remember that it's a ~0.7V drop (~0.3V for a Schottky) as long as the diode is "sufficiently biased" with current, but as the capacitor nears full charge, it draws less and less current until the current is so low the diode voltage begins to approach 0! So I wouldn't worry about that too much.

A high ESR is actually a good thing for a battery-backup application. It limits inrush current (really, you do NOT want to load down your circuit with charging a 0.22F capacitance through nearly-0 ohms!) and the current draw out of the cap when it's backing up the RTC is so tiny that any series resistance causes negligible voltage drop.

--
The Gadget Shield: accelerometer, RGB LED, IR transmit/receive, speaker, microphone, light sensor, potentiometer, pushbuttons

CrossRoads:
That high ESR will also limit the current flow out of the cap.

Diodes like this
DFLS130LDICT-ND
http://search.digikey.com/us/en/products/DFLS130L-7/DFLS130LDICT-ND/673203
0.31V drop.

Thanks!

RuggedCircuits, I guess that's right! 0V across a diode that is turned off (very little current). I'll have to do some testing then. I have one super cap on a circuit board I bought as scrap. Will remove that for the test.

Sony made a great miniature radio (sony icf-sw100) whose super capacitor failed after a while and needed replacement.
That should tell you something about using capacitors for your clock.
I think it is much easier for the average customer to change a battery than to replace the capacitor.
Plus, some customers may find unacceptable to set the clock every time they come back from one-week vacations.

How many people lose power for a week tho?

Some people turn off their appliances when they go away (I am one of them).

That's a good point, hadn't considered that.

Some RTCs have built-in battery charging circuits, perhaps a super capacitor can be substituted.

I saw one suggestion to use a DS1302 because it has a charging circuit already in the RTC. I don't know how if differs in power consumption or if you might be able to charge the cap to a higher voltage and give you a long power off time.

I know a coin supercapacity.I hope it can help you.

http://www.kingwellbattery.com/products/coin-super-capacitor-c-v-h-series-ultracapacitor.htm