How to figure out needed capacity for backup power?

[mclien]

I have a RTC modul (microcrystal RV-3032-C7, App Manual attached).
Power consumption can be as low as 160nA/3.0V and I try to figure out what kind of backup capacitor I do need to keep it running for some days in the low power state, so it keeps time when the main power is not available.
Operating voltage is 1.3 to 5.5V,.
Main energy source will be a LiPoly single cell (that is min 3.0V for a usual protection circuit integrated in the cell IIRC).
Loading the cap is described Chapter 4.3/page61 in the doc. (as I understand it, that would be adjusting the according reg. right).

Chapter 8.3/8.4 pages142ff show the setup for rechargeable caps as backup (ceracharge is a fancy new TK tec only available for OEM atm).
And here is where my knowledge is way to thin, even though I tried to figure that part out with the help of online calculator and tutorials.

I do understand, that the cap reloads according to the Voltage/Resistor combination. But do I have to attach the right resistor to achieve the 160nA current or does the RTC "take" the low current having the right resistance?
By try and error with some only calculator, I get 14 days for a 70mF cap at 3V (applied voltage). But I'm not far from guessing with this .

I still assume, I might be way of with my limited understanding, though.

[RoGeorge]

I do understand, that the cap reloads according to the Voltage/Resistor combination. But do I have to attach the right resistor to achieve the 160nA current or does the RTC "take" the low current having the right resistance?
By try and error with some only calculator, I get 14 days for a 70mF cap at 3V (applied voltage). But I'm not far from guessing with this

The resistor in series with your 70mF, marked with (1) on that schematic at p142, has to be anything between 100 and 1000\$\Omega\$, and it is there only to protect in case of an accidental short-circuit.

The clock chip will draw from the capacitor whatever current it needs to keep the oscillator running.  It does the same for VDD, you apply a voltage, and the circuit will take only as little as it needs to do its job.  So you don't have to limit the current from outside to only 160nA.

As for the 70mF supercap charged at 3V, you have to consider only the difference between the 3V and the 1.3V (minimum V to keep the clock running), so the useful voltage on the supercap is between 3V and 1.3V (3V-1.3V=1.7V, 1.7V is the maximum drained voltage before reaching the minimum 1.3V where the clock dies).  Assuming the clock chip will constantly draw 160nA, by definition I=Q/t, and C=Q/V.

So Q=I*t, and also Q=C*V, so we have I*t=C*V, from which we can calculate the maximum time to empty a 70mF supercapacitor from 3V to 1.3V, under a constant current of 160nA:

t = C*V/I = 70mF*(3V-1.3V)/160nA = 70*10^-3F*1.7V/(160*10^-9)A = 70*1.7/160*10⁶ seconds = 0.74375 millions of seconds, about 206 hours.

Note:
- the 100..1000k protection resistor was neglected in the above time estimation, to keep it simple, because at 160nA the voltage drop on that resistor will be so small that it won't affect much the estimated backup run time
- !!! beware many supercaps have a maximum voltage, some may not support charging them up to 3V
- !!! beware that some supercapacitors might not allow to discharge them under a certain voltage (Li based supercaps are similar with Li-ion recheargeables), I didn't check if you TDK supercap have a minimum allowed voltage, probably they don't have this issue
- there was a note before that page, where they say the current might be 260nA or so (not sure for which power mode was that, I've accidentally spot that value at the description of the power backup modes, not sure where, or in which conditions the bigger consumption current was expect, I guess was for Vdd, but not sure)
- if the TDK supercapacitor allows higher than 3V, then the clock IC can charge it at a higher voltage than Vdd, it has an internal charge pump where the voltage is selectable from software
- typical Li-ion cells I have tested (for examples the ones from mobile phones) will have a circuit protection inside the battery, the shutdown voltage is about 2.4V, but it is recommended to not discharge a Li-ion under 2.8..3V or so for a longer battery life.
- !!! beware that most of the Li-ion or LiPo have an internal protection (for undervoltage/overvoltage/short-circuit/thermal protection) but not all commercialized Li recheargeable have it.  If the one you use doesn't have internal protection, you MUST add one, they can be bought ready made.  It's not to prolong the life of battery, is to protect against fire hazards in the first place.
- and, of course, I've didn't double check my calculations  (I hope my calculations are not totally out of range, but even if they are, the idea of how to estimate the time is correct)

[mariush]

Any reason you're not using a CR2032 battery?  You want to avoid that for shipping reasons (because it's lithium)?

You could also use rechargeable ML-1220 or VL-1220 batteries and put a resistor in series to limit the charge current

Or there's batteries that look like capacitors, but have lower voltage (2.4v), see for example https://www.digikey.com/en/products/detail/nichicon/SLB08115L1401PM/14005849  and https://www.digikey.com/en/products/detail/nichicon/SLB12400L1511CV/13590644

[mclien]

Thanks for the explanations.
Here is some background. I'm designing a tiny PCB as some sort of "universal arduino for wirst watches".
Main parts are: the mentioned RTC modul, a ATSMAD21E18 and a power latching circuit.
Most to all GPIOs will have connections at the edge of the board. Starting with a matching landing pattern it should be possible to design quite a range of daughter boards, which than make up the watch/ watchface itself.
So the main power source should be planned for that daughterboard and connected to main modul.
Th goal is, that most of the time, the RTC only is powered, but when activating (via the power latching) the MCU and the daughterboard is powered up, showing the time in whatever way it's designed (LEDs in my case). How the power ist cut can be decided via software (timebased or pushbutton).
To prevent to re-set the time every time the main power runs out, I'm currently trying to find the best backup power source. (Based on what I expect for me to be the longest time I might end up with no recharge source).

As for the cr 2032: Well I have reason to think. I manage to get the whole PCB as tiny as 15.5x15.5mm

My (possibly very wrong) thought was: If I do have a cap as backup, it is kind of long living and "self-organized" with the trickle charging of the RTC.

Exploring a bit more, I think I underestimated the difference in energy density between caps and batteries (and further more between rechargeable and non-rechargeables).

So perhaps a silver oxide battery with 4.8mm is the best choice.
Those have about 7mAh capacity down to 1.3V which is 20-30.000 hours runtime at 200nA, which is a few years in backup mode only..
(if I have not made a silly miscalculation in the 1000-steps).
And that space is still available in my design.

So perhaps I was completely on the wrong path with the caps idea?

[RoGeorge]

From the engineering standpoint, it makes sense to use a 7mAh battery at 200nA to get 35000 hours lifetime (though, in practice the shelf life of the battery might not be that long even at 0nA).  I've been told there are CR2032 format rechargeable batteries, too.  Maybe other rechargeable button size battery are available, too, but in a smaller format.

Li-ion also exists in miniature square formats, the smaller I have is 30mAh and it is roughly about 1x2cm, has a battery cell protection board included, looks something like this:  https://www.ufinebattery.com/products/3-7-v-30mah-lithium-ion-battery-370820/  Maybe there are even smaller secondary (rechargeable) batteries out there.

If I were to buy such a clock module, I would prefer to never worry about a primary (single use) battery.  Those sometimes leak with time.  And when they have to be replaced, you have to find another one of just the right format.  A rechargeable would be better.  Would give the peace of mind for no (or less) future problems.

Many super/ultra/whatever-named very big capacitors also use to have limited life, and they are expensive.  Maybe use a normal smd capacitor that can keep the module running only a few minutes, during the time a user change/plug/unplug the other boards.

If you want to keep the clock module running unplugged for longer than minutes, maybe make another "docking station" module, with just a Li-ion or a casual 2x1.5V AA batteries and nothing else, so to keep the clock module running "forever", while the module is waiting to be reused again (waiting plugged into its power "docking station").

Of course, these are only my personal preferences, can't say which version would be more appealing to others.

[mclien]

The scenario is a bit different. Both batteries are built in the enclosure with the Main (LiPoly) being recharged in a clock (via a magnetic connector).
So the Main battery might be cut of by built in the protection most likely at times where you won't be in reach of the dock in minutes. So 24h is the very least the backup should work.

Shelf life (as you suspected) realy is the bummer for the SilverOxyde Type. Datasheet says 80% left after 60 days...
One idea was this: https://www.nichicon.co.jp/_assets/pdf/products/slb/datasheet0307_2204_e.pdf (but might need an extra and special protection circuit).

Maybe the expensive one is the best solution:
https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/694/B73180A0101M062_DS.pdf
Even pessimistic approach works:
Page 4 Discharge: with 60% at 1.3V it is 60µAh capacity which equals 300h runtime, 12.5days.
That is if it is working to discharge with 0.002C instead of the lowest documented in the datasheet, which is 0.2C

But as this backup solution is mentioned in the datasheet of the RTC, it might be a good (while quite expensive) solution.

[PGPG]

To avoid "EU Battery directive" (don't remember exact symbol as just not use it) and not to worry about battery replacement ever, for many years I was using 0.22F, what with integrated in microcontroller RTC was giving about 3 days time-keeping (uC in RTC mode was not as effective as your RTC).
Few years ago, redesigning it I made two steps to get it longer:
- external RTC (may be 200-400nA),
- 1F capacitor.

I started from purchasing 10 such 1F capacitors, connecting them paralel and measuring what is going on with their self-discharge.
I wrote results down somewhere, but don't have it here at home.
What I remember:
- for the first time they self-discharged relatively quickly (it may be was about 3 days),
- for the second it was much longer - may be 7 days,
- then longer and longer.

[mclien]

Yeah, a soldered Battery is out of question. Plus those have quite bad self discharge.
And if it should happen it is shortened while building (which will happen, as i´the board is designed to be a starting point/ core for watch projects), a cap can be "just recharged", while a battery needs to be replaced.
Even a coin cell with a clip has the problem. As the space is tight it has to be removes over the connector pins plus the way might be blocked by the daughter board attached to the core board.

This leaves me with caps and I found these:
https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/6465/STDE-B-CPH3225A1AG-0020-2%20DK.pdf
3.2x2.5mm (1mm height), 3.3V, 11mF
I have room for 2 (1 top, 1 bottom side), connected in parallel should give me:
2V (3.3 - 1.3, highest voltage for the cap, lowest operating voltage RTC)
22mF (adding through parallel connecting)
Which should theoretically run 200nA for 60hours.

That should at least give a day or 2 till next charge possibility.