NRF52810 using CR2032 in freezing temps

[haricane]

Hello Nordic team, 

We have developed a smart tag using the NRF52810. We are looking at using it in asset monitoring application. 

It is powered by 2x cr2032 in parallel. 

My question:

1. Have your customers been able to power the nrf52 via cr2032 coin cell in freezing temperatures (-10 deg C)? Is there a better coin cell for this application. 

2. what would be a good capacitor size to ensure brown outs do not happen, given that BLE adverts take (~10ma for 380 us) x 3, while the CR2032 can only supply a 200ua current. 

The reason I ask is that I've tested 8 tags and they all fail in the freezer in 2 days. When I inspect the cell voltage, its at 1.5v or 1.6v. When I run the tags outside, they last months. I am assuming there was a brownout event, which pulled the mcu into reset which then discharged the cells. 

Kind regards, 

Hari 

[JONATHAN LL]

haricane said:

Does the NRF power profiler support logging over a long duration, such as 48 hours?

It could, it is possible to adjust the "resolution" of the sampling so with a low sampling rate you will be able to extend the time your are sampling. 


So just adjust it to cover 48hours 


What circuit configuration are you using ? There is a small difference in efficiency with DCDC and without depending on the voltage. So the higher the voltage the more efficient the DCDC configuration is, but there is a voltage range\level where the LDO(no DCDC) is just as efficient if not a small amount more efficient. 

haricane said:

Also, could you please recommend a capacitor size for such an application. 

Low ESR caps are the way to go but for size it can be hard to just say a number, the size could depend on the current consumption during the critical stage where the device sizes to function as intended. I am not sure what the typical BOR current draw is, but I suspect it will depend on the circuit configuration. 

Also, some SoC's will differ in how low the voltage can be before there is a brown out. There will be some variance there as well so not all cells will stop working at the same voltage. 

Regards,
Jonathan

[haricane]

Hi Johnathan, 

Thanks for the followup. 

I will run some tests where we sample the supply voltage over 48 hours. 

Regarding your questions, we are using the DCDC. Its the Laird BL651 which has the DCDC circuitry built in. 

I am experimenting with different capacitance values. 

Kind regards, 

Hari 

[hmolesworth]

Stored charge Q is the ticket; Q=CV so either use  lot of (ultra low-leakage = ceramic) capacitance C or use a high voltage V or better still do both. The current surges on the coin cell are predictable, typically during BLE transmission but also (say) for flashing an LED. The CR2032 pair in parallel are giving a useable voltage of typically between 2.4 and 3.4 volts but due to the high internal impedance at low temperatures or near end-of-life cannot source anything like the pulse transmission current required. The nRF52810 can PWM a voltage doubler or tripler to raise the voltage V just prior to a high-current event. Ceramic capacitors allow this high voltage to be maintained for longish periods but the voltage rating on ceramics has to be at least 2 x the maximum voltage to get the full capacitance value. High leakage Tantalum capacitors don't have that voltage rating problem, but can only be used as charge storage devices for short periods due to the extremely high leakage current. What is a high voltage V? Without an external DC-DC it is limited to the max input voltage the nRF52810 DC-DC can handle, 3.6V, but with an external DC-DC regulator much higher voltages can be used, say 5 or 6 volts or more.

Naturally using an external DC-DC means a buck-boost regulator can be used instead to boost the voltage without the nuisance of having to PWM a voltage doubler, although now the maximum voltage is that the nRF52810 DC-DC can handle, 3.6 volts, or (safer) a tad less such 3.55V. 150uF at 3.5 volts is much better than the same capacitance at 2.4 volts. The coin cells should be treated as a trickle charger to the true power source, which is the capacitors. 150uF (6.3V rated ceramics passable or better 10V ceramics) is a starting point; better to use a lot more. The nRF52810 DC-DC is a lot more efficient at 3.55V than 2.4V and less likely to be susceptible to a brown-out.

Should there be other high-discharge components which need to be managed, typically those not subject to a brown-out, isolate the nRF52810 supply with an ideal schottky diode such that the nRF52810 storage capacitors are not drained by those other components. Ensure the buck-boost DC-DC does not drain backwards or supply other stuff; if it does add an ideal diode in that path as well. A good ideal diode for this purpose is the Maxim MAX40203 max40203.html

[haricane]

Hello Hugh, 

thank you very much for the detailed info. I have increased my capacitor size to 47 + 47 uf and I'm running some tests now. I earlier had only 4.7uf. I think your analogy of using the cr2023 as a trickle charger is great. Will drop you an update later . Thanks again. 

[hmolesworth]

Hopefully you are having some luck; meanwhile here is a Nordic white paper about the use of ceramic capacitors and the effective capacitance of the same with some specific references to the nRF52840 which may be of use: nwp_030

[haricane]

Hi Hugh, 

I have since resolved the issue with a 47uf cap. 

thanks for the whitepaper, it will surely be useful for me in the future.