A few months ago, I replaced my "wired" bike speedometer with a wireless one from Cat Eye Co., see "Wireless or not? When are short-distance, point-to-point RF links worthwhile?". [Note: A wired one has a cable connecting the sensor on the front fork to the handlebar readout; a wireless one does not.]
In general, I am not a fan of wireless links for this type of short-distance, simple link, since it seems to me that the additional cost, and need for two batteries, as well as potential for problems due to interference and other factors is greater than the problems a wireless link purports to overcome. But it the case of the bike speedometer, that short cable connected the sensor and the readout is a real nuisance, as it gets snagged on branches or when you put the bike in a car, it interferes with maintenance, and it needs to be checked for damage due to flex and stress.
When the display on the handlebar readout started to fade, it was a sure sign of a battery starting to run down. I replaced the battery in the unit and all was well. But then I started to wonder: should I also replace the battery in the fork-mounted sensor/transmitter (aka sender) as well? And that became a thought problem that has nagged at me since.
Here's the situation. First, keep in mind that entire setup has a very low duty cycle. The unit generates one pulse/revolution of the wheel, which means only about 0.2 pulses/sec for each mile/hour of speed (about 0.15 pulses/sec per km/hr speed). At a typical cruising speed of 15 mph/hr (25 km/hr) we're looking at just 30 pulses/sec. Both the sender and the handlebar readout use a CR2032 coin cell with 200 mA-hr capacity.
On one hand, the sending unit has less internal circuitry, but transmitting usually takes more power than receiving does. It also has fairly little to do, since it only has to send data when it senses an input pulse via the Hall-effect transducer. The rest of the time, it can go into a deep-sleep mode, and wake-up latency is not an issue here.
In contrast, the a receiver in the handlebar readout unit probably needs less power for its front end, but the complete handlebar readout also has to do calculations, and drive the readout, so its microcontroller is more power-hungry than the one used in the sender, assuming the sender even has one (which I doubt). In addition, the readout is on all the time, once a first pulse is received, and it only goes into sleep more when it gets no signal for about 10 or 20 seconds. Even then, it has to always be on alert for a wake-up pulse from the sender.
My problem in determining the relative life of the battery in each is twofold. First, even if I do build a fixture to get into the battery holder to measure the current, it's very hard to measure such short current spikes with accuracy. Second, so much of the actual battery life depends on how the bike is used, the amount of time it is not being used, and other factors which determine the sleep/active ratio.
So for now, I'll just give it more thought, but I'll probably replace the sender battery as well, just to be sure. After all, both units have to be OK with good batteries for the speedometer to work, which is one drawback of a wireless unit compared to the wired one. Or maybe someone will come out with a design which uses vibration energy harvesting to power the sender?–theirs is plenty of unused vibration available on a bike!♦