CycleBridgeEstimates, not measurements
Battery life while you ride
Nobody has yet run a phone flat with a stopwatch while CycleBridge recorded a ride, so every hour figure on this page is modelled from component power draws rather than measured. The model is written down below so you can argue with it — and there is a plan for replacing it with real numbers at the bottom.
Estimated hours on a 15 Wh phone battery
The phone’s GPS radio is never switched on — every number arrives over Bluetooth from the watch.
Nothing to draw and nothing to locate. The watch is doing the work; the phone is only listening.
The phone’s own GNSS running continuously, with a Bluetooth heart rate strap.
The most expensive way to ride: GNSS, plus map tiles fetching and redrawing.
Still locating you once a second, but with nothing to light up.
The finding that surprised us
With the screen on, the display costs more than the GPS does. Swapping the phone’s GNSS for a watch’s saves 0.25–0.5 W, while the screen at riding brightness costs 1.0–1.8 W — so watch mode buys you perhaps an extra hour, not double.
With the phone in a pocket the same swap roughly doubles what is left, because the display is no longer there to hide the difference. If you want the phone to survive a very long day, turning the screen off matters more than which GPS you use — and doing both is what turns a five-hour phone into an all-day one.
And the watch itself?
Garmin publishes up to 42 hours of GPS-only recording for the Forerunner 955 (Garmin’s published specification for GPS-only mode). Recording a ride while broadcasting to the phone once a second and reading wrist heart rate costs more than that figure assumes — but it is far beyond the length of any ride the phone will survive, which is why the watch is the better GPS in every case except not owning one.
What CycleBridge already does about it
Only the screen you are looking at exists
Avoids the map’s 0.3–0.8 W whenever you are not looking at itThe ride readout is six screens you swipe between, but only the visible one is ever built. The other five are not rendered, not laid out, and cost nothing until you swipe to them — including the map, which fetches no tiles while it is off screen.
One tick a second, not one per packet
Cuts UI rebuilds by roughly 3–5×Heart rate packets and GPS fixes arrive several times a second. The readout deliberately ignores that rhythm and redraws on a single one-second timer, so the screen is steadier to read and the CPU wakes far less often.
Black pixels are off pixels
Large, but already counted in the display figures aboveThe readout is pure black with white digits. On the OLED panel in most modern phones an unlit pixel draws essentially nothing, so the design that reads best in sunlight is also the cheapest one to display.
Brightness on the ride screen
Every step down is worth ~0.15–0.3 WPlus and minus buttons sit next to Stop, because the one control you actually reach for mid-ride is brightness, and the display is the biggest cost on this page.
In watch mode the phone’s GPS never starts
0.25–0.5 WWatch mode does not run the phone’s GNSS in parallel "just in case". Position, speed and distance all come over Bluetooth from the watch, which is already locating you far more efficiently.
No ad while you ride
Small, but it is the network radio, which is never cheapThe single banner ad appears on the idle screen only and is gone the instant a ride starts — no ad requests, no ad rendering, nothing on the network for the whole ride.
What could still be done
None of these are built. They are listed with what they would plausibly be worth, so the list can be argued down as easily as up.
A deliberate pocket mode
Up to ~0.25 W, and it makes the pocket figures above the normal case rather than the lucky oneToday, putting the phone away relies on the system turning the screen off. An explicit pocket mode could stop the one-second rebuild entirely while still recording, and wake on the next look — the recording loop and the drawing loop do not have to run at the same rate.
Let the rider choose GNSS accuracy
0.1–0.25 W in phone modeThe phone currently asks for the best fix it can get. A "good enough" setting — single-band, or a distance filter that only reports after you have actually moved — would trade a little precision for a real slice of the biggest non-display cost.
Auto-dim between glances
Potentially 0.3–0.8 W, the single largest remaining leverRiders look at a bike computer in short glances. Dimming after a stretch with no interaction, and coming back to full brightness on a tap, would cut the dominant cost most of the time without ever hiding a number you were reading.
Cache map tiles between rides
Most of the map’s network share on familiar roadsTiles are fetched fresh today. Keeping them would make a repeated route cost almost no network at all, which matters more than the drawing does.
A slower watch-to-phone rate
Small on the phone, more meaningful on the watchThe watch transmits once a second because the readout updates once a second. A two- or five-second option would cost freshness the rider may not miss, and save radio time on both devices at once.
How these stop being estimates
- Run a real ride to flat on each mode with the same phone, screen brightness fixed, and log the percentage every ten minutes.
- Repeat with the screen off to separate the display’s share from everything else.
- Read the per-app energy figures the phone already collects (iOS Battery settings, Android Battery usage) rather than trusting the model’s split.
- Publish the measured numbers here and mark the model as superseded — an estimate that is never checked is just a guess with a table around it.