How Touch-Screen work

While touch screens have become more prevalent in everyday electronics, few people understand how they really work. From the capacitive change from a human finger to the final motion on screen, the technical aspects of touch screens remain a mystery to most users. In fact, most users understanding of a touch screen consists of either “it works well” or “it doesn’t work well.” The question is how to understand the system well enough to determine what makes a touch screen work well or poorly. In order to understand this better, we would like to de-mystify how a touch screen registers the human touch, expose critical aspects of touch screen construction, and to show how touch motions on the touch screen are interpreted and result in screen action. 

What’s Inside the Touch Screen?

A capacitive touch screen is commonly constructed of several layers, called the “stackup.” The top layer is a protective layer composed of glass, or PMMA (commonly called Plexiglas or acrylic) with an anti-scratch coating. This layer is often shaped and printed with the company logo, button indicators, and is the outward most facing material of the phone or touch product.  Directly underneath this top layer is a thin layer of optically clear adhesive and then the exciting part—the touch-sensing layers.

The primary technical “secret” of the touch sensor layers is that they are constructed of indium-tin-oxide (ITO). The ITO is optically transparent, even though it is a ceramic. Typical capacitive ITO touch screens are ~92% transparent, which is a very large selling feature. Below the ITO sensing layers is an air gap or spacing that helps separate these touch sensing layers of ITO from the electrical noise of the bottom most layer—the LCD electronics. In Figure 1, there are two layers of ITO shown. Touchpanel vendors will use one, two, or even three layers of ITO depending on the specific product design considerations. Ultimately, however, the ITO layering is chosen depending on the touchpanel supplier’s technical capabilities as well as technical requirements from the customer. 

Figure 1: Cross-sectional view of capacitive touch screen “stackup”

If you examine a touch screen device from the right angle (with the LCD turned “off”), you may even be able to see the ITO pattern in the glass. Today, typical touch screens use a diamond pattern as shown in Figure 2. Diamonds have been chosen as a “best practice” for design because the ITO patterning must meet several requirements. First, the ITO layer must cover the whole screen, so that there is no visible variation in optical clarity. Second, it is easiest for system designers if the pattern corresponds to an X/Y grid—writing software that comprehends X and Y locations is simpler and more logical when making LCD update decisions. Finally, the pattern must provide continuity across the entire electrode. Diamonds with small bridges between them meet all of these needs. 

Figure 2: ITO patterning on a typical touch device provides X & Y touch information

The Physics of Touch

The key to making a capacitive touch screen work is that the system is designed to detect minute changes in capacitance introduced by a human finger. In fact, as a finger or other conductive object approaches the screen, it creates a parallel-plate capacitor between the sensors and the finger. This capacitor is small relative to the others in the system (about 0.5 pF out of 20 pF), but it is measurable with the right technology and filtering.  

A parallel-plate capacitor is composed of two conductive surfaces with an insulator between them. The ITO layer is a conductor, a user’s body is a conductor, and the glass or PMMA is an insulator. When a user touches the screen, they’ve become part of a capacitor. This same phenomenon is used in Cypress’s CapSense devices, which provide buttons and sliders in products from cell phones to washing machines.

One mechanism used to sense capacitance is Cypress’s CSD (CapSense with Sigma Delta). In this method, the capacitor is measured by a mechanism similar to a Sigma Delta ADC. The capacitor is connected to a bleed resistor (RBleed). If the voltage across the capacitor is above a threshold, a fixed charge is added to the cap. The circuit then measures the amount of time (counts) required to discharge the cap before the threshold is hit again. These counts are averaged over time to obtain an accurate capacitance measurement.

The CSD block diagram in Figure 3 appears complex at first, but it can be understood. Let’s start on the left hand side. The PRS (Pseudo-Random Source) causes Switch 1 (SW1) and Switch 2 (SW2) to alternate connections in a break-before-make pattern. This means that CSENSOR will be connected to VDD, then connected to CEXT over and over again. CEXT is an external capacitor used to integrate charge from CSENSOR. Every time CSENSOR is connected to VDD, it fills with charge. When CSENSOR is connected to CEXT , it charges CEXT. Note that CSENSOR is much smaller than CEXT. Imagine a bicycle pump and a tire. Each time SW1 is closed, the handle is pulled up and the air cylinder is filled. Every time SW2 is closed, the pump handle is pushed down and the air moves from the pump to the tire.

Figure 3: Capacitive charge circuit for sensing capacitive change and finger detection

The circuit elements in the middle portion of Figure 3 discharge CEXT and measure CSENSOR. When the voltage at CEXT is above VREF, SW3 is closed, which discharges CEXT. Think of this discharge as a safety valve on the bicycle tire. One important feature of this safety valve is that it discharges at a known rate.  Once we measure the discharge time, we know CSENSOR.

Figure 3: Capacitive charge circuit for sensing capacitive change and finger detection

The right-hand portion of the diagram is simply measuring the amount of time that SW3 is closed. The PWM provides a pause between scans to allow the processor to read samples. The counter measures time whenever SW3 is closed. As the capacitance increases at CSENSOR the charge sent to CEXT increases, so the amount of time that SW3 is closed must also increase.

It is this capacitive change that allows the touch screen controller to recognize that a touch (resulting in an increase in sensed capacitance) has occurred. The problem, of course, (and one of the key contributors to a “good” or “bad” touch screen), is the ability to tell a true finger touch from environmental noise. Noise in the system can really cause problems. If the touch controller is not able to distinguish an actual finger touch from just a capacitive noise spike, the touch screen will not react properly.

Finger Location Determination

Once the system can properly recognize a change in charge due to a finger touch, the exact location and motion path can be determined. Measuring the capacitance of all sensors in the touch screen produces information on exactly which sensors are active and the magnitude of capacitance change. Even though the screen may only contain an X/Y grid of 10x15 sensors, users want the system to report touch location accurately even within a millimeter. A simple interpolation technique can be used to determine finger position to a very fine degree. For example, if sensors 1, 2 and 3 see signals of 3, 10 and 7, the center of the finger is at (1*3+2*10+7*3) / (3+10+7) = 2.2. This value is normally scaled to match the pixel resolution of the LCD before it’s reported to the host CPU.

In a real design, there are several factors that complicate matters. When the finger reaches the edge of the panel, the interpolation breaks down because there is no “next sensor” to average into the computation. One method used to mitigate this effect is to assume that all finger touches produce the same total capacitance change. In the example used above, the finger created a capacitance change of 20 units. If a touch near the edge is measured with a capacitance change of only 15 units, it can be presumed that the remaining 5 units are over the edge.  

Additionally, customers may have unique industrial design requirements that dictate that a touch screen does not detect a touch when finger capacitance is witnessed on the touch screen edge (perhaps from holding the phone edges). In designs such as this, the touch screen controller must be flexible enough to be programmed to recognize these types of touches and to ignore them. 

Communication to the User

Now that a valid touch signal is present and the X/Y coordinates of the touch are known, it’s time to get the data to the host CPU for processing. Most touch screen devices communicate using the venerable I2C interface. The touch screen controller IC uses I2C (or SPI if desired) to pass X/Y or gesturing data to the product’s host processor where the command of product operation takes place. An example flow of this data transfer is shown in Figure 4.

Figure 4: Data processing flow from touch screen to host processor

In a real system, most of these touch detection and feedback operations take place in parallel because the touch screen is in constant detection mode when being operated by a user. As the touch data is collected, the touch controller device can either process the data to interpret a finger motion (or gesture) itself, or it can pass the raw data through a low-level driver to the host operating system for interpretation and action—and the rest is up to the user interface designers. Creating unique human interaction and exciting features that rely on touch will be an exciting place to watch for new development ideas in the near future. 

The fact that touch screens are becoming more and more popular is not a mystery. Consumers continue to call for cell phone, notebook, PC monitor, netbook, MP3 player and IP phone manufacturers to produce more versions of touch screen-based devices. With a more clear understanding of how capacitance change is sensed, how a finger location is determined and refined, and how the finger location data is passed up through the host system for operation, developers will be more able to supply innovative touch screen technology to meet the ever-increasing demand from users for this compelling interface.