Single-ended amplifier-based touch readout circuit with immunity to display noise

To suppress errors in a touch readout circuit due to display noise, differential sensing using fully differential amplifiers is widely used. However, conventional differential sensing methods require additional circuits or increase circuit complexity, thus increasing power consumption and circuit area or requiring additional dummy RX line on the touch panel. In this study, we propose a compact touch readout circuit composed of a single-ended amplifier while keeping the display noise suppression of differential sensing so as to minimize power consumption and circuit area increase and avoid the need for additional dummy RX line. The proposed touch readout circuit for is fabricated in circuit area of 0.4 mm2 with 0.35 µm 3.3 V CMOS process and measured for 10.1-inch touch panel on top of a TFT-LCD panel. With the proposed touch readout circuit, the signal-to-noise ratio (SNR) is improved by up to 13.1 dB compared to the conventional touch readout circuit using single-ended amplifiers.


Introduction
The touch readout circuit, which reads out the touch signal from the capacitive touch panel, must be robust to display noise, because large display noise is injected from the operating display panel [1][2][3][4][5][6][7]. In particular, as the add-on type touch panel has advanced into an on-cell and further, an in-cell type touch panel, the gap between the display panel and the touch panel has been greatly decreased, and the display noise injected into the touch panel has become a more serious problem [3][4][5][6][7][8][9][10]. As shown in Figure 1, display noise is injected to the RX electrode of the touch panel through the common electrode located at the top of the display panel (for example, the cathode electrode in a top-emission OLED panel). Therefore, display noise is considered a common noise independent of the position in the panel [1,[3][4][5]8,11,12]. Therefore, the display noise can be effectively suppressed by using a differential sensing method [1][2][3]6,[11][12][13]. There are two common circuit implementations to realize the differential sensing method. One is to additionally use a fully differential amplifier that receives the outputs of the pair of readout amplifiers of two adjacent channels as differential inputs [6,12]. However, the added circuits increase the power consumption and the circuit area. The other is to design each channel's readout amplifier as a fully differential amplifier [1][2][3]11,13]. This circuit, however, can only detect the difference in capacitance between the adjacent two channels, then it is not sufficient for touch detection. To determine a touch, an additional dummy RX line is needed as the reference for the touch panel.
In this study, we propose a compact touch readout circuit that consists of a single-ended amplifiers instead of fully differential amplifiers, but at the same time effectively suppresses errors in a touch readout circuit due to display noise. The proposed compact touch readout circuit neither increases in circuit area or power consumption nor requires additional dummy RX lines.
The rest of this paper is organized as follows: Section 2 describes the principle of the proposed touch readout circuit. Section 3 shows the simulation and measurement results. This paper is concluded in Section 4. Figure 2 shows the effect of display noise on the conventional touch readout circuit composed of single-ended amplifiers. In Figure 2, C M [N] represents the mutual capacitance between the RX line of the N-th channel and a TX line running perpendicular to the RX line. C D is the parasitic capacitance between the RX line and the common electrode. C P represents all the other parasitic capacitances between the RX line and the other electrodes. In the conventional single-ended amplifier-based touch readout circuit, the positive input of each amplifier is connected to the DC voltage, V REF , so the display noise introduced through the common electrode appears at the output of the touch readout circuit, as shown in Figure 2.

Principle of the proposed touch readout circuit
To suppress the display noise, fully differential amplifiers are widely used to suppress the display noise, and the display noise is injected equally into both the positive and negative inputs so that it does not appear at the output of the touch readout circuit by commonmode rejection. In this study, we propose the use of compact single-ended amplifiers in which the display noise is still injected into both the positive and negative inputs of each single-ended amplifier. This allows a touch readout circuit to be constructed with singleended amplifiers, which is relatively simple compared to fully differential amplifiers, while suppressing the display noise at the output of the touch readout circuit. The concept of the proposed touch readout circuit is shown in Figure 3, where the positive input of any amplifier except for the first channel's amplifier is connected to the RX line of the adjacent channel. The first channel's amplifier must have its positive input connected to a DC voltage. Otherwise, the positive input of any amplifier is undetermined because the voltage of the RX lines is undefined. In the proposed touch readout circuit configuration, two adjacent RX lines are connected to the positive and negative inputs of the amplifier, respectively, so the display noise is commonly injected into the two inputs. As mentioned above, because the display noise is independent of the position in the touch panel, the noise applied on the two RX lines are considered the same. Therefore, the display noise does not appear at the output of the touch readout circuit, because commonmode rejection is also possible even with a single-ended amplifier.

Analysis and design of the proposed touch readout circuit
The display noise suppression of the proposed touch readout circuit can be analyzed as follows: The voltage change of common electrode is denoted as V COM (1).
In (1), the display noise through TX lines is not considered because TX lines are driven by TX drivers. This assumption is found in many literatures, for example [2,4,11]. As shown in Figure    the amplifiers are denoted as A, V OUT [1] is obtained as (2).
On the other hand, the positive input of the amplifier in the second channel in Figure 3 is connected to RX [1], which is the RX line of the first channel. The output voltage of the second channel, V OUT [2] and V OUT [2] are obtained as (5) and (6), respectively.
In the same way, the (N-1)-th channel and the N-th channel, V RX[N−1] and V RX[N] are obtained as (11) and (12), respectively.
V OUT[N] of the N-th channel can be obtained as (13) in the same way as (6), and if we put V RX[N−1] in (11) and V RX[N] in (12) into (13), V OUT[N] is expressed as (14).
If the open-loop gain of the amplifier (A) for V COM is not much high, α in (14) is smaller than one and is multiplied to the power of N, which suppresses the display noise. Whereas, in the case of conventional single-ended amplifier-based readout circuit, V OUT for all channels is the same as when N = 1 in (14), V OUT for V COM of the proposed circuit decreases with N. On the other hand, V OUT for V TX can be easily obtained from (14) by substituting the injected charge due to display noise (C D V COM ) with the injected charge due to V TX (C M V TX ). Since the frequency of the V TX is higher than V COM , the open-loop gain of the amplifier (A) for V TX is higher than the open-loop gain (A) for V COM . If A is sufficiently high, α for V TX is close to one, and the signal due to V TX is not suppressed. Meanwhile, V OUT for V TX of the proposed circuit which is the touch signal is similar to that of the conventional circuit, thus the output of the proposed circuit depends on the mutual-capacitance (C M ), not the difference of the mutual-capacitances of the adjacent RX lines.
Since V OUT for V COM of the proposed circuit decreases with N, this means that the display noise suppression is effective for channels far from the first channel but might be insufficient for channels close to the first channel. To address this problem, the proposed touch readout circuit uses a bidirectional scanning method, where a frame is divided into two subframes and reverses scanning direction between the subframes. More specifically, in the first subframe, V REF is connected to the first channel (light-green in Figure 4), and the outputs of channels farther away from the first channel than the (N/2)-th channel are used as valid outputs of the touch readout circuit. Conversely, in the second subframe, V REF is connected to the N-th channel which is the last channel (light-blue in Figure 4), and RX[i + 1] is connected to the positive input of the amplifier in the i-th channel, so that the output of the (N/2)-th channel and the outputs of the channels closer to the first channel than the (N/2)th channel are used as valid outputs of the touch readout circuit. Throughout the first and second subframes, all outputs of the touch readout circuits can be obtained with the display noise suppressed.

Simulation and measurement results
Display noise suppression of the proposed touch readout circuit was measured with Lenovo's 10.1-inch add-on type mutual capacitive touch panel. Also, for SPICE simulation, the circuit model of the touch panel is configured  as shown in Figure 5 [5,14], and the circuit parameters in Table 1 were obtained through the measurement of the touch panel. Therefore, C D in (14), which is the capacitance between one RX line and the common electrode, is 4.3 pF because it is composed of as many numbers of C RX (100 fF) as TX channels (43 channels).
The simulation results are shown in Figure 6. The display noise used in the simulation was collected from the LCD panel of a 13.1-inch laptop (ASUS, U35JC-RX006V) displaying a full-white screen. In Figure 6(a), the display noise is modeled as a 500 mV impulse with a frequency of 48 kHz and 160 mV impulse with a frequency of 96 kHz. In the first subframe in which the positive input of the first channel is fixed to V REF , as described in sub-section 2.2, the display noise is not sufficiently suppressed at the output voltages of the close channels to the first channel. In Figure 6(b), it is clearly observed that the output of the first channel is affected by the display noise, while the display noise is sufficiently suppressed at the output voltage of the 14th, the last channel. Therefore, in the first subframe, the output voltages of the eighth to 14th channels, which belong to the half part far from V REF , are accepted as valid touch information. Conversely, as shown in Figure 6(c), in the second subframe in which the positive input of the 14th channel is fixed to V REF , the output voltages of the first to seventh channels, which belong to the half part far from V REF , are accepted as valid touch information. The results of the simulated signal-tonoise ratio (SNR) will be presented and compared with the measured SNR in Figure 8. Figure 7 is a micrograph of the proposed touch readout circuit of 14 channels fabricated in a 0.35 μm 3.3 V CMOS process. The main part of the fabricated readout circuit is the amplifier, and its specifications are presented in Table 2. To accurately read out the touch signal, the input-referred offset of the amplifier is tightly controlled with an average of 0.4 mV by applying the chopper stabilization technique. The TX drivers are constructed separately on the printed circuit board using discrete devices and generate 5 V TX signals. The proposed touch readout circuit is measured with the add-on type 10.1-inch touch panel mounted on top of the 13.1-inch LCD panel such that display noise is injected into the touch panel. If the output of each channel is measured 100 times each when touched and not touched, the SNR of each channel is calculated as (15) [9,11,15]. Figure 5. Touch panel circuit model for simulation [5,14]. where V OUT_TOUCH [i] is the output voltage of i-th measurement when touched, and V OUT_TOUCH and V OUT_w/oTOUCH are the average of the output voltage of each channel for 100 measurements when touched and not touched, respectively. In Figures 8, the SNR of the proposed readout circuit for the full white and the zebra-pattern screens, both simulated and measured, are presented. Figure 8(a,b) shows the SNR in the first and second subframe, respectively. In the first subframe, V REF is connected to the first channel, and in the second subframe, to the last channel (14th channel). In Figure 8(a), the first channel has the lowest SNR because it has a fixed positive input to V REF in the same way as the conventional single-ended amplifierbased touch readout circuit, but the SNR increases as the channel becomes farther away from the first channel. The SNR for the full-white screen is measured to increase up to 13.1 dB at the last channel compared to the first channel and up to 12.6 dB for the zebra pattern screen. Therefore, in the first subframe, we only accept the outputs of the eighth to 14th channels as valid touch information. In Figure 8(b), on the same principle, the measured SNR increases up to 13.1 and 12.7 dB compared to the last channel for the full-white and the zebra pattern screens, respectively. This is because, in the second subframe, the amplifier of the 14th channel has its positive input fixed at V REF . In the second subframe, we only accept the outputs of the first to seventh channels as valid touch information. Finally, by collecting the touch information from two subframes, the proposed touch readout circuit improves the SNR by a minimum of 6.3 dB to a maximum of 13.1 dB for the full-white screen and by a minimum of 7.2 dB to a maximum of 12.6 dB for the zebra pattern screen. The slightly lower SNR for the zebra pattern screen than that for the fullwhite screen is because the measured display noise for the zebra pattern screen is a maximum of 550 mV, while the measured display noise for the full-white screen is maximum of 500 mV. A 10% larger noise reduces the SNR by 0.8 dB in calculations, which is in good agreement with the measurement results. In addition, the measured SNRs in Figure 8 are lower than the simulated SNRs. This is because the simulations only include display noise, while the measurements also include internal circuit noise, power noise, jitter of the control signal, and other environmental noise. The SNR of the fabricated proposed touch readout circuit is compared with the previous works in Table 3. Given the fact that all the previous works in Table 3 use fully differential amplifiers to reject display noise, the proposed single-ended amplifier-based touch readout circuit exhibits good SNR, although not the best. Due to the simple circuit structure of the proposed touch readout circuit, the circuit area and power consumption are much lower than those of the previous works.

Conclusion
A touch readout circuit composed of single-ended amplifiers that effectively suppresses display noise is proposed, where the positive input of each amplifier is connected to the RX line of the adjacent channel instead of a fixed reference voltage and the negative input is connected to the RX line of its own channel. As a result, display noise is injected equally into both the positive and negative inputs of the amplifier and suppressed at the output of the touch readout circuit by common-mode rejection of the single-ended amplifier. Compared with fully differential amplifier-based touch readout circuits,    which are commonly used to reject display noise, the proposed touch readout circuit consumes much lower chip and power consumption due to its simple circuit structure.
The proposed touch readout circuit was fabricated in a 0.35 μm CMOS process and measured to have SNR of 47.2 to 56.2 dB, which is up to 13.1 dB improved compared to the conventional single-ended amplifier-based touch readout circuit. Even compared with the reported fully differential amplifier-based touch readout circuits, the SNR is comparable with much lower circuit area (0.029 mm 2 per RX channel) and power (450 μW per RX channel) consumed.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Notes on contributors
Yongsang Yoo received the B.S. and M.S. degrees in electronic engineering from Hanyang University, Seoul, South Korea, in 2014 and 2016, respectively, where he is currently pursuing the Ph.D. degree. His research interests include sensor readout circuits, memory core, and peripheral circuits.
Byong-Deok Choi (Member, IEEE) received the B.S., M.S., and Ph.D. degrees in electronic engineering from Hanyang University, Seoul, South Korea, in 1994Korea, in , 1996Korea, in , and 2002, respectively. Since 2005, he has been with Hanyang University, where he is currently a Professor with the Department of Electronic Engineering. His current research interests include power management IC, highvoltage power devices and circuits, analog and low-power circuits, driver ICs for displays, and hardware security.