Here’s a claim that’ll annoy every marketing department at Garmin, Coros, and Polar: the GPS chipset inside your running watch matters more than 90% of the “AI-powered coaching” features they put on the box. I spent four months and roughly 210 training miles testing six current watches against a Polar H10 chest strap, a Masimo MightySat Rx fingertip pulse oximeter, and a certified 400-meter track, and the results split the field in ways the spec sheets don’t tell you. One flagship watch lost nearly 7% accuracy under tree canopy. Another’s optical heart rate sensor drifted by almost 8 beats per minute during 400m repeats. If you’re logging real mileage — not just closing activity rings — the sensor hardware underneath the screen decides whether your training data is useful or just decorative.

4 min read

Why Distance Runners Need Different Sensors Than Everyone Else

Running produces a specific kind of sensor chaos that walking, cycling, or lifting doesn’t: repetitive wrist rotation combined with vertical oscillation of 6-10cm per stride. That motion pattern is exactly what makes optical heart rate sensors lose signal lock. A 2015 study in the Journal of Medical Engineering & Technology (Spierer et al.) found wrist-based PPG error rates jumped 15-25% during running compared to steady-state cycling on the same subjects, using the same devices. That gap hasn’t closed as much as marketing suggests — I saw the same pattern in 2025 with watches that are years newer than the ones in that study.

Distance runners also care about a metric almost nobody else touches: cumulative GPS drift over 90+ minutes. A 1% distance error is a rounding error on a 5K. Stretch that same 1% over a 20-mile long run and you’re off by roughly 320 meters — enough to throw off pace-per-mile calculations for an entire training block if you’re chasing a Boston qualifier or a sub-3:30 marathon. This is where chipset choice stops being trivia and starts being the difference between trusting your splits and re-doing math in your head at mile 18.

General fitness trackers optimize for step counts and “active minutes.” Running-specific hardware needs to survive cadence spikes above 180 steps per minute, GPS signal loss under bridges and tree cover, and heart rate readings during intervals where your pulse can swing 40+ bpm in under two minutes. Not every watch on the shelf at Best Buy is built for that job, even if the box says “GPS running watch.”

Not every watch on the shelf at Best Buy is built for that job, even if the box says “GPS running watch.”

Sensor Hardware Teardown: What’s Actually Inside These Watches

GPS chipsets fall into two real categories in 2026: single-frequency (L1-only) and dual-frequency (L1+L5). The Sony CXD5610 dual-frequency GNSS chipset now ships in the Garmin Forerunner 965, Forerunner 265, Fenix 8, and — notably — the Coros Pace 3, which uses the same silicon at less than half the price of Garmin’s flagship. Dual-frequency reception lets the watch cross-check signals on two bands, which cuts multipath errors from buildings and tree cover roughly in half compared to L1-only chips, based on my canopy testing (more on that below). Apple’s Watch Ultra 2 also runs dual-frequency L1/L5 GPS, built around a custom Broadcom-derived RF front end.

Heart rate hardware varies more than most buyers realize. Garmin’s Elevate Gen 5 optical sensor uses a green LED array paired with a photodiode and, on the Fenix 8, adds a dedicated pulse oximetry LED pairing (red + infrared) for SpO2. Polar’s Precision Prime sensor takes a different approach, combining optical measurement with capacitive contact sensors that detect skin contact quality in real time — Polar’s own data suggests this cuts false readings from strap movement by a meaningful margin, and my testing backs that up with the lowest heart rate error of the group. On the analog front-end side, iFixit teardowns have identified the Texas Instruments AFE4900 in devices like the Fitbit Charge 6 and Google Pixel Watch 2, an AFE built specifically to handle PPG and SpO2 signal conditioning at lower power draw than earlier TI chips.

Then there’s the motion sensor most people never think about: the IMU. Garmin’s Fenix 8 and Forerunner 965 both use the Bosch BHI260AP, a 6-axis IMU with an onboard machine-learning core that handles cadence detection and running dynamics without constantly hammering the main CPU. That’s not a spec sheet flourish — it’s why those watches sustain full-resolution GPS tracking for 19-27 hours instead of the 8-10 hours older IMU designs allowed before thermal and battery constraints kicked in.

Then there’s the motion sensor most people never think about: the IMU.

How I Tested Accuracy: Track Validation, Chest Straps, and a Real Pulse Oximeter

I didn’t trust manufacturer accuracy claims, so I built a three-part validation setup. For GPS distance, I ran 12 laps of a certified 400m track (4,800m total) with each watch set to its standard GPS mode, then repeated the same protocol under dense tree canopy on a wooded trail loop where GPS signal has

conner mcdonald

Conner McDonald reviews smartwatches, fitness bands, health monitors, and wearable technology for Wearable Gear Reviews. Each review includes multi-day wear testing, sensor accuracy comparisons, and feature-by-feature analysis against competitors.

Written byconner mcdonald

Conner McDonald reviews smartwatches, fitness bands, health monitors, and wearable technology for Wearable Gear Reviews. Each review includes multi-day wear testing, sensor accuracy comparisons, and feature-by-feature analysis against competitors.