Behavioral Circuit Models of Stereoscopic 3-D Liquid ... - IEEE Xplore

3302

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 10, OCTOBER 2015

Behavioral Circuit Models of Stereoscopic 3-D Liquid Crystal Displays and Shutter Glasses Seung-Hyuck Lee, Jong-Man Kim, and Seung-Woo Lee, Senior Member, IEEE

Abstract— In this paper, behavioral circuit models of a stereoscopic 3-D liquid crystal display (LCD) and shutter glasses (SGs) are proposed. Until now, no study has been conducted on a model that can demonstrate the optical behavior of a 3-D LCD panel, because it includes black frame insertion. In addition, a circuit model of SGs should be established to completely illustrate the behavior of a stereoscopic 3-D LCD system. These circuit models can accurately predict the optical responses of dynamic transitions in 3-D LCD systems and those of SGs. We can then develop a complete behavioral model of the stereoscopic 3-D LCD system by combining the behavioral models of both the 3-D LCD panel and SGs. SmartSPICE simulations are performed to verify our model, which is implemented by an analog hardware description language, Verilog-A. On the basis of these simulation and measurement results, we can conclude that the proposed behavioral circuit models can accurately describe the optical responses of combined displays such as stereoscopic 3-D LCDs. Index Terms— Behavioral model, liquid crystal display (LCD), shutter glasses (SGs), stereoscopic 3-D.

I. I NTRODUCTION OWADAYS, liquid crystal displays (LCDs) have been widely used in various devices. A number of technologies have been developed and implemented in LCDs mainly to achieve high performance and low power consumption [1]–[6]. With the development of LCD technologies, 3-D LCDs have become one of the most common devices [7], [8]. Among several 3-D technologies, a time-multiplexing stereoscopic 3-D display with shutter glasses (SGs) is common in the market [9]. For achieving higher image quality and high performance in 3-D LCD, the electrical and optical characteristics of liquid crystals (LCs) should be accurately analyzed. A stereoscopic 3-D display with an SG contains two types of LCDs: an LCD panel that displays binocular images and SGs that select either left or right images. To describe the optical and electrical behaviors, the macro-models of an LC cell are investigated, which is suitable for circuit simulation [10]. In our previous works, we successfully developed a behavioral circuit model by applying a first-order macromodel to

N

Manuscript received February 16, 2015; revised July 29, 2015; accepted July 31, 2015. Date of publication August 20, 2015; date of current version September 18, 2015. This work was supported by the ICT Research and Development Program through the Ministry of Science, ICT and Future Planning/IITP under Grant 10041416. The review of this paper was arranged by Editor K. C. Choi. The authors are with the Advanced Display Research Center, Department of Information Display, Kyung Hee University, Seoul 130-701, Korea (e-mail: [email protected]; [email protected]; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TED.2015.2464237

Fig. 1. Conceptual diagram of the first-order circuit model. (a) Simplified mechanical model. (b) Schematic of the circuit model.

active-matrix LCD (AMLCD) panels such as twisted nematic, in-plane switching, and patterned vertical alignment (VA) modes [11]–[14]. Moreover, we confirmed that this macromodel could be applied to an AMLCD with an even more complex pixel structure, such as the charge-shared VA (CS-VA) mode [13]. In this paper, we develop first-order macromodels for the active SGs and the CS-VA-mode LCD panel of a stereoscopic 3-D display. In Section II, we explain our behavioral model of the LCD panel and SGs. In Section III, we show the results of our model to 3-D LCD products. II. R EVIEW OF P REVIOUS W ORK Here, we review the macromodel proposed in [10]. Fig. 1(a) and (b) shows a schematic of the equilibrium state and the circuit diagram of the model, respectively. In Fig. 1, E and d represent the applied electric field and the cell gap in the LC layer, respectively. As shown in Fig. 1(a), an LC is affected by three main forces in the equilibrium state: 1) electric force, which aligns the molecules parallel or perpendicular to the electric filed; 2) elastic force that pulls

0018-9383 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

LEE et al.: BEHAVIORAL CIRCUIT MODELS OF STEREOSCOPIC 3-D LIQUID CRYSTAL DISPLAYS AND SHUTTER GLASSES

3303

the LC molecules back to the original position; and 3) viscous force, which is proportional to the velocity of LC molecules in a fluid. Thus, the motion equation can be expressed as d x(t) . (1) dt Here, E, c, k, and γ are an electric field across the pixel, a linear proportional constant, the elastic constant, and viscosity, respectively. x(t) is an average director orientation of LC molecules with time. We can replace the electric field E with the applied voltage Vext as follows: cE 2 = kx(t) + γ

γ d 2 d x(t) Vext kd 2 x(t) + , E= . (2) c c dt d Here, Vext and d are the applied voltages between two electrodes and the cell gap, respectively. Finally, the equilibrium motion equation can be rewritten as 2 Vext =

γ dvi (t) kd 2 , vi (t) = x(t). (3) k dt c Any first-order linear differential equation can be modeled by a simple resistor–capacitor (RC) circuit. Fig. 1(b) shows a schematic of our circuit model. Equation (3) can be behaviorally described by the RC circuit. To improve the accuracy, Watanabe modified the time constant (t) concept on the basis of Smet’s approach. As a result, time constant τ = RC is expressed in (4), where a1 and a2 are physical parameters to represent the LC characteristics 2 = vi (t) + Vext

τ=

1 . 2 a1 + a2 Vext

Fig. 2. Schematic of the CS-VA LCD. (a) Pixel structure. (b) Timing diagram of gate and data signals.

(4)

In addition, we defined the effective voltage, Veff (t), as follows:  Veff (t) = vi (t). (5) Veff (t) helps us determine the LC characteristics, because Veff (t) is directly related to x(t), the orientation of LC molecules that defines the capacitance and transmittance of the LC cell. We implemented this by using the analog hardware description language, Verilog-A. We simulated the CS-VA panel by importing the behavioral circuit model into the circuit simulator SmartSPICE. Fig. 2 shows an equivalent circuit and timing diagram of the CS-VA pixel. As shown in Fig. 2(a), one pixel consists of two driving transistors, T 1 and T 2, for charging subpixels (CLC−A and CLC−B ), a charge-sharing transistor, T 3, and an additional capacitor, CCS , for sharing charges between CLC−B and CCS . The driving principle of the CS-VA mode is as follows: first, the Nth gate signal (G n ) turns on TFT switches T1 and T2 , corresponding to both A and B subpixels, thereby charging the pixel capacitors CLC−A and CLC−B . For example, positive or negative data voltages (VD·P or VD·N ) are applied to both CLC−A and CLC−B . Then, when the next (n + 4)th gate signal (G n+4 ) turns on, subpixel B shares the charge with sharing capacitor (CCS ) by a small amount through an additional TFT (T 3) with a small value of W /L, and hence, the voltage of the B subpixel becomes a different value at a different level. Fig. 3 shows the

Fig. 3. Transient optical responses starting from gray level 0 to other levels. Symbols: measured data. Lines: simulated data.

rising transitions of the CS-VA panel starting from the gray level 0. The symbols and lines represent the measurement and simulation results, respectively. In the simulation, the timeconstant parameters a1 and a2 were −143 s−1 and 27 s−1 V−2 , respectively. In the previous work [13], the frame rate of the CS-VA panel was 120 Hz, whereas the frame rate in this work is 240 Hz. As shown in Fig. 3, the simulation results are well matched with the measurement results. III. B EHAVIORAL M ODELS OF 3-D LCD AND S HUTTER G LASS A. Driving Scheme of 3-D CS-VA Panel To predict the optical characteristics of the CS-VA panel driven in a 3-D environment, it is necessary to understand the

3304

Fig. 4.

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 10, OCTOBER 2015

Timing diagram for the driving scheme of the 3-D CS-VA panel.

Fig. 6. Measured capacitance of an LC cell used for the SG, depending on the frequency.

Fig. 5.

Optical characteristics of the CS-VA panel with and without BFI.

driving scheme. A commercial 46-inch FHD (1920 × 1080) 3-D CS-VA panel is used in our experiments. Fig. 4 shows a basic driving scheme for the 3-D CS-VA panel that uses an active SG. This driving scheme adopts black frame insertion (BFI). The black data are inserted to separate the left eye images from the right ones. One frame (1/60 s) in the conventional 2-D driving scheme corresponds to four fields in 3-D driving, and each field is 1/240 s long. All the data are refreshed every 1/240 s, which is equal to the period of vertical sync (VS) signal of 3-D CS-VA panel. VS signal is the timing signal that indicates the beginning of the field. To realize the 3-D visual perception, 3-D CS-VA panel displays the left and right images with inserted black field in sequence. The left image is displayed for 1/240 s, and then, the full-screen black image is inserted during the next field. Subsequently, the right image is displayed for 1/240 s, and then, the black image is displayed again during the last field. Thus, the data sequence during one frame time (1/60 s) is left–black–right–black (L–B–R–B). The active SG transmits the synchronized left and right images from displayed images on the LCD panel. Meanwhile, back light unit (BLU) is controlled as synchronized with panel data because the data are progressively applied to the panel, line by line. B. Behavioral Model of CS-VA LCD Panel Fig. 5 shows the behavioral characteristics of the CS-VA panel. The red solid line and blue circles represent the

transmittance characteristics of static image with and without BFI driving, respectively, when the panel displays the full 255 Gy. From Fig. 5, we find that two curves are quite different. When 2-D static image is displayed, there is no fluctuation of transmittance. However, when 3-D static image is displayed, it is shown that transmittance fluctuates severely. It is caused because the BFI is used to make 3-D vision. The most remarkable aspect is that transmittance with BFI cannot reach the target level of 255 and 0 Gy level. It is because LC molecules should always return to the black level in a short time (1/240 s). Because of the slow LC response, transmittance cannot reach the target in that short time. Meanwhile, in the VA panel, a sudden application of high voltage can cause randomly tilted domains of LC molecules. Before the propagation of the tilting waves by the fringe field from the edges of the electrode, LC molecules in the center area start tilting randomly, owing to a vertically applied electric field. Thus, it takes a long time for the randomly tilted LC molecules to return to the right direction. However, with BFI, LC molecules cannot reach the target when black image is displayed. As a result, LC molecules with vertical state do not exist, and delay does not appear in the case of 3-D vision with BFI. C. Behavioral Model of SG Fig. 6 shows the frequency dependence of the dummy LC cell used for the SG. As shown in Fig. 6, the capacitance changes when the frequency changes. However, the SG is driven by a signal at a fixed frequency (60 Hz), as shown in Fig. 4. Thus, we used the capacitance–voltage (C–V ) characteristics of the dummy cell operating at 60 Hz. To establish the circuit model, the C–V characteristics should be expressed by a formula, because the capacitance values should be obtained corresponding to the effective voltage values. Fig. 7 shows the C–V characteristics of the panel. The solid line represents the estimated C–V characteristics. We used the hyperbolic tangent (tanh)

LEE et al.: BEHAVIORAL CIRCUIT MODELS OF STEREOSCOPIC 3-D LIQUID CRYSTAL DISPLAYS AND SHUTTER GLASSES

3305

Fig. 7. Capacitance–voltage characteristics. Symbols: measured data. Line: estimated data. Fig. 9. Comparison of simulation results with measurement data of the transient responses of the 3-D CS-VA panel when BFI is applied. Symbols: measured data. Dotted line: estimated data.

Fig. 8. TV characteristics of SGs. Symbols: measured data. Line: estimated data.

IV. R ESULTS AND D ISCUSSION

function, as shown in (6), to fit the measured data 2 C(Veff ) = Cmin + (Cmax − Cmin ) · tanh(P) π 1

Veff − Vtc α + (α 2 + δ 2 ) 2 , α= P = . (6) 2 Vmc Here, Cmin , Cmax , Vmc , Vtc , and δ are the minimum capacitance, the maximum capacitance, the capacitance modulation voltage, the threshold voltage, and the smoothing constant, respectively. Fig. 8 shows the transmittance–voltage (T –V ) characteristics of the SG. The line represents the estimated TV characteristics. We used the hyperbolic tangent (tanh) function, as shown in (7), to fit the measured data 2 T (Veff ) = Tmin + (1 − Tmin ) · tanh(Q) π 1

Veff − Vth β + (β 2 + η2 ) 2 , β= Q= . 2 Vmo

Tmin , Vmo , Vto , and η are the minimum transmittance, the optical modulation voltage, the threshold voltage, and the smoothing constant, respectively. We applied different parameters to high-, middle-, and low-transmittance regions to improve the accuracy of lower gray levels. As shown in Fig. 8, the estimated T –V curve is in good agreement with the measured data over the entire Veff range. Note that the maximum transmittance of the SG is approximately 35%. We adopted these empirical formulas, (6) and (7), because C–V and T –V characteristics are extracted exactly as referred in [10]. However, we approximated the C–V and T –V characteristics with arc tangent or hyperbolic tangent optionally depending on the opto-electrical property of panel. If the C–V or T –V characteristics are saturated in extra-high voltage, we preferred hyperbolic tangent formula.

(7)

Fig. 9 shows the simulation results of the CS-VA panel with BFI. The target gray level was 255. The symbols and lines represent the measurement and simulation results, respectively. As shown in Fig. 9, the optical responses were accurately predicted. Note that the transmittance of a certain gray level cannot reach its target value when the panel operates without BFI. For example, 255–0–255–0 is a repeating data pattern when the gray level of 255 is displayed. Thus, the normalized maximum and minimum transmittance should be approximately “1” and “0,” respectively. However, the gray levels of 255 and 0 at each field do not reach the target transmittance. As we mentioned in the previous section, this is caused by the slow response characteristics of the LC molecules. Fig. 10 shows the simulation result of the SG. The symbols and lines represent the measurement and simulation results, respectively. As shown in Fig. 10, the simulation results are in good agreement with the measurement results. The frame

3306

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 62, NO. 10, OCTOBER 2015

while the left eye perceives the left image, right eye perceives the black image in the field from 0 to 2. Contrarily, from field 2 to 4, the left eye perceives the black image while the right eye perceives the corresponding right image. In this way, left and right image can be separate. With our model, we can precisely predict not only the optical response but also the degree of crosstalk. In this respect, we can evaluate the performance of the 3-D LCD before production. Crosstalk is the critical artifact of 3-D display, which hinders natural 3-D vision and causes visual fatigue. We expect that 3-D performance can be optimized through prediction of the optical and electrical characteristics of 3-D LCDs in advance. V. C ONCLUSION

Fig. 10. Simulation result of transient response of the SG. Symbols: measured data. Dotted line: estimated data.

In this paper, we combined the behavioral models of stereoscopic 3-D LCDs and SGs. The circuit model of a stereoscopic 3-D LCD that adopts BFI is established. Using this model, we can accurately predict the optical responses of a 3-D LCD panel. In addition, the circuit model of SGs is also established. The optical response of a stereoscopic 3-D LCD with SGs can be predicted successfully. Finally, we verified our model with a SPICE simulation. In the simulation results, we could correctly predict the crosstalk in stereoscopic 3-D LCDs. R EFERENCES

Fig. 11. Comparison of simulation results with measurement data of transient responses, including both the 3-D CS-VA panel and SG. Symbols: measured data. Dotted line: estimated data.

frequency of the SG is 60 Hz. During the first half frame, the left or right image is transmitted to the user’s corresponding eye. During the second half frame, the SG is still open, but the image from the CS-VA panel is black. After the SG transmits the image to the user’s corresponding eye, it is turned OFF to isolate the image of the corresponding eye from the other eye. Fig. 11 shows the simulation results by considering the CS-VA panel and SG. The symbols and lines represent the measured and simulation results, respectively. In measured data, transmittance has to be near zero, but it is not. It is called crosstalk of the 3-D LCDs and is expressed at the fields 2 and 6. At field 2 and 6, Crosstalk is the severest at 255 Gy, and it is decreased at lower gray. As shown in Fig. 11, we could accurately predict the complete optical response that left or right eye can perceive. For instance,

[1] J.-M. Kim, J. Kim, Y. Cho, M. Kim, and S.-W. Lee, “Liquid crystal displays with temperature-independent characteristics,” Opt. Eng., vol. 51, no. 7, pp. 077402-1–077402-5, Jul. 2012. [2] D. Cho, W.-S. Oh, and G.-W. Moon, “A novel adaptive dimming LED backlight system with current compensated X-Y channel drivers for LCD TVs,” J. Display Technol., vol. 7, no. 1, pp. 29–35, 2011. [3] X.-B. Zhang, R. Wang, D. Dong, J.-H. Han, and H.-X. Wu, “Dynamic backlight adaptation based on the details of image for liquid crystal displays,” J. Display Technol., vol. 8, no. 2, pp. 108–111, Feb. 2012. [4] K. Käläntär, “A monolithic segmented functional light guide for 2-D dimming LCD backlight,” J. Soc. Inf. Display, vol. 19, no. 1, pp. 37–47, Jan. 2011. [5] J. Lee, K. Oh, H. Kim, and S.-T. Wu, “Novel surface-stabilized vertical alignment mode for fast-response liquid crystal display,” J. Display Technol., vol. 8, no. 5, pp. 296–298, May 2012. [6] H.-C. Cheng, J. Yan, T. Ishinabe, C.-H. Lin, K.-H. Liu, and S.-T. Wu, “Wide-view vertical field switching blue-phase LCD,” J. Display Technol., vol. 8, no. 11, pp. 627–633, 2012. [7] P.-J. Lee and Y.-C. Lai, “Perceptual awareness rate control for multi-view video encoder in stereoscopic display,” J. Display Tech., vol. 9, no. 7, pp. 552–560, Jul. 2013. [8] A. Cellatoglu and K. Balasubramanian, “Autostereoscopic imaging techniques for 3D TV: Proposals for improvements,” J. Display Technol., vol. 9, no. 8, pp. 666–672, Aug. 2013. [9] Z. Xia, X. Li, Y. Cui, L. Chen, and K. Teunissen, “A study on the perceptual relevance of crosstalk equations for stereoscopic displays,” J. Soc. Inf. Display, vol. 21, no. 6, pp. 271–279, Jun. 2013. [10] H. De Smet, J. Van Den Steen, and D. Cuypers, “Electrical model of a liquid crystal pixel with dynamic, voltage history-dependent capacitance value,” Liquid Crystal, vol. 31, no. 5, pp. 705–711, 2004. [11] J.-M. Kim, Y. Cho, J. Kim, S.-H. Lee, K. Kim, and S.-W. Lee, “Precise prediction of optical responses of all types of liquid crystal displays by behavioral models,” J. Soc. Inf. Display, vol. 21, no. 1, pp. 2–8, Mar. 2013. [12] Y. Cho, C. Park. J.-M. Kim, S.-H. Lee, and S.-W. Lee, “A behavioral circuit model of active-matrix liquid crystal displays for optical response simulation,” IEEE Trans. Electron Devices, vol. 59, no. 5, pp. 1430–1438, May 2012. [13] J.-M. Kim, Y. Cho, S.-H. Lee, J. Kim, and S.-W. Lee, “Behavioral circuit model of active-matrix liquid crystal display with charge-shared pixel structure,” IEEE Trans. Electron Devices, vol. 60, no. 5, pp. 1673–1680, May 2013.

LEE et al.: BEHAVIORAL CIRCUIT MODELS OF STEREOSCOPIC 3-D LIQUID CRYSTAL DISPLAYS AND SHUTTER GLASSES

[14] J.-M. Kim, S.-H. Lee, and S.-W. Lee, “Behavioral model of patterned vertical alignment pixel in active-matrix liquid crystal displays,” IEEE Trans. Electron Devices, vol. 61, no. 11, pp. 3783–3789, Nov. 2014.

Seung-Hyuck Lee received the B.S. and M.S. degrees from the Department of Information Display, Kyung Hee University, Seoul, Korea, in 2011 and 2013, respectively, where he is currently pursuing the Ph.D. degree with the Department of Information Display. His current research interests include driving methods and circuits for LCDs and driving technology for color motion performance of LCDs.

3307

Jong-Man Kim received the B.S. and M.S. degrees in physics from the Department of Information Display, Kyung Hee University, Seoul, Korea, in 2010 and 2012, respectively, where he is currently pursuing the Ph.D. degree with the Department of Information Display. His current research interests include driving methods and circuits for LCD and E-paper displays and driving technology for color motion performance of LCDs.

Seung-Woo Lee (SM’10) received the B.S. and M.S. degrees in electrical engineering and the Ph.D. degree from the Korea Advanced Institute of Science and Technology, Daejeon, Korea, in 1993 and 1995, respectively. He joined Samsung Electronics, Company, Ltd., Suwon, Korea, in 2000. He is currently an Associate Professor with the Department of Information Display, Kyung Hee University, Seoul, Korea. Prof. Lee has been active with SID as a Senior Member.