Hi All,

I'm relatively new to design, so I would sincerely appreciate any help, advice and feedback to get me started on the right direction.
Here's the scoop:
I'm designing a battery powered tracking device. This is device is powered using a CR2450X battery by Murata. I have linked the data sheet (Japanese followed by English).

What I can share is:
My device can read the battery voltage and temperature (this device will be used outdoors).
My device has a cutoff voltage of 2.0V (CR2450X nominal voltage is 3V)
The device's default state is 'asleep' (2uA consumption) and it periodically wakes up to do a RF transmission to a base station (50 ms transmission at 10mA, this is the largest load placed on the device)

I think I need to consider my device 'dead' when it can no longer perform the critical function of performing its RF transmission.

My thought is this - the battery must actually be considered dead long before it hits the actual 2.0V cutoff value (application specific), because putting a 10mA load on a battery at that point will cause the voltage to drop much further below 2.0V, rendering me unable to do my RF transmission.

So to provide a clear question - At what battery voltage should I consider my device actually 'dead' ? And How can I know this before hand? I think i can do a temperature and battery voltage reading, and pair that information together, but I am lost as where to take it from there. I know the answer will be temperature specific, and I will have to account for these temperature variances into my strategy for warning the low battery event is imminent.

Thank you in advance.

 

I think the voltage under load should be followed rather than the unloaded. For instance, if it's 70°F and the voltage observed at the end of the last transmission was 2.1V, and the temperature is now 30°F, there's no way a transmission would now succeed.

Defining precisely where the wall is placed - the combinations of temperature and voltage that define the "dead" zone - will take some thought and understanding.

I also recommend collecting some data. There's really no substitute. It looks like you're already on the right path. Use the data sheets to estimate where the wall is, then test your model to validate it.

 

So...what happens when the battery is dead? I assume someone changes it?

Why not just report the battery voltage with each transmission, and have your master server alert (via email?) that a particular device needs attention.

The alert trigger level can be decided later after you've collected and analyzed a bunch of data.

 

This Panasonic data sheet shows some measurements with temperature.
This Energizer sheet give some info on its pulse voltage.

You can use that to estimate the temperature coefficient of the battery (appears to be about 4.3mV/°C) and that, with the battery voltage at a 10mA pulse load, calculate for a given ambient temperature, at what low-current voltage the battery is dead for your purposes.

For example, at -25°C, the battery voltage would be down about 215mV from ambient.
The 10mA pulse causes a voltage drop of about 300mV near the end of the battery life.
This means the minimum quiescent battery operating voltage for a 2V, 10mA pulse at -25°C would be about 2.5V at 25°C, and 2.3V at -25°C .
So that way you can calculate the minimum battery voltage required for any anticipated ambient temperature.
Make sense?

The above doesn't include any possible change in the battery pulse resistance with temperature, so you may have to do some tests by putting the battery in a freezer if you want to include that.

 

Log the temperature and voltage values each transmission.
You will generate a profile for the device you have created.
When no data has been received for a period of time, you'll know the battery has died. Determine what the period between transmissions is and use that in your profiling.
That will help determine the time to replace the battery.