I took my kitchen probe thermometer out of the drawer the other day, to check on some burgers, but it was dead. Later in the day, I tried to measure some critical dimensions with my electronic calipers, and had the same problem. When I went to ride my bike before it got dark, the handlebar-mounted speedometer/odometer was also dead. As a final touch, the tiny digital voice recorder I take in the rides, in case a brilliant idea hits, also was looking dim and sluggish.
What's going on? Have the gods of electrochemical storage conspired to give me a hard time?
No, what happened is a function of several factors adding together. First, there is the unavoidable self-discharge of a battery. Second, the so-called button or coin cells used in these devices have limited energy storage; the relative large CR2032 cell specs between 200 to 250 mA-hr, depending on the vendor.
But the factor that really aggravates things is the use of soft on/off in the design, rather than a definite, “off really means off” design with zero current drain in the off mode. As a result, the device is always drawing a little current when nominally off. Even though it is in a sleep or power-down mode, it needs to sense the contact close or activation of the physical on/off switch, among other functions.
I understand why this makes sense. A soft switch is easier to implement mechanically and electrically, and is compatible with other user controls, if any. It also allows some internal functions, such as a clock, to keep running and thus accurate when the unit is allegedly off. There are other design advantages as well: in some cases, key set-up and calibration factors can be retained in memory, if it's not a non-volatile design.
But doing a good, low-power design of a circuit, even when using low-power IC, is not trivial. Do a quick estimate, and you'll see the impact of even a little quiescent drain on battery life. At 100 microamps, even that 200 mA-hr cell (again that's one of the larger ones) will run for only 2000 hours, which is about 4 months (a year is 8760 hours). You can easily see how modest changes in quiescent drain and battery capacity seriously affect battery life of these “off” units.
But it's easy to complain, while engineers solve problems, right? My approach was logical: I decided to buy spare batteries and keep them on hand. First, of course, I did a check among the dozen devices I identified around the house, and found five different battery types in use. This was not good from an MRP (material requirements planning) or inventory-investment perspective, so I just bought the two types that were most common. (Does anyone want to take a bet on which battery type I will actually need next time? I didn't think so!)
So while I now understand the problem and have a partial solution (as they say in calculus class), I still have one puzzle. I have used a no-name desktop clock with dual time-zone display since 1984. It runs on two CR2032 cells, and I have only replaced them twice since I got the clock (I know this because I wrote the dates of replacement on the back, each time). How cans a device made that may years ago–with a much older process and larger device geometry, and thus more power-hungry, and which is running 24/7/365 with a display–be so low power? Were its designers unappreciated and clever engineers, who took their mandate of low-power design very seriously? Perhaps so.
And if you are out there now: I appreciate your effort and successful results, even if no one else knows about what you did.♦