Small supercapacitor as 3V backup

[nibblie]

I'm building a device that is normally powered from a 3V battery, but I want the device to keep running for some seconds if the battery is suddenly removed. The current draw is quite small, about 1 mA (including a 50ms 10mA peak). The solution I've come up with is to use a supercapacitor. I must detect when the battery is removed, but I'm not sure how to do that in an efficient way. This is what I have right now, but a diode is not really acceptable here because of the voltage drop.

1. Is there a way to implement the detection more power efficiently?

This is the supercap I'm considering:

https://media.digikey.com/pdf/Data%20Sheets/Seiko%20Instruments%20PDFs/CPH3225A_CP3225A.pdf

No info about leakage current, but I'm guessing (hoping) <1uA after it's been charged up, since more than 4uA is not acceptable.

2. Is it OK to connect a supercapacitor directly like this? Or is a charging circuit required?

(This is just a sketch, so no bypass cap, and unconnected pins...)

 

[KlausST]

Hi

I recommend to read about
* "ideal diode"
* microcontroller supervisory circuit with backup supply
* backup battery controller

Klaus

 

[danadakk]

I would pick a supercap manufacturer that fully specifies its performance over
T and V and self discharge. Wanting a part to perform a specific way is not
the same as a part fully characterized, and speced in a datasheet. Part you have
picked is about 4 - 6 pages short of specs.

You have to do the computations, worst case, to make sure you get the
needed backup time. What processor are you using ?

Ideal diode :

https://www.analog.com/media/en/news-marketing-collateral/solutions-bulletins-brochures/idp-web.pdf

https://www.analog.com/en/analog-dialogue/raqs/raq-issue-179.html

Supercapacitor and battery backup power supply design | Video | TI.com

Supercapacitor and battery backup power supply design

www.ti.com

https://www.renesas.com/us/en/document/apn/1195-supercapacitor-based-backup-solutions-design-toolkit

Regards, Dana.

 

[nibblie]

Hi

I recommend to read about
* "ideal diode"
* microcontroller supervisory circuit with backup supply
* backup battery controller

Klaus

I would pick a supercap manufacturer that fully specifies its performance over
T and V and self discharge. Wanting a part to perform a specific way is not
the same as a part fully characterized, and speced in a datasheet. Part you have
picked is about 4 - 6 pages short of specs.

You have to do the computations, worst case, to make sure you get the
needed backup time. What processor are you using ?

Ideal diode :

https://www.analog.com/media/en/news-marketing-collateral/solutions-bulletins-brochures/idp-web.pdf

https://www.analog.com/en/analog-dialogue/raqs/raq-issue-179.html

Supercapacitor and battery backup power supply design | Video | TI.com

Supercapacitor and battery backup power supply design

www.ti.com

https://www.renesas.com/us/en/document/apn/1195-supercapacitor-based-backup-solutions-design-toolkit

Regards, Dana.

Thanks for the tips.

"Ideal diode" like MAX40203 looks suitable, ~15 mV voltage drop and 300nA supply current.

Yes, I agree, probably best to go for something more well specified, I was just sold on the fact that it's a small SMD part.
Anyway, found one from Eaton, better specified and bigger, https://www.eaton.com/content/dam/e...-kr-supercapacitors-coin-cells-data-sheet.pdf, KR-5R5H104-R specifically, 100mF, 0.6 uA leakage.

I'm using a NRF52 microcontroller, so it can operate down to 1.8V.

Simplified calculation:
100 mF with 30% degradation and voltage drop from 3V to 1.8V, or 2.1V when
factoring in ESR + degradation (75*4*0.0001=0.3).
Energy capacitor = 0.5*C*(Vmax^2-Vmin^2) = 0.5*(0.1*0.7)*(3^2-2.1^2) ~= 0.16 J

Assuming 1mA at 3V for 10 seconds, and 10mA peak for 50 ms
Energy avg= U*I*t = 3*0.001*10 = 0.03 J
Energy peak= 3*0.01*0.05=0.0015 J
Energy required tot = 0.0315 J

I still don't know if I need a charger. I don't need cell-balancing, but perhaps I need to limit the current somehow at charge-up to limit voltage drop from battery. I also need to figure out if I can have it connected/charge all the time, I guess it could help out the battery with some of the peaks, but that means it's going to be consistently discharged/charged (waste power on series ESR), and not just activated as a backup supply. Edit: https://www.ti.com/lit/an/slvaf06/s...search%3DSearch-EN-everything%26usecase%3DGPN , this might be a solution, ideal diode+resistor and simple power mux.

 

[KlausST]

Hi,

Calculation.
I do it just with coulomb, not energy. But it should work both ways.

A more detailed calculation includes the ESR of the supercapacitor.
It says max 75 Ohms ... this means an immediate voltage drop of 750mV on a 10mA current.

But we don´t know the circuit and eventually connected fast capacitors .. so impossible to tell if this matters in your case.

Klaus

 

[FvM]

Regarding charging circuit. Small supercaps like planned for your circuit have relative large ESR and can be directly parallel connected to a 3V power supply.

 

[danadakk]

Some interesting comments here about ESR and internal DC R and impact on charging rate -

Supercapacitor - Wikipedia

en.wikipedia.org

Under the heading "Internal resistance".

Some interesting ref material, google "super capacitor design handbook"


Regards, Dana.
"

 

[nibblie]

Hi,

Calculation.
I do it just with coulomb, not energy. But it should work both ways.

A more detailed calculation includes the ESR of the supercapacitor.
It says max 75 Ohms ... this means an immediate voltage drop of 750mV on a 10mA current.

But we don´t know the circuit and eventually connected fast capacitors .. so impossible to tell if this matters in your case.

Klaus

Regarding charging circuit. Small supercaps like planned for your circuit have relative large ESR and can be directly parallel connected to a 3V power supply.

Some interesting comments here about ESR and internal DC R and impact on charging rate -

Supercapacitor - Wikipedia

en.wikipedia.org

Under the heading "Internal resistance".

Some interesting ref material, google "super capacitor design handbook"


Regards, Dana.
"

I hadn't actually taken the time to study the Wikipedia, just assumed it would be brief, but a lot of good info there.

It says ESR is not to be confused with internal resistance. Internal resistance is much larger than ESR even. With that assumption, I might not be able to use a supercap with an ESR of 75 Ohm, especially if it's also going to degrade and increase the resistance by as much as 4x. So even if the ESR at best was equivalent to internal resistance, that would give 300Ohm, which at 10 mA is a voltage drop of 3V! I will order some capacitors to measure to confirm this.

But I also now see there are other supercaps available with a much lower ESR, at the expense of slightly more leakage current.

https://www.eaton.com/content/dam/eaton/products/electronic-components/resources/data-sheet/eaton-pb%20supercapacitors-cylindrical-pack-data-sheet.pdf

https://www.eaton.com/content/dam/eaton/products/electronic-components/resources/data-sheet/eaton-b-supercapacitors-cylindrical-cells-data-sheet.pdf

One of them is 2.5V, however, so might not be an option in my case unless I drop the voltage between the 3V battery and supercap.

 

[KlausST]

Hi,

Again: to get useful answers you need to provide important informations fisrst.

The "drop on as short 10 mA pulse" easily may be overcome with the use of a capacitor.
And capacitors at VCC usually are mandatory. It just depends on the circuit and the part values.
Maybe the installed capacitors already take care for this short pulse - so the problem does not even exist - we just don´t know....

Klaus

 

[nibblie]

Hi,

Again: to get useful answers you need to provide important informations fisrst.

The "drop on as short 10 mA pulse" easily may be overcome with the use of a capacitor.
And capacitors at VCC usually are mandatory. It just depends on the circuit and the part values.
Maybe the installed capacitors already take care for this short pulse - so the problem does not even exist - we just don´t know....

Klaus

Sorry for not providing information, what important information is needed? I don't have finished circuit yet, but I'm planning on having a 1uF and 100nF installed at VCC.

Correct me if I'm wrong, but I don't believe a typical SMD bulk capacitor has chance of handling this kind of pulse?

Required energy (10mA@50ms): U*I*t = 3*10mA*50ms=0.0015J

Even a 47uF bulk and voltage drop from 3 to 1.8V is far from supplying that kind of energy:
47uF capacity: 0.5*C*(Vmax^2-Vmin^2) =0.00013536J

 

[KlausST]

Hi,

We asked for a schematic. Finished or not. Mainly the according 3V part. No need for detailed logic ....

C = I * t / V = A * s / V = 10mA * 50ms / (3.0V - 1.8V) = 417uF

I don´t think 470uF is untypical.

It´s available as tantal in size D for 6.3V.
Even more different types of polymere and as electrolytics.
Even higher value..

***
If I imagne you just compensate this with another supercapacitor in parallel .. it would be more expensive and way bigger.

Klaus

 

[nibblie]

Hi,

We asked for a schematic. Finished or not. Mainly the according 3V part. No need for detailed logic ....

C = I * t / V = A * s / V = 10mA * 50ms / (3.0V - 1.8V) = 417uF

I don´t think 470uF is untypical.

It´s available as tantal in size D for 6.3V.
Even more different types of polymere and as electrolytics.
Even higher value..

***
If I imagne you just compensate this with another supercapacitor in parallel .. it would be more expensive and way bigger.

Klaus

Here is what I have now.

Looked at some Tantals, but they seem to have rather high leakage current, ~0.01CV, which is >10uA. Same for Electrolytics at >= 470uF. But I might need to search more to see if I can find some with lower leakage. I would rather avoid adding another bulky through-hole component. By high leakage I mean, for this application which is going to last for several years on the battery, every uA counts.

 

[KlausST]

Hi,

Looked at some Tantals, but they seem to have rather high leakage current, ~0.01CV, which is >10uA. Same for Electrolytics at >= 470uF

Let´s do some math:
1mA for 5 seconds + 10mA for 50ms (during the 5s) makes an average current of 1.1mA.
so the 10uA leakage current make just 0.9% of the total consumption.

Let´s focus on the capacitance:
The EATON super-capacitor has a tolerance of -20% ... +80% in capacitance.
Let´s just focus on the -20% ... this is a 22 times higher ratio than what´s caused by the 10uA leakage current.

Let´s focus on the backup time:
If the application runs for 5.000s without the capacitor it will run 4.955s (just 45ms less) with the capcitor installed.

Let´s focus on the load:
You say it is 1mA. How sure can you be that an identical circuit does not draw 1.01mA?
I see on your schematic there are I/Os left unconnected. "unconnected" indeed is not the problem, as long as they are pulled to either supply rail internally. But if left floating an input may increase the supply current way higher than 10uA.

Taking this 10uA leakage current into consideration makes sense when your application draws less than 0.05mA in average.(to be within the 20% uncertainty of the capacitor)

In my eyes, if you want to improve a system you have to start with the "highest uncertainties".

Klaus