A sensor that has run out of battery does not tell you it has run out of battery. It stops broadcasting, and a thing that has stopped broadcasting looks exactly like a thing that has nothing to report. The wet bay sensor that has been reading a comfortable four degrees all winter and the wet bay sensor that died in November look identical from the app, right up until you check the timestamp.
This is the least interesting subject in RV monitoring and one of the two or three most common reasons the whole thing quietly stops working.
What these sensors actually run on
Nearly everything on the compatibility list falls into three groups.
CR2032 coin cells. The most common by a wide margin. Cheap, everywhere, and the one you want a drawer full of.
Two AA or AAA cells. Bigger sensors, and the ones with a screen. Far more capacity, and the screen is usually what eats it.
Larger lithium cells such as the CR123A, in sensors built for outdoor use. Fewer of these, longer life, harder to find in a small town.
Check what yours takes before you need one. The moment you need one is a Sunday.
Cold is the thing nobody warns you about
A coin cell loses a substantial part of its usable capacity below freezing. Not permanently, and not because it is faulty: the chemistry simply cannot deliver current as readily when it is cold, and a radio transmission is a current spike.
The behaviour this produces is genuinely confusing the first time.
- The sensor in the exterior bay goes silent during a cold snap.
- You bring it inside to change the battery, and on the bench it works perfectly.
- You put it back. It works. It goes silent again on the next cold night.
Nothing is broken. The cell has enough left to run warm and not enough to run cold, and it will keep doing this until it is replaced. A battery percentage reported just before it went quiet will have looked entirely healthy, because it was measured warm.
If a sensor goes quiet in the first hard frost, replace the cell before you suspect anything else. It is the answer far more often than a fault, and it is the cheapest thing to rule out.
Lithium, not alkaline, in anything that gets cold
For AA and AAA sensors this is the single highest-value change you can make. Lithium AAs cost noticeably more and hold up far better in the cold, and in an unheated rig over winter the difference is not marginal.
They also do not leak the way an alkaline does when left flat for a year, which matters when the sensor is behind a panel you would rather not take off twice.
Coin cells are lithium already, so there is no equivalent upgrade. Buy a known brand rather than the cheapest multipack; the very cheap ones are often old stock, and a coin cell that has been sitting in a warehouse for three years starts partway through its life.
Change them all at once, on a date
Sensors bought together and installed together fail together, and they will pick the same fortnight to do it. Replacing them one at a time as each goes quiet means finding out one at a time, which means several separate trips and at least one gap you did not notice.
Pick a date instead. Once a year before the rig goes into storage is the obvious one, because that is the start of the longest stretch where you are least able to check and most reliant on it working.
Write the date on the sensor. A strip of masking tape and a pen is a genuinely good answer, and it survives the moment eighteen months later when you cannot remember whether you did the bay one.
Watch the reported battery, but do not trust it
Most of these sensors report a battery percentage and RoamVitals reads it. It is worth having and it is worth knowing what it actually is: a voltage measurement, converted to a percentage by a curve the manufacturer chose.
Coin cells hold a fairly flat voltage for most of their life and then drop quickly, which is the same shape of problem that makes voltage a poor way to measure a lithium house battery. So the number sits at a comfortable-looking value for months and then falls off a cliff, and the cliff is short.
Treat any sustained drop as urgent rather than waiting for a threshold, and remember it reads high when warm.
The failure worth designing around
Silence is indistinguishable from good news. That is the real problem here, and it is not solved by better batteries.
It is solved by treating last seen as a reading in its own right. A sensor that has not reported in two days has told you something, and it has told you something more urgent than most of the readings it sends when it is working. Check the timestamps, not just the values, and be suspicious of any number that has been reassuringly steady for a very long time.
The short version
Keep spare CR2032s in the rig. Use lithium AAs in anything unheated. Change them all on one date every year before storage and write the date on the tape. And if a sensor goes quiet in the first frost, it is the battery, not the sensor.
