Let's Make Robots!

Weather station project: death, autopsy and resurrection

On April 2nd a bit after 11:33:22.546588 AM GMT+3 it happened. My weather station's (http://letsmakerobots.com/node/25773) remote unit died. I was having my morning coffee and noticed that weatherlogger hadn't received any updates for last 10 to 15 minutes or so. Last update was at previously mentioned timestamp.

I took a look at remote units voltage data that weatherlogger had been saving for a bit over a month and here's what I found:

Remote unit's battery voltage. Clickety-click for bigger picture.

My first though after seeing the voltage drop was that there's probably some condensed water inside the casing and it's causing shorts in the circuit. Yesterday I did an autopsy and found out that, to my surprise, the casing was completely dry inside. No trace of any water or moisture.

Next thing to check: the battery. And this is how it was:

Yup. That's a dead 6V battery.

Battery's voltage was barely over 2V so it's practically dead. The battery had 4AH capacity and if I calculate 31 days without a glitch I get about 5mA average current. That figure also includes the self-discharge of the battery. Originally I estimated the average current to be in microamp range so either I have calculated it wrong, have a bug in the software (no sleep), battery's self-discharge was bigger than I thought or there's a hardware fault. Or maybe a bit of all of these.

Anyway, I really didn't bother to investigate the mysterious power drain and bought 3000mAh lithium batteries. I'll just try them and see how it works. If the average current really is around 5mA then those batteries should only last 25 days.

Weather station has been resurrected and once again it's logging weather data. Let's see how long it lives this time.



Update Nov 13, 2010: 

PaulJD suggested that the battery might have frozen. So I decided to compare temperature and battery voltage (pics are clickable):

Temperature and battery voltage together.

Close up of last couple of days.

In the end battery voltage seems to follow air temperature after some delay. Delay can be explained by battery's mass. Bigger mass equals slower temperature change in the battery. Right?

So did it freeze? Dunno, may be it did. Those two spikes on 30th and 31st during afternoon certainly suggest that the battery got frozen during the night and "unfrozen" at daytime.


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Have ya tried setting it up with a solarcharger?

It is interesting the voltage drop btw...do leadcells always die this spectacularly?

How are you mesuring the voltage? Are you using a reference voltage source or is it just a from the cell itself?

Solar charger was in my plans but then work got on the way and I didn't have enough time and energy to build it (even though it's fairly simple). I might try that during summer holidays.

I don't know if lead cells always die like this one. Voltage drop certainly is interesting.

I'm measuring the voltage like this: Voltage divider halves the voltage from the battery and that's connected to ATMega's ADC. ADC's reference voltage is 3.3V from regulator. Voltage divider is 2 x 100K resistors so that doesn't waste too much power. Now that I think of it I'm not sure if I took regulators power loss into account.

If you were using an ldo the Vref should be fairly accurate up to that voltage. If you used a zener diode, that would help in reducing the power consumption. You could also find some lower than 3V, maybe 2.x, though at that V you might start to get brownouts. I'm guessing that you might be set in the proc, but if not, the "popular for a day because it was cheap MSP430" also known as the "launchpad" would be a great platform and extremly low power consumption(this is what it was designed for). I've been playing around with it for a couple weeks when I've had time in the evenings and have found it to tbe a great proc. I'm planning on converting my current picaxe weather station setup to use it when I have learned it well enough.

was it atmega or attiny in your remote device? (the battery is used to power a remote device, isn't it?) 

and still, did you measure what was finally the power consumption? what led to that extreme power drain? 

There's ATMega inside the remote unit and it's powered by lithium batteries. I had plans solar cells + lead acid battery but I haven't had time or energy to build that.

Calculated average power consuption is now around 5mA (3000mAh batteries last about 25 days). There could be something fishy going on with ATMega's, RFM12B's and SHT15's sleep modes but I haven't checked it.


Sulfates?  I think voodoobot's idea of a solar charger would be great for your weather station - you could also monitor the charging, which you could log as amount of sunlight.


It might be that 2 cells are totally sulfated and that's why the voltage is ~2V (=voltage of one cell that's still alive). I'm not sure. Luckily it was a cheap battery.

I already have a LDR in the remote unit to log light levels. But getting some kind of current measurement from solar charger would be nice and it would be interesting to see if the system would survive winter when there's not that much sun light.


I thought I read a way to use an opamp to get current meausurements. sort of a current to Voltage output that you could then read into an adc...Just a thought.  :)

when building ammeters, I use a very small ohm resistor (like .1 ohm) in line with the signal you want to measure, and take the voltage drop across it.   the more current you have, the higher the voltage.   ammeters are actually very low voltage meters.     I suppose an opamp would amplify the voltage I'm talking about up to a range that ADC's can work with.

was it really cold during the final dying days?   did it freeze?

that battery is rechargeable, yes?

you could probably charge it overnight and get another 30 days from it.

I agree, even the smallest solar panel, 9v or 12v, would do wonders for that 6v battery, especially when draw is only 133mah per day (4ah / 30days).   you would get that in little over 1 hour of using $10 solar panels.   and it would probably run forever.

as I recall when you were planning the weather station, 5ma draw is about what you expected, and you were exactly right.