How a touchscreen feels your finger: a grid of tiny capacitors

Your fingertip never touches anything electrical when you tap a phone. It rests on glass. Under the glass is a grid of transparent wires, and at every crossing of the grid sits a tiny capacitor. Your finger, a lump of salty, conducting tissue, bends the electric field at the crossings nearest to it. The phone finds you by measuring which crossings changed, and by how much.

Why it is worth a look

The change being measured is absurdly small. Gary Barrett and Ryomei Omote, writing in the Society for Information Display’s magazine in 2010, say these projected-capacitive screens “must sense changes in capacitance as small as a few femtofarads”. A femtofarad is a thousandth of a picofarad, or 10⁻¹⁵ farad. Your whole body’s capacitance to earth is 100 to 200 picofarads for an adult, according to Microchip’s touch-sensor design guide. A 3 femtofarad change is about one 33,000th of 100 picofarads (our arithmetic). Signals that faint are easily swamped by electrical noise, and the same authors note this is what makes the screens hard to build into a display.

A capacitor you can disturb from outside

A capacitor is two conductors separated by an insulator. Put a voltage across it and charge collects on the two sides. Its capacitance says how much charge collects for each volt. For two flat plates it is ε₀ε_rA/d: bigger plates hold more, a thicker gap holds less, and the insulator’s permittivity ε_r scales it.

The Microchip note, by Feargal Cleary, treats your fingertip as one such plate. An 8 mm disc of finger over an electrode, through 1 mm of plastic with ε_r = 2, gives 0.89 picofarad. That is for a single large button. The crossings of a screen grid are far smaller, and the changes Barrett and Omote say they must sense are a few femtofarads.

The finger’s capacitor does not stand alone. It sits in series with your body (100 pF, say) and with the coupling between the device’s ground and the earth. In series, the smallest capacitor dominates. Cleary’s worked example: a 1 pF finger, 100 pF body and 100 pF ground coupling combine to 0.98 pF. Drop the ground coupling to 2 pF, as in a small battery-powered device, and the total falls to 0.662 pF, about a third less signal. Nothing here is a closed circuit through your body. Every link in the chain is a capacitor, an insulator with charge on either side.

Rows, columns, and the crossing between them

The transparent conductors are laid out in two layers, one set of rows and one set of columns, separated by an insulator. Where a row crosses a column the two form a capacitor. Barrett and Omote describe how the controller reads them: it “drives a single column (Y) and then scans every row (X) that intersects with that column, measuring the capacitance value at each X-Y intersection”, then moves to the next column. Their typical phone has nine columns and 16 rows, 144 crossings; current controllers scan at roughly 20 to 200 times a second.

When a finger comes near a crossing, their account is that the body “‘steals’ some of the charge”. Cleary explains the two effects at work. The finger adds coupling between row and column, but it also opens a path to ground that drains away charge the column was sending to the row. The second effect usually wins, so a touch shows up as a fall in the crossing’s capacitance. Water, by contrast, raises it, which is one way these sensors tell a drop from a finger.

From 144 crossings to a precise point

Each crossing has a stored baseline, its value with nobody touching. The controller subtracts the baseline to get a touch delta, and a crossing counts as touched when that delta passes a threshold. Cleary says the threshold is often set at half the largest delta. A fingertip spans several crossings, so several of them change by different amounts. Barrett and Omote write that the controller uses those neighbouring values to interpolate, and that the resolution is “usually at least 1024 × 1024”. That is far finer than a grid of 9 by 16 wires. The precision comes from comparing neighbours, not from packing in more wires.

Measuring each crossing separately is also what lets a screen follow several fingers. The other method, self-capacitance, measures each whole row and each whole column against ground. Two fingers on a diagonal then light up two rows and two columns, and the controller cannot tell which row goes with which column. Barrett and Omote call the two false answers “ghost points”.

What it does not do

Nothing is pressed. This is not the analog-resistive screen that Barrett and Omote compare it with, and the grid does not read how hard you push or the ridges of your skin. It reads field. They note that the screen does not actually have to be touched, and that the sensing range can be extended to work through cotton or surgical gloves. A thicker glove keeps the finger farther from the glass, and the plate formula above shows the capacitance shrinking as that gap grows (our reading).

In short

Under the glass, rows and columns cross at a grid of tiny capacitors. A finger nearby lowers the capacitance of the crossings closest to it by a few femtofarads. The controller scans every crossing, subtracts its baseline, keeps the ones above a threshold, and weighs the neighbours against each other to place your finger far more finely than the wires are spaced.