Smartphone Touch Screen Analysis Tests Finger Fidelity

Illustration for article titled Smartphone Touch Screen Analysis Tests Finger Fidelity

MOTO Development Labs devised a simple method of analyzing capacitive touch screens using drawing programs. They put the iPhone, the Nexus One, the Droid, and the Droid Eris through the paces and proved not all touch screens are created equal.


Using only your fingers and a drawing app, MOTO shows how you can test out the accuracy of your smartphone's touch screen. The test is simple: draw some slow, steady lines across the screen with your finger. If they're smooth and straight, your touch screen is tracking with relative accuracy. If they're wavy or jagged, your phone might not be giving your fingers the attention they deserve.

MOTO's test showed the iPhone tracking the most accurately of the four, with smooth, straight lines. The Motorola Droid fared worst of the bunch, its crossing lines tracking so jaggedly that the screen looked like a jigsaw puzzle. The Eris and the Nexus One landed somewhere in between.

If jagged lines are the symptoms of a subpar touch screen, MOTO suggests that the affliction can be any combination of too large a sensor, too low a touch-sampling rate, or too inaccurate an algorithm. [MOTO Development Labs - Thanks Sabrina]


The test is only somewhat accurate for certain types of input and usage.

For prolonged motion along a line predictive systems do their best to follow finger motion using interpolation to avoid unnecessary overhead in processing. Since resistive touch screens need to average their inputs based on voltage dividers, this can cause software latency issues for pixel-accurate tracking (especially based on variances in humidity, skin moisture, and texture such as callouses, etc).

Touch screen software also measures fixed locality in contrast to edges of soft buttons, and this is where resistive plays its strengths. The software can account for inaccuracy by using calculations for likelihood of user intent, based on the region most largely affected, or prior action taken.

The smoother a line is, on a test such as this, would most likely be indicative of more software interpolation, and less about the touch screen quality itself. I'd be more interested in seeing a test that measured the latency of touch input, the minimum vs. maximum touch input registered, and the polling time between voltage and resistance division.

Varying software methods implemented per app can also make a huge difference, as well as the amount of CPU overhead being utilized for other applications and background tasks. I would need to see the source code used, the methods for constituting light vs. medium touch, and accurate measurements of the various screens resistive properties before I could lend this so-called "test" any credence.