
Touch Screen Settling
Temperature Measurement
A/D
Converter
MUX
X+
Temperature Select
TEMP0
TEMP1
www.ti.com ........................................................................................................................................................................................... SBAS454 – DECEMBER 2008
In some applications, external capacitors may be required across the touch screen for filtering noise picked up by
the touch screen (i.e., noise generated by the LCD panel or backlight circuitry). These capacitors provide a
low-pass filter to reduce the noise, but they also cause a settling time requirement when the panel is touched.
The settling time typically shows as a gain error. The problem is that the input and/or reference has not settled to
its final steady-state value prior to the ADC sampling the input(s) and providing the digital output. Additionally, the
reference voltage may still be changing during the measurement cycle.
To resolve these settling time problems, the TSC2003 can be commanded to turn on the drivers only without
performing a conversion (see
Table 1). Time can then be allowed before the command is issued to perform a
conversion. Generally, the time it takes to communicate the conversion command over the I2C bus is adequate
for the touch screen to settle.
In some applications, such as battery recharging, a measurement of ambient temperature is required. The
temperature measurement technique used in the TSC2003 relies on the characteristics of a semiconductor
forward diode voltage (VBE) has a well-defined characteristic versus temperature. The temperature can be
predicted in applications by knowing the 25°C value of the VBE voltage and then monitoring the delta of that
voltage as the temperature changes. The TSC2003 offers two modes of temperature measurement.
The first mode requires calibrations at a known temperature, but only requires a single reading to predict the
ambient temperature. A diode is used during this measurement cycle. The voltage across the diode is connected
through the MUX for digitizing the diode forward bias voltage by the ADC with an address of C3 = 0, C2 = 0, C1
= 0, and C0 = 0 (see
Table 1 and
Figure 8 for details). This voltage is typically 600 mV at 25°C, with a 20-
A
current through it. The absolute value of this diode voltage can vary a few millivolts; the temperature coefficient
(TC) of this voltage is very consistent at –2.1 mV/°C. During the final test of the end product, the diode voltage
would be stored at a known room temperature, in memory, for calibration purposes by the user. The result is an
equivalent temperature measurement resolution of 0.3°C/LSB.
Figure 8. Temperature Measurement Mode Functional Block Diagram
The second mode does not require a test temperature calibration, but instead uses a two-measurement method
to eliminate the need for absolute temperature calibration and for achieving 2°C/LSB accuracy. This mode
requires a second conversion with an address of C3 = 0, C2 = 1, C1 = 0, and C0 = 0, with a 91 times larger
current. The voltage difference between the first and second conversion using 91 times the bias current is
represented by kT/q × 1n (N), where N is the current ratio (91), k is Boltzmann's constant (1.38054 × 10–23
electron-volts/degree Kelvin), q is the electron charge (1.602189 × 10–19 C), and T is the temperature in degrees
Kelvin. This mode can provide improved absolute temperature measurement over the first mode, but at the cost
of less resolution (1.6°C/LSB).
Copyright 2008, Texas Instruments Incorporated
15