How a phone measures car vibration (ISO 2631-1, explained)
The short answer
A smartphone’s accelerometer measures acceleration in three directions, hundreds of times a second on most Android phones and about 100 times a second on iPhones. Laid flat on a car seat, it records the vibration that reaches you. Software can then do two things with the recording: weight it the way ISO 2631-1 says the body feels vibration, giving a comfort value in m/s², and split it into frequencies, so a peak that rises with road speed can be matched to the wheels and one that follows engine speed to the engine. A phone can’t see inside parts: it screens, and a mechanic confirms.
What the sensor records
The accelerometer reports force along three axes, including gravity, which tells software which way is down. The gyroscope reports rotation, which shows whether the phone moved. RidePulse checks before every test that the phone is within 10° of flat and still, and stops if it tips more than 15° during the test.
How fast the sensor reports sets the highest frequency you can see: at most half the sample rate. Most Android phones deliver 200 to 500 samples a second; RidePulse analyzes at 400, 200 or 100 a second, whichever the phone sustains, and looks at vibration up to 100 Hz. iPhones deliver about 100 a second, so the limit is 50 Hz and RidePulse analyzes up to about 45 Hz, marking those results as limited.
Every phone has its own faint electronic noise. A 15-second calibration on a table measures it, so vibration that close to it is flagged, not reported.
ISO 2631-1 in plain words
ISO 2631-1 (“Evaluation of human exposure to whole-body vibration, Part 1: General requirements”, 1997, amended 2010) is the international standard for measuring how vibration affects people. For a seated person’s comfort it measures at the seat surface, which is why RidePulse asks for the phone flat on the passenger seat cushion.
The body isn’t equally sensitive at all frequencies. The standard’s weighting curves (Wk for up and down, Wd for side to side and front to back) count vertical vibration most between about 4 and 8 Hz, and horizontal vibration most below about 2 Hz. The weighted values on the three axes combine into one number, the vibration total value a_v, in m/s². When a ride has sharp jolts (a crest factor above 9), the standard says the RMS value alone understates them, so RidePulse also uses the fourth-power vibration dose value, and the score counts those jolts extra.
Annex C of the standard gives rough comfort reactions for a_v. RidePulse maps them onto a 0 to 100 score, always shown next to the measured value:
| ISO 2631-1 comfort reaction | a_v (m/s²), per ISO | RidePulse score |
|---|---|---|
| Not uncomfortable | below 0.315 | 85 to 100 |
| A little uncomfortable | 0.315 to 0.63 | 65 to 85 |
| Fairly uncomfortable | 0.5 to 1.0 | 45 to 65 |
| Uncomfortable | 0.8 to 1.6 | 25 to 45 |
| Very uncomfortable | 1.25 to 2.5 | 10 to 25 |
| Extremely uncomfortable | above 2.0 | 0 to 10 |
The ranges overlap on purpose, because people differ; the score runs through the middle of each. Scores from different roads aren’t comparable between cars, so comparisons with other cars use the idle test, where the road plays no part.
Matching vibration to parts
Parts that turn shake at the rate they turn, so a peak in the spectrum can be matched to a part. This is called order analysis.
| Source | How its frequency behaves | Example |
|---|---|---|
| Wheels and tires | Speed ÷ (π × tire diameter): once per turn (1st order), twice for some tire and rim faults | About 12.8 Hz at 60 mph on a 27-inch tire |
| Engine (four-stroke) | rpm ÷ 60 × cylinders ÷ 2: the firing rate | 25 Hz for a four-cylinder at 750 rpm; 37.5 Hz for a V6; 50 Hz for a V8 |
| Brake rotors | The wheel’s rate, but only while braking | Sweeps down as the car slows |
| Road and body | Not tied to wheel or engine speed | Bounce at 1 to 2 Hz; most felt at 4 to 8 Hz |
RidePulse calls a peak a match when it’s within 5% of the predicted frequency (allowing for the tire squashing under the car’s weight), and gives more confidence when the match holds at two or more speeds. At idle it finds the engine’s firing peak and estimates rpm from it; a peak at half the firing rate can mean one cylinder isn’t pulling its weight. While braking, the wheel frequency drops four-fold in a few seconds, so instead of a fixed spectrum it follows the wheel’s turning angle from GPS speed.
On iPhone, a V8’s firing rate at idle sits right at the 50 Hz edge; on most Android phones it’s well inside the range.
What a phone can’t tell you
- Which corner. One phone in the middle of the car hears all four wheels.
- What’s inside a part. It can say a vibration follows the wheels; a mechanic finds the bent rim or worn joint.
- Noise. RidePulse doesn’t use the microphone, so hums and growls (often wheel bearings) aren’t measured.
- The driveline, yet. RidePulse doesn’t track driveshaft speed, which needs the gear ratios.
- Lab-grade numbers across phones. Results depend on the phone, where it lies and the road, so compare like with like.
Research and real-world use
Using a phone this way isn’t new. A 2015 MIT paper (Siegel et al.) detected wheel imbalance from a dashboard-mounted phone’s accelerometer, with more than 90% accuracy on its test data from two cars. A truck maker’s 2018 service article on file with NHTSA tells technicians to check wheel-end and driveline vibration with a phone app: phone flat, top forward, a second person operating it. Professional apps such as NVH work the same way.
Safety notes
- A passenger runs any test while moving; the driver keeps to the limit. Never brake hard with traffic behind you, and run idle tests outdoors.
How RidePulse can help you measure it
RidePulse puts this behind a few guided, spoken tests. Each checks its own conditions (flat, still, a fast enough sensor, a GPS fix better than 20 m, a steady speed) and asks you to try again rather than give a misleading result. The idle test and comfort drive are free.
RidePulse is a screening tool, not a professional diagnosis. Results depend on phone, placement and road. Have a qualified mechanic inspect any suspected fault.
Questions
Can a phone measure car vibration accurately?
A modern phone’s accelerometer is good enough to measure the vibration you feel and to see clear peaks from wheels and engines, as research and professional apps show. It isn’t a calibrated lab instrument: results depend on the phone and where it lies, so compare like with like.
What is a good ride comfort score?
In RidePulse, 85 or more matches ISO 2631-1’s “not uncomfortable” range (a_v below 0.315 m/s²). Rough roads lower any car’s score, so compare scores on the same road, or use the idle test.
Why are iPhone results marked limited?
iPhones report motion about 100 times a second, so they can’t see vibration faster than 50 Hz. RidePulse analyzes up to about 45 Hz on iPhone and marks the result so you know.
Sources
- ISO 2631-1:1997, Mechanical vibration and shock: Evaluation of human exposure to whole-body vibration (ISO catalog page (amended 2010))
- Siegel, Bhattacharyya, Sarma, Deshpande: Smartphone-Based Wheel Imbalance Detection (2015) (MIT DSpace)
- Determine wheel speed or driveline vibration using a smart phone or tablet (service article, 2018) (NHTSA document)