Archived Forum

One problem is how to monitor battery condition?

You could probably run the +ve terminal of the battery to a spare analog pin on the Arduino, and check its value every second or so.

Naive answer. Is it not enough to use the readVcc() function of class EnergyMonitor ? This is what I was planning to do.

See here:

https://github.com/openenergymonitor/emonTxFirmware/blob/master/emonTx_C...

BTW, I agree with you, a 85% efficiency is better than a battery that can only be used at 20% of its capabilities.

Sorry. Loose connection. Double-posted.

Is it not enough to use the readVcc() function of class EnergyMonitor ?

Possibly.  It would be worth experimenting with I guess.  That NCP1402 is going to try to maintain 3.3V until the end.  It may just hit a digital cliff and be 3.3V one minute and 0 the next, so you may get little warning that the battery is tired.  As an experiment, you could initially measure both the battery and Vcc and plot them as the battery exhausts, then decide which gives you the best results.

I'm with dBC here. readVcc() function compares the output voltage of the regulator/ the supply voltage to the processor to the internal bandgap reference, it will not see the true battery voltage.

My guess is the regulated output will fall in a matter of some tens of minutes, what you want is enough warning to replace the battery at your convenience.

I agree Vcc is unlikely to give a clear warning of battery failure.

It might take a while, as unfortunately I don’t have any near dead AA cells, but I’m measuring battery voltage and Vcc to plot the death of a pair of batteries.  It could be the extra energy in a battery past 1.0V is not worth worrying about, as indicated by the Duracell data sheet and their demise could therefore be very quick beyond this point. 

The battery voltage profile could be a basis for a depth of discharge function, perhaps referenced to Vcc instead of the internal 1.1V internal ref used by the readVcc() function.  This would save an analogue to digital conversions, but assumes Vcc stays stable beyond the point of determining battery near exhaustion.

I guess the ultimate goal would be to combine the knowledge of Vcc and battery condition in one piece of data.

I'm going to be experimenting with these ......

http://www.saftbatteries.com/Produit_LS_cell_range_303_6/Language/en-US/Default.aspx

Eamonn

I must admit I am having pretty good results from a 3.6vAA battery.

Its been temperature logging since mid January.  The battery voltage was 3.6v when I popped it out today to check the voltage using a multimeter.

I've not tried Lithium-thionyl chloride batteries, but looking at the data sheet they certainly seem impressive.  The voltage holds up at 3.6V right to the end of their life and with 3x the energy of an alkaline battery at 9.36Wh vs 3Wh they should certainly last.

They are, however  £3 each vs 50p for alkaline, so 6x more expensive.  There is no need for an stepup regulator to maintain voltage though.

I got the idea to use these batteries after seeing them used in commercial wireless temperature sensors.  If I remember correctly they estimate the battery life in years.

My sketch reads the temperature every 10 seconds but only sends the temperature value to the base station roughly every hour or 0.2°C temperature change.

The batteries are not cheap but no one wants to keep replacing batteries every month.

I have not tried the 3.6v batteries without a LDO voltage regulator.  The voltage seems pretty close to the limit of the RFM12b.

w.r.t. the Lithium-thionyl chloride batteries; I was concerned about the rfm12b limit also, and at 3.6V it's right on the limit for all the rfm12b alternatives! (SRF, XBEE, NRF24L01)....

but I really think for a remote monitoring node we need to have a solution that means you can 'forget' about changing batteries for a few years.

I'm using two alkaline batteries with a 3.3V step-up regulator on my temperature monitoring nodes.  They're part of my remote central heating control system so need to be reliable.

I calculate the battery life to be approx. 645 days with a measurement interval of 1 minute.  This is assuming 3W.hr per battery and 85% efficiency from the regulator.

If a 3.0V step-up regulator is used then two years is probably feasible.

The difficulty is determining / monitoring battery condition without actually affecting it.  Connecting the battery to an analog pin will increase the drain on the battery, so its not ideal.  There's some info here on how to reduce this battery drain.

My goal however is to be able to periodically switch the NCP1402 step-up regulator off thus returning the output to the battery voltage, then measure the battery voltage condition on VCC as normal.  The NCP1402 chip has an enable pin, which when high or floating turns the regulator on, when low its off. 

This sounds just the ticket, but unfortunately the breakout boards I bought don't have this feature and on such a small smt chip I don't believe I can modify it.

To my mind the ideal would be to have an NCP1402 or similar regulator included on the emontx and have the means of enabling or disabling it.  The NCP1402 copes with an input voltage range of 0.8 to 6.0V so would also deal with an overvoltage concern when using Lithium-thionyl chloride or similar batteries.

What about LiPo batteries ? Wouldn't they compete in terms of voltage stability ? It looks like they keep a constant voltage during the discharge.

Good idea John,

This is something we're looking into for the emonTH battery powered temperature and humidity node which I'm currently working on http://openenergymonitor.blogspot.com/2013/06/emonth-prototype.html.

The LTC3525 seems to have greater efficiency and fewer and smaller external components are required. 

It could be offered as an optimal extra. I would very much like to be able to use AA rechargeable batteries in these nodes. 

I agree the LTC3525 seems more efficient, but I couldn't source them at a decent cost, so ended up with the NCP1402. 

I'm not sure about using rechargeable batteries for temperature nodes.  In general secondary cells have a lower life time cost, but when used for low current applications self discharge becomes an issue.  Most secondary cells self discharge at 10 to 30% per month severely limiting their usable energy capacity for low current applications.  Even new Li-Po cells discharge at 5% per month.  So with this in mind I think for low power applications alkaline or lithium become cheaper / greener.

I'm using two AA alkaline cells in each node and they are on target to last around two years.  That works for me.

Thanks John, I didn't know about self-discharge. Explains why Li-Po are used for model vehicles where power is needed for a limited amount of time.

I suppose the AA batteries can be replaced by refillable ones, the lower voltage issue being addressed by the regulator.

Self-discharge is a "feature" of all rechargeable cells. http://batteryuniversity.com/ is a useful source of general information, especially in combination with the manufacturer's data sheets.

I see on the schematics  that you finally selected the LTC3525.
Could you explain the reason why the NCP1402 was not a good solution?

Does Glyn Hudson's post above not explain?

I had seen Glyn Hudson's post. But there was no conclusion with regard to price (john.b was not convinced).
The ltc3525 is far more expensive (3.61€) than ncp1402 (0.67€).
Is that little efficiency gain justifies the extra cost?

If we consider about 20uA additional quiescent current, this represents 175mAh per year.
To be compared with the capacity of AA batteries (approximately 2500mAh per battery)