Summary: Vadim Konradi of Silicon Labs reviews the various EMI problems associated with projected-capacitance touchscreens in today’s portable devices. The author then outlines design and optimization techniques to deal with the interference coupling paths.
A projected-capacitance touchscreen is capable of precise touch location based on a light finger touch to the screen. It determines finger position by measuring miniscule changes in capacitance. Developing a mobile handheld device with a touchscreen interface can be a complex design challenge, especially for projected-capacitance touchscreens, which represent the current mainstream technology for multi-touch interfaces. A key design consideration in this type of touchscreen application is the impact of electromagnetic interference (EMI) on system performance. In this article we’ll explore the sources of interference-caused performance degradation that can negatively impact touchscreen designs and how to mitigate their effects.
Projected capacitance touchscreen geometry
A typical projected-capacitance sensor is assembled to the underside of a glass or plastic cover lens. Figure 1 shows a simplified edge view of a two-layer type sensor. Transmit (Tx) and receive (Rx) electrodes are drawn in transparent indium tin oxide (ITO), forming a matrix of crossed traces, each Tx-Rx junction having a characteristic capacitance. The Tx ITO lies below the Rx ITO, separated by a thin layer of polymer film and/or optically-clear adhesive (OCA). As shown, the Tx electrode runs from left to right, and the Rx electrode runs into the page.
Figure 1: Sensor geometry reference
Sensor normal operation
The operator’s finger is nominally at ground potential. The Rx is held at ground potential by the touchscreen controller circuit, and the Tx voltage is varied. The changing Tx voltage induces current flow through the Tx-Rx capacitance. A carefully-balanced Rx integrating circuit isolates and measures the charge movement into the Rx. This measured charge indicates the “mutual capacitance” linking Tx and Rx.
Sensor condition: not touched
Figure 2 indicates flux lines in the untouched condition. Without a finger touch, the Tx-Rx field lines occupy considerable space within the cover lens. These fringing field lines project beyond the electrode geometry — thus the term “projected capacitance.”
Figure 2: Flux lines untouched
Sensor condition: touched
When a finger touches the cover lens, flux lines form between the Tx and finger, displacing much of the Tx-Rx fringing field, as shown in Figure 3 . In this manner, the finger touch reduces Tx-Rx mutual capacitance. The charge measurement circuit recognizes this changed capacitance (delta C) and the presence of a finger over the Tx-Rx junction is detected. A map of touch across the panel is generated by making delta C measurements at all intersections in the Tx-Rx matrix.
Figure 3: Flux lines touched
Figure 3 demonstrates an important additional effect: capacitive coupling between the finger and the Rx electrode. Through this path, electrical interference may couple onto the Rx. Some degree of finger-Rx coupling is unavoidable.
Interference in projected capacitance touchscreens is coupled through parasitic paths that are not entirely intuitive. The term “ground” is commonly used interchangeably in reference to either the DC circuit reference node or a low-resistance connection into planet Earth. These are not the same terms. In fact, for a portable touchscreen device, this difference is the essential cause of touch-coupled interference. To clarify and prevent confusion, we’ll use the following terminology when assessing touchscreen interference.
- Earth – connection to planet Earth, for instance via the earth pin of a 3-pin AC mains socket
- Distributed Earth – capacitive connection of an object to earth
- DC Ground (GND) – DC reference node of a portable device
- DC Power – Battery voltage of a portable device. Alternately the output voltage of a charger connected to the portable device, e.g. 5V Vbus for a USB-interface charger.
- DC VCC – regulated voltage which powers the portable device electronics, including LCD and touchscreen controller
- Neutral – AC mains return, nominally at earth potential
- Hot – AC mains voltage, energized with respect to neutral
LCD Vcom coupling to the touchscreen receive lines
Theportable device touchscreen may be mounted directly over an LCDdisplay. In a typical LCD configuration, a liquid crystal material isbiased between upper and lower transparent electrodes. The lowerelectrodes define the individual pixels of the display. The upper commonelectrode is a continuous plane across the visible front of thedisplay, biased at voltage Vcom. The AC Vcom voltage, as implemented in atypical low-voltage portable device such as a cell phone, is a squarewave oscillating between DC ground and 3.3V. The AC Vcom plane typicallyswitches once per display line, so the resultant AC Vcom frequency isone half the display frame refresh rate multiplied by the number oflines. A typical portable device AC Vcom frequency might be 15 kHz. Figure 4 shows how the LCD Vcom voltage couples into the touchscreen.
Figure 4: LCD Vcom interference coupling model
A two-layer touchscreen is implemented with the Tx and Rxarrays on separate ITO layers, spaced by a dielectric layer. The Txtraces occupy the full width of the Tx array pitch, separated only bythe minimum trace-trace gap required for manufacturing. This type ofconstruction is referred to as self-shielded because the Tx arrayshields the Rx array from LCD Vcom. However, there is still potentialfor coupling to occur through the gaps between Tx strips.
For economy of construction and to achieve bettertransparency, a single-layer touchscreen implements the Tx and Rx arrayson a single ITO layer, with individual discrete bridges applied tocross one array over the other. As a result, the Tx array does not form ashielding layer between the LCD Vcom plane and the sensor Rxelectrodes. This represents a potentially severe Vcom interferencecoupling situation.
A potential source oftouchscreen interference is the switching power supply in amains-powered cell phone charger. Interference is coupled through thefinger to the touchscreen, as shown in Figure 5 . Small cell phonechargers typically have AC mains hot and neutral inputs but no earthconnection. The charger is safety-isolated, so there is no DC connectionbetween the mains input and the charger secondary. However, there isstill capacitive coupling through the switching power supply isolationtransformer. The return path for charger interference is through thefinger touching the screen.
Note that charger interference in this context is voltageapplied to the device with respect to earth. The interference may bedescribed as “common mode” as it appears equally on DC ground and DCpower. Power supply switching noise appearing between the charger outputDC ground and DC power could be a problem for the touchscreen operationif not adequately filtered. This power supply rejection ratio (PSRR) isa separate issue, which is not addressed in this scenario.
Charger coupling impedance
Charger switchinginterference is coupled by the transformer primary-secondary windingleakage capacitance on the order of 20 pF. The effect of this weakcoupling is offset by parasitic shunt capacitance to distributed earthoccurring in the charger cable and in the powered device itself. Holdingthe device in the hand applies more shunting, often enough toeffectively short the charger switching interference and preventinterference with touch operation. A worst-case charger-generatedinterference situation occurs when the portable device is connected tothe charger and placed on a desktop, and the operator’s finger contactsonly the touchscreen.
Charger switching interference component
Typicalcell phone chargers use a flyback circuit topology. The interferencewaveform they generate is complex and varies considerably betweenchargers, depending on circuit details and output voltage controlstrategy. The interference amplitude varies considerably depending onhow much design effort and unit cost the manufacturer has allocated toshielding in the switching transformer. Typical parameters include:
- Wave shape: complex, consisting of pulse-width modulation square wave followed by LC ringing
- Frequency: 40 – 150 kHz under nominal load, with pulse-frequency or skip-cycle operation dropping frequency to < 2 kHz when very lightly loaded
- Voltage: up to one half mains peak voltage = Vrms / sqrt(2)
Charger mains interference component
Inside thecharger front end, the AC mains voltage is rectified to generate thecharger high voltage rail. As a result, the charger switching voltagecomponent is riding on a sine wave of one half the mains voltage.Similar to the switching interference, this mains voltage is alsocoupled through the switcher isolation transformer. At 50 or 60 Hz, thiscomponent is much lower frequency than the switching frequency, so itseffective coupling impedance is proportionally higher. The importance ofmains voltage interference depends on the character of shunt impedanceto earth and on the touchscreen controller sensitivity to low frequency.
Mains interference special situation: 3-pin plug with missing earth
Poweradapters rated for higher power, such as laptop PC AC adapters, may beequipped with a 3-pin AC mains plug. To suppress EMI on the output, thecharger will likely have the mains earth pin connected internallythrough to the output DC ground. Such chargers typically connectY-capacitors from mains line and neutral to earth to suppress conductedEMI on the mains. Provided the earth connection is present as intended,this type of adapter does not create an interference problem for thepowered PC and a USB-connected portable touchscreen device. Thisconfiguration is represented by the dotted box in Figure 5 .
A special case charger interference situation occurs for aPC and its USB-connected portable touchscreen device if the PC chargerwith 3-pin mains input is plugged into a mains socket with no earthconnection. The Y-capacitors couple the AC mains through to the outputDC ground. The relatively large values of the Y-capacitors couple themains voltage very effectively, resulting in a large mains frequencyvoltage coupled at relatively low impedance through the finger on thetouchscreen.
Projected-capacitance touchscreenscommonly used in today’s portable devices are vulnerable toelectromagnetic interference. The interference voltages are coupledcapacitively from sources that are both internal and external to thetouchscreen device. These interference voltages cause charge movementwithin the touchscreen, which may be confused with the measured chargemovement due to a finger touch on the screen. Effective design andoptimization of the touchscreen system depends on understanding theinterference coupling paths and mitigating or compensating for them asmuch as possible.
Interference coupling paths involve parasitic effects suchas transformer winding capacitance and finger-device capacitance. Proper modeling of these effects yields a detailed understanding of thesources and magnitudes of the interference.
For many portable devices, the battery charger can be akey source of touchscreen interference. The charger interferencecoupling circuit is closed through the capacitance of the operator’sfinger on the touchscreen. The quality of the charger’s internalshielding design and the presence of a proper charger earth connectionare key factors affecting charger interference coupling.
Vadim Konradi is Staff ApplicationsEngineer, Silicon Labs. He joined Silicon Labs in 2010 as a staffapplications engineer, developing human interface devices. Previously,he has worked in companies ranging from startups to large multinationalcorporations in a variety of technical and management positions focusedon occupancy sensors, automatic lighting controls, hospitality Internet,projectors, mainframes and astronaut tools. Mr. Konradi has workedacross a variety of technical disciplines including embedded, analog andpower electronics, optics, audio and acoustics. His engineeringexperience includes taking products from concept and requirementsthrough R&D, commercialization, sales, manufacturing and fielddeployment. He holds a BS in electrical engineering from the Universityof Texas and an MS in electrical engineering from Berkeley. He haspatents in sensors, power conversion and lighting controls.