Very narrow and minimal switch node - low power consumption experiment

[AWI]

The minimalistic Very narrow and minimal switch node lends itself for some battery experiments. A how low can you go study with a standard MySensors node.

The objective was to explore how long it can operate with a small i.e. CR1620 (3V, 75mAh, until 2V) battery.

The node is attached to a 'reed contact' magnetic sensor.
This one is connected between arduino pin D2 (connector) and GND.

The node is programmed with the bootloader as described by @m26872 for the My slim battery node.

The findings:
Summary: minimum of 4 years operation

Basic setup.. a uCurrent current to voltage converter. This one is set to 1mV per nA ! (0.001uA) The photo below shows the sleep current at ~3.00V supply voltage (nominal battery voltage (ignore the minus sign on the voltmeter. ;-))). Supply current during sleep ~0.9 uA!

Dropping to ~0.5uA for a supply voltage of ~2.2V

The current during wake is ~13mA during ~4ms. This is for reading two magnetic door contacts and the battery voltage.

A simple and conservative calculation (using the Android Electrodroid battery calculator):
Sleep: < 1uA
Wake: 13mA during 4ms and awake once every hour
Average consumption: 1.014uA,
Battery: 75mAh
---> expected lifetime: ~74000 hours = ~8.4 years !
For a wake-up around every minute this would still be over 4.5 years.

A CR1620 (dimension 16mm x 2.0mm) with a (75 mAh) capacity would fit perfectly with the Very narrow node in a 16x16mm housing. I am off to build some MySensors keyfobs (remote controls)

[hek]

Cool! What the self discharge of the CR1620s?

[AWI]

@hek for an A quality Energizer type: 1%/ year

these will probable be different but at Euro 0.16 a piece ....

[scalz]

@AWI : nice report

imho I am not sure it could last so long time (4.5years with a 75mah !). Unfortunately, because of the rated capacity vs effective capacity, coincell brand/vendor, ...
or maybe for a very basic keyfob..I know it's an estimation lol
for a keyfob (depends on application type), we would need signing which would take more than 4ms for the process (and coincell don't like pulse drain > 5mA). But for simple application/sleep mode, I have reported same as you for coincell. And I had not the same sleep current for coincell vs aa/aaa
If you are interested in coincell resources, maybe you already know it ;), always good to know :
https://www.dmcinfo.com/Portals/0/Blog Files/High pulse drain impact on CR2032 coin cell battery capacity.pdf
http://www.ti.com/lit/wp/swra349/swra349.pdf
http://www.embedded.com/electronics-blogs/break-points/4429960/How-much-energy-can-you-really-get-from-a-coin-cell-
I use this link for battery calc : http://oregonembedded.com/batterycalc.htm

[AWI]

-- revisited ---

I changed the bootloader to a 8mHz version. As the processor runs 8 times as fast the active period pulse width is now < 0.5 ms (earlier 3.8 ms) The sleep and active consumption have not changed. This would reduce the 'pulse' burden on the coin cell as attentively mentioned by @scalz.

Something comparable to 'internal resistance' increases as the charge reduces. This will impact the available voltage for the node. I used a conservative figure for battery capacity (50mAh) and a load current of 0.8uA which will be less if battery voltage goes down.

Have three nodes ready for real life test with constant battery measurement. I hope to report just within five years from now