I am working on a project that requires the use of BLE Beacons in very low-temperature environments, on average -15°C but on occasion as low as -25°C.
We are currently trialling a brand of BLE Beacon that receives power from 4x AA 3.6V Lithium (Li-SOCl2) batteries in parallel. You may find the beacon product page here https://accent-systems.com/product/ibks-plus/.
We need the beacons to broadcast exclusively one Eddystone-TLM frame per second for a 1Hz advertising rate. At room temperatures, the beacon app estimates a useful life of 86 months. We will be satisfied if they last half that amount of time, but I am trying to determine a rational estimate for their true life in cold temperatures.
We are currently trialling them under these environmental conditions and we will monitor the battery voltage after 2 months, but I am concerned that battery voltage measure does not offer the full picture. My understanding of what causes battery performance to reduce in cold temperatures is not the absence of potential energy but the capacity for the battery to convert the chemical energy into electricity to supply current.
To determine how much current draw is needed to power the BLE beacon I used my multimeter and observed a current draw of
200μA (microAmps). The current draw was also not consistent with a baseline draw of 30μA with spikes of 100-400μA every second (Editor's Note: I was reading the multimeter wrong) 20μA (microAmps). The current draw was also consistent with a baseline draw of 3μA with spikes of 10-40μA every second. This seemed to make sense given the pulse-like behaviour of BLE beacon tech, additionally, at 86 months or roughly 62000h at approximately 0.2mA the 4 batteries supply 12400mAh which seems about right for 4 AA Lithium.
From this observation, my hypothesis is the amount of chemical conversion going on in the battery has to be extremely small. Perhaps small enough that even very cold temperatures would not demonstrably change the estimated life of the BLE beacon than that of room temperature.
I have attempted to research the subject, but most of the discussion is around cell phones and car batteries, the former requiring a pretty constant draw of current that is not sustainable in the cold, and the latter requiring a great burst of current upon ignition. Neither of these use-cases seems to be applicable.
The question is, with knowing how much current the circuit supplies, how do I make a practical estimate for the life of the BLE beacon at these temperatures?
Update April 15, 2019
In performing greater research on the question of profiling the depth of discharge of Lithium batteries I soon learned the question is meaningfully dependent on the type of lithium-metal chemistry of the lithium battery. It was more productive to narrow my research specifically towards lithium-thionyl chloride cells. The exact battery provided by Accent Systems in their Lithium modified beacon is the SAFT LS 14500 (manufacturer information can be found here).
One of the serious problems associated with exploiting lithium–thionyl chloride cells (LTCC) is estimating their depth of discharge (DOD). Diagnostics of discharging cells of this type is very difficult. This is due to the distinctive feature that LTCC voltage changes very little over the whole duration of discharge to the catastrophically rapid drop in cell voltage at almost total discharge. In fact, LTCC discharge curves cited in most literature sources and advertising material (especially at currents that are not extremely large) are nearly parallel to the capacity axis.
Kanevskii, L. "Special Features of Discharge Characteristics of Different Types of Lithium-Thionyl Chloride Cells and the Problem of their Diagnostics." Russian Journal of Electrochemistry 45.8 (2009): 835-46. Web. 15 Apr. 2019
The following reference chart was pulled from the Tadiran Batteries Technical Brochure here with the LTCC battery chemistry highlighted in blue at the top showing a distinct steady voltage with and immediate drop-off.
The closest research I could get for the study of low-temperature discharge of LTCC batteries was a paper written in the late 80s written about LTCC batteries a memory back-up power source where the produced the following, very relevant result on an Li-SOCl2 battery of R6 size (ER6C) [ R6 is the IEC 60086 system nomenclature for a AA battery] :
When the cell was discharged at -40°C under a 360Ω load (corresponding to about 10 mA discharge), the working voltage reached 3.1 V and the discharge capacity was 900 mAh.
Uetani, Y., and T. Iwamaru. “Characteristics of a Lithium-Thionyl Chloride Battery as a Memory Back-up Power Source.” Journal of Power Sources, May 1987, pp. 47–52.
Based on the manufacturers stated nominal capacity of 2600 mAh under 2mA at 20°C with a 2.0V cut-off (found here) I am willing to assume a very similar battery capacity at a 10 mA discharge. Further, if I am to assume a linear proportional decrease in battery capacity from 20°C to -40°C I calculate a 28.33 mAh decrease per decrease in °C operational temperature. Put another way, I can solve for battery capacity in mAh (BC) with the following equation where T is the temperature in °C:
So at -20°C I estimate 1466.6mAh per cell (about half the capacity at room temperature). So at 4x batteries, I should have 4x the battery capacity for a total of 5866.4mAh.
At an average approximate operational current consumption of 20µA (0.020mA) when programmed to advertise 1 Eddystone TLM frame at a 1000ms broadcast rate I should theoretically have a battery life of 263320 hours or 30 years! This strikes me as pretty dramatic and I feel perhaps I am underestimating the current draw of the circuit but even the Accent System IBKS Plus datasheet seems to attest to a 3.8µA Idle Current Consumption which is what I observed from my multimeter between advertisement broadcasts.
The question I have now is there anything particularly wrong with my calculations and stated assumptions. I know what battery capacity is variable on the current draw, but it seems from the literature that the battery capacity only increases as the current draw decreases which would only stand to increase my calculated estimate.
Perhaps a 2.0V cut-off is too liberal and if I were to consider a cut-off of something like 3.0V I would find that the total battery capacity would reduce. However, based on the documented discharge behaviour of LTCC batteries, it seems like the voltage drop from 3.4 to 0 is almost immediate and all at once so I'm not sure setting a higher voltage cut-off would decrease my estimate substantially either.
Ultimately I feel like 4-year battery life is almost guaranteed at the current advertisement settings, but if anyone could offer some corrections and/or validation to my estimates it would be greatly appreciated.