3.

Experimental Results

For the experiment, a 17-in. LCD panel with a $1280\times 1024$ resolution was used. Total resistance and capacitance of one column line were $30\mathrm{k}\mathrm{\Omega }$ and $110\mathrm{pF}$ , respectively. To find out optimal pulse widths according to the locations of pixels from the CD, five vertical locations with an interval of 256 pixels were selected. I recorded the extended pulse widths when the horizontal lines disappeared at those five locations. In Fig. 4, a solid line with solid squares shows the minimum extended times to compensate the failure at the five locations. Figure 4 shows that there was no horizontal-line defect at the nearest location to the CD, and at least $2\mathrm{\mu }\mathrm{s}$ were necessary to compensate the insufficient charging at the fifth position from the CD. Figure 4 shows that the modulated time became saturated, as the distance from the CD is longer. I extended pulse widths of even row signals further after the horizontal line defects were gone. New line defects were visible after some amount of extra-extension. The dotted line with open squares in Fig. 4 shows overcompensated conditions at the five locations. The new line defects are caused by a charging delay through switch transistors. The OTM gets rid of the voltage difference at column lines. The OTM, however, reduces turn-on times of the odd pixels, which causes insufficient time to transfer the compensated column voltages to the pixels. Thus, the new line defects become visible because the odd lines are dimmer after overcompensation. I found that it is impossible to fix the pulse widths of even and odd lines, regardless of position, because of the overcompensation problem. Hence, it is necessary to extend pulse widths of even lines according to their positions from the CD. The extended times of even lines should be shorter, as their positions are closer to the CD. The extended time $t_{\mathrm{ex}}$ can be calculated using

where $t_{\max }$ is a maximum extended time, $k$ is the order of the equation, and $R_n$ and $R_N$ are the distances of the $n$ ’th and the last rows from the CD, respectively. I found that if the order of the equation is higher than three, the horizontal defects are gone throughout the LCD screen. Figure 5 shows a comparison between the OTM with a fourth order and eye threshold, over which the horizontal defects are gone. I measured luminance of several selected gray levels at the upper, center, and lower parts of the 17-in. LCD screen. The overall luminance improvements after the OTM are less than 0.7% on average over the gray levels at the upper part. This means luminance increase by the OTM was not effective, because the upper part has a relatively sufficient charging time. At the lower part, however, luminance of most gray levels increases after applying the OTM by 2.7% on average, with 6.1% as a maximum. Thus, the horizontal lines were gone. We can easily calculate the actual luminance increase of even pixels. Before applying the OTM, the measured luminance is the same as the average luminance value of even and odd pixels, i.e., $(L_{\mathrm{EVN}}+L_{\mathrm{ODD}})∕2$ , where $L_{\mathrm{EVN}}$ and $L_{\mathrm{ODD}}$ are luminances of pixels at even and odd rows, respectively. The luminance after applying the OTM is the same as $L_{\mathrm{ODD}}$ , because even pixels are compensated by the OTM. I found that the OTM could increase the luminance of even pixels by a maximum of 13.1% for gray level 64 and by 5.7% on average over the gray levels.