Men42 out11
Ib4
out22
Mep1
Men 1 Ib1
Mep12
Men12 out32 Iout
Ix
Ix
Ix
Ix
s3 segment circuit s4 segment circuit
s2 segment circuit s1 segment circuit
Moutx1
Fig. 4. Multi-segment approximation circuit for tanh.
IV. SIMULATIONRESULTS
To verify the multi-segment approximation circuit, we ini-tiate a simulation by using the 0.35 µm CMOS TSMC tech-nology. The supply voltage VDD of the circuit in Fig. 4 is set to be 2.5V . When we set I0= 30µA, we can observe Ixfrom
−300µA to 300µA for input dynamic range from x = −2.5 to x = 2.5. Figure 5 shows the Ioutsimulation results. The multi-segment approximation method nearly match the ideal tanh function for I from −256µA to 240µA with relative error
The total input dynamic range is about 4.13. The measured frequency response of Iout is shown in Fig. 6. The −3dB bandwidth is 138 MHz. The THD measurements of Iout are shown in Fig. 7 for input DC current of 30µA and 130µA with sinusoidal signals of amplitude Ia from 2µA to 20µA at the frequency of 100 KHz.
V. CONCLUSION
In this paper, we propose a multi-segment method to design
TABLE I
FOUR-SEGMENTS CONTROL TABLE.
Segment s1 on s2 on s3 on s4 on
(Ix< It3) (It3≤ Ix< It2) (It2≤ Ix< It1) (Ix≥ It1)
Vout11 low low low high
Vout12 high high high low
Vout21 low low high high
Vout22 high high low low
Vout31 low high high high
Vout32 high low low low
−300 −200 −100 0 100 200 300
−40
−30
−20
−10 0 10 20 30 40
Ix (µA) Iout (µA)
ideal
multi−segment method
Fig. 5. Simulation results for current-mode tanh function.
10−1 100 101 102 103
−91
−90
−89
−88
−87
−86
−85
−84
MHz dB ( measured at Iout )
Fig. 6. Bandwidth measurement of the multi-segment approximation circuit.
The proposed multi-segment approximation circuit has been verified with wide input dynamic range from −256µA to 240µA for relative error less than 3%, which is wider than the Taylor’s method of expanding series to the term of x7. Moreover, we measure −3dB bandwidth of 138MHz. The THD measurements are less than 2.9% for large sinusoid
2 4 6 8 10 12 14 16 18 20
0 0.5 1 1.5 2 2.5 3
Sinusoidal amplitude I a (µA)
THD (%)
Ix = 30 µA Ix = 130 µA
Fig. 7. THD measurement of the multi-segment approximation circuit.
ACKNOWLEDGMENT
The authors would like to thank the National Chip Im-plementation Center (CIC) of the National Applied Research Laboratories of Taiwan for technical support. The research is supported by the National Science Council of Taiwan, under Grant No. NSC100-2221-E-216-033.
REFERENCES
[1] M. Skrbek, and M. Snorek, ”Shift-add neural architecture,” Proc. 6th IEE Int. Conference on Circuits and Systems, vol. 1, pp. 411-414, 1999.
[2] S. Marra, M. A. Iachino, and F. C. Morabito, ”High speed, programmable implementation of a tanh-like activation function and its derivative for digital neural networks,” Proceedings of International Joint Conference on Neural Networks, pp. 506-511, 2007.
[3] M. Carrasco-Robles and L. Serrano, ”A novel CMOS current mode fully differential tanh(x) implementation,” ISCAS, pp. 21582161, 2008.
[4] Y. Berg, S. Aunet, O. Naess, and M. Hovin, ”A novel low-voltage floating-gate CMOS transconductance amplifier with sinh (tanh) shaped output current,” The 8th IEEE International Conference on Electronics, Circuits and Systems, vol. 3, pp. 1461-1464, 2001.
[5] C. A. De La Cruz-Blas, A. Lopez-Martin, and A. Carlosena, ”1.5-v mos translinear loops with improved dynamic range and their applications to current-mode signal processing,” IEEE Trans. Circuit Syst.-II, vol. 50, pp. 918927, 2003.
[6] W. Liu, S. I. Liu, and S. K. Wei, ” CMOS differential-mode exponen-tial voltage-to-current converter,” Analog Integrated Circuits and Signal Processing, vol. 45, pp. 163168, 2005.
[7] M. Mottaghi kashtiban, A. Khoei, and K. Hadidi, ”A current mode, first-order Takagi-Sugeno-Kang fuzzy logic controller, supporting rational-powered membership functions,” IEICE Trans. Electron, vol. E90-C, pp.
[8] M. T. Abuelmaatti, ”Universal CMOS current-mode analog function synthesizer,” IEEE Trans. Circuit Syst.-I, vol. 49, pp. 14681474, 2002.
[9] M. Kumngern, J. Chanwutitum, and K. Dejhan, ”Simple CMOS current-mode exponential function generator circuit,” Proceedings of ECTI-CON, Krabi, pp. 709712, 2008.
[10] M. Akay, Nonlinear Biomedical Signal Processing, Fuzzy Logic, Neural Networks, and New Algorithms, 2000.
[11] A. Motamed, C. Hwang, and M. Ismail, ”CMOS exponential current-to-voltage converter,” Electron. Lett., vol. 33, pp. 9981000, 1997.
[12] K. J. Lin, C. J. Cheng, S. F. Chiu, and H. C. Su, ”CMOS current-mode implementation of fractional-power functions,” Circuits, Systems, and Signal Processing, vol. 31, pp. 61-75, 2012.
IEICE TRANS. ELECTRON., VOL.Exx–??, NO.xx XXXX 200x
1
PAPER Special Section on Electronic Displays
A CMOS Current-Mode S-Shape Correction Circuit with Shape-Adjustable Control
Kuo-Jen LIN†, Member, Chih-Jen CHENG††, Hsin-Cheng SU†, and Jwu-E CHEN††, Nonmembers
SUMMARY A CMOS current-mode S-shape correction circuit with shape-adjustable control is proposed for suiting different LCD panel’s char-acteristics from different manufactures. The correction shape is divided into three segments for easy curve-fitting using three lower order polyno-mials. Each segment could be realized by a corresponding current-mode circuit. The proposed circuit consists of several control points which are designed for tuning the correction shape. The S-shape correction circuit was fabricated using the 0.35µm TSMC CMOS technology. The measured input dynamic range of the circuit is from 0µA to 220 µA. The -3 dB bandwidth of the circuit is up to 262 MHz in a high input current region.
key words: TFT-LCD, S-shape correction, current-mode circuit, quadratic circuit
1. Introduction
In the past decade, with the rapid increase in mobile cam-eras, mobile phones can now capture and store still images or moving pictures as digital files. As a result, there is a significant difference in the color appearance when captured images are displayed on a mobile LCD. Therefore, real-time color-matching between mobile camera and mobile LCD in a mobile phone needs to be considered to ensure a better im-age quality [1], [2]. The color-correct visualizations of the LCD displays become more and more important.
The electro-optical transfer function of an LCD display is called S-shape curve. Previous works [3]–[6] used sev-eral resistors and operational amplifiers to construct di ffer-ent grey scale’s voltage for correcting the S-shape. In [7], authors proposed an analog compensating S-shape circuit which is composed of BJTs. Comparing to [3]–[6], the tech-nology in [7] is more suitable for mobile device due to the low circuit complexity. In [8], a new model was developed for the S-shaped electro-optical transfer function of the LCD device. However, due to different LCD panel’s spectral and gamma characteristics at different manufacturers [9], one should make different corrections. In [10], a CMOS current-mode S-shape correction circuit with four selectable curves was proposed to meet for four corrections. The authors adopted a polynomial form to represent an S-shape and gen-erated different shapes by adjusting two variables. However, the CMOS current-mode circuit, including a multiplier, a square circuit, and an approximation fractional-power
cir-Manuscript received January 1, 2011.
Manuscript revised January 1, 2011.
†The authors are with the Department of Electronics Engineer-ing, Chung Hua University, Hsinchu, Taiwan
††The authors are with the Department of Electrical Engineer-ing, National Central University, Taoyuan, Taiwan
DOI: 10.1587/transele.E0.C.1
cuit, has weaknesses in power, bandwidth, complexity, and curvature.
In this paper, we also use the CMOS current-mode circuits to design the S-shape correction function, because the current-mode circuit has the advantages of higher band-width, large dynamic range, and simplified circuitry. We di-vide a pseudo S-shape correction curve into three segments.
Each segment has a corresponding circuit with adjustable control points. By adjusting the control points, we can ob-tain a suitable S-shape correction curve.
2. S-shape Correction Curves
A CRT display and an LCD display have electro-optical transfer functions shown in Fig. 1, where the transfer func-tion of an LCD is an S-shaped curve. As same as the gamma correction curve shown in Fig. 1(a), Fig. 1(b) shows an S-shape correction curve. Because different manufactur-ers have different electro-optical characteristics, we can use Fig. 2(a) to show some examples of different electro-optical transfer functions. Similarly, Fig. 2(b) shows some exam-ples of S-shape correction curves.
correction curve s-shape curve
gamma curve correction curve
(a) (b)
Input voltage
Brightness
Brightness
Input voltage
Fig. 1 (a) The electro-optical transfer function of a CRT display and its correction curve. (b) The electro-optical transfer function of an LCD dis-play and its correction curve.
The curves shown in Fig. 2(b) are not mathematical functions, which are hard to describe by some polynomial approximations. We observe that these curves rise quickly on the high voltage part and low voltage part, but slowly on the middle voltage part. Hence, we can certainly divide the S-shape correction curves into three segments. Segment A, B, and C are corresponding to the curve on high, low, and middle voltage part, respectively. We observe that di ffer-ent curves have different curvatures and segment-lengths. The curvature and length will be the important pa-rameters for designing the circuit. Moreover, each segment Copyright c⃝ 200x The Institute of Electronics, Information and Communication Engineers
2
IEICE TRANS. ELECTRON., VOL.Exx–??, NO.xx XXXX 200x
Input voltage
Brightness
Input voltage
Brightness
(a) (b)
B
A C A
A
B B
C
C
Fig. 2 (a) Different electro-optical characteristics for different LCD manufacturers. (b) Different S-shape correction curves for different LCD manufacturers.
could be approximated by using lower order polynomial due to low curvature.
3. Circuit Design and Simulations
We can generate a current-mode quadratic function by using a simple circuit shown in Fig. 3 [11]. From [10], we have
Mi1
Mi2 Iin Im1
Im2 Vc
Fig. 3 Simple current-mode quadratic circuit.
two current-mode functions:
Im1 = K(VDD− |Vt p| − Vtn
2
− Iin
2K(VDD− |Vt p| − Vtn))2 (1)
Im2 = K(VDD− |Vt p| − Vtn
2
+ Iin
2K(VDD− |Vt p| − Vtn))2 (2) Obviously, Im1 and Im2 are the quadratic functions of Iin. As cascading the simple quadratic circuits, we can obtain a higher order polynomial function which should be used to generate different curvatures by controlling the bending fac-tors. In Fig. 4, we create a pseudo S-shape correction curve to illustrate the characteristic of the correction S-shape. In addition, we may take some operations such as bending, ro-tation, condensation, and shifting on the curves in Fig. 4 to realize an S-shape correction. The proposed circuits were simulated by using the 0.35µm TSMC CMOS process and HSPICE. The supply voltage is set to be 3.3 V. The input current is from 0µA to 220 µA.
Iin Iout
Iseg1 Iseg2
Segment A
Segment B
Segment C
0
Fig. 4 Three segmented curve with different curvatures.
3.1 Curve Bending Circuit
Figure 5 shows the curve generator of Segment A which sketched in Fig. 4. When Iin is small, Ima is large from Eq. (1). Because Ima+ Ia = Ir2− Ir1, Ir1will be small for small Iin. As Ima+ Iais large enough, Ma1 is forced to op-erate in cutoff. Therefore, we can use the bias current Iato control the cutoff point. When Iinis large enough, Imawill be small from Eq. (1), Ma1 will be on. If Iinincreases, then Ir1and Isa will increase. Figure 6 shows the simulation re-sults of Isa. As we tune the width of Ma3, we can change the curvature of Segment A. Therefore, Fig. 5 shows the curve bending circuit of Segment A. The size of Ma3 is the curve bending factor.
Mi1
Iin
Vin
Isa
Vout
Mi2
Ma2
Ma1 Ma3
Mi4 Ia
Ima Ir 1
Ir2
Fig. 5 Curve bending circuit of Segment A.
Figure 7 shows the curve generator of Segment B which sketched in Fig. 4. When Iin is small, we find the Imbis large. Because Imb− Ib= If 2− If 1, If 2will be large, the large amount of Imb− Ibwill drive Mb2 operating in sat-uration. When Iin increases, Imb will decrease, If 2and Isb
will decrease. If Iin is large enough, then Mb2 will oper-ate in cutoff. Figure 8 shows the simulation results of −Isb. Therefore, when we tune the width of Mb3, we can change the curvature of Segment B. Consequently, Fig. 7 shows the curve bending circuit of Segment B. The size of Mb3 is the curve bending factor.
3.2 Curve Rotating Circuit
Figure 9 shows the linear circuit. It is a simple
current-LIN et al.: A CMOS CURRENT-MODE S-SHAPE CORRECTION CIRCUIT WITH SHAPE-ADJUSTABLE CONTROL
3
0 50 100 150 200 250
0 20 40 60 80 100 120 140 160 180
Iin (µA) Isa (µA)
Ma3 = 10 µ m Ma3 = 20 µ m Ma3 = 30 µ m
Fig. 6 Different curvatures for different widths of Ma3 at Segment A.
Mi1
Iin
Vin Isb
Vout Mi3
Mi2
Mb1
Mb2 Mb3
Ib
Imb If 1
If2
Fig. 7 Curve bending circuit of Segment B.
0 50 100 150 200 250
−100
−90
−80
−70
−60
−50
−40
−30
−20
−10 0
Iin (µA)
−Isb (µA)
Mb3 = 10 µ m Mb3 = 20 µ m Mb3 = 30 µ m
Fig. 8 Different curvatures for different widths of Mb3 at Segment B.
mirror. When the outputs of curve bending circuits are com-bined with the current-mirror, the curves in Fig. 6 and Fig. 8 are rotated. The linear characteristics and their rotating re-sults are shown in Fig. 10 and Fig. 11, respectively. Fig-ure 12 is the combined circuit of the bending circuit and the rotating circuit, where Iout = Isa− Isb+ Isc+ Idc. The current source Idc is used in DC level bias. Different size of Mc2 represents different rotation angle. The size of Mc2 is the
rotating factor.
Mi1
Iin
Vin Isc
Vout
Mi2
Mc1 Mc2
Mi5
Mi6 Iin
Fig. 9 Curve rotating circuit constructed by simple current-mirrors.
0 50 100 150 200 250
0 10 20 30 40 50 60 70 80 90
Iin (µA) Isc (µA)
Mc2 = 1 µ m Mc2 = 2 µ m Mc2 = 3 µ m
Fig. 10 The linear characteristics of curve rotating circuit.
0 50 100 150 200 250
−40
−20 0 20 40 60 80 100 120 140 160
Iin (µA) Iout (µA)
Mc2 = 1 µ m Mc2 = 2 µ m Mc2 = 3 µ m
Fig. 11 Rotation results including Segment A and Segment B for di ffer-ent widths of Mc3.
3.3 Curve Condensing Circuit
When we tune the width of Mi3 and Mi4 in Fig. 12, we
4
IEICE TRANS. ELECTRON., VOL.Exx–??, NO.xx XXXX 200x
Mi1
Iin
Vin
Iout
Vout
Mi3
Mi2
Mb1
Mb2 Mb3
Mc1 Mc2 Ma2
Ma1 Ma3
Mi5
Mi6 Mi4
Ib
Ia
Isb Isa
Isc
Idc
Fig. 12 A specified S-shape correction circuit.
could condense or expand the Segment C. Figure 13 shows the curve condensing results. The size of Mi3 and Mi4 is the condensing factor.
3.4 Level Shifting Circuit
In the current-mode circuit, we can easily make a level shift circuit by adding a constant current source Idc at the out-put port shown in Fig. 12. Figure 14 shows the simulation results of level shift.
0 50 100 150 200 250
−100
−80
−60
−40
−20 0 20 40 60 80 100
Iin (µA) Iout (µA)
Mi3 = 0.5 µ m, Mi4 = 20 µ m Mi3 = 0.6 µ m, Mi4 = 17 µ m
Fig. 13 Curve condensing results for different widths of Mi3 and Mi4.
3.5 Adjustable S-Shape Correction Circuit
Finally, we design an adjustable S-shape correction circuit
0 50 100 150 200 250
−80
−60
−40
−20 0 20 40 60 80 100 120
Iin (µA) Iout (µA)
Idc = 0 µ A Idc = 10 µ A Idc = 20 µ A
Fig. 14 Level shifting results for different value of Idc.
shown in Fig. 15. We can tune a correction shape by giving a proper VDD or GND at the control points which are named with Cx for x from 1 to n.
4. Experimental Results
We trim the circuit in Fig. 15 into the circuit shown in Fig. 16 for fabricating. In Fig. 16, C1 is used for condensing the Segment C shown in Fig. 4 when C1 connects to VDD. C2 is designed to bend up the curve of Segment A when C2 connects to VDD. C3 and C4 are used to rotate the S-shape counterclockwise when C3 and/or C4 connect to VDD. As C5 connects to GND, the curve of Segment B will bend down. As C6 connects to GND, we can shift down the
out-LIN et al.: A CMOS CURRENT-MODE S-SHAPE CORRECTION CIRCUIT WITH SHAPE-ADJUSTABLE CONTROL
5
Mi1
Iin Vin
Iout
Vout
Mi3 Mib1
Mi2
Mx1 Mx3
Mx2 Mx4
Mb1
Mb2 Mb3 Mb4 Mbn Vn
Ms1
Ms2 Ms3 Ms4 Msn
Mc1 Mc2 Mc3 Mcn Ma2
Ma1 Ma3 Ma4 Man
Mi5
Mi6 Mia1 Mian
Mi4 Mibn
Cn C1
Fig. 15 An adjustable S-shape correction circuit with control points C1, C2, · · · , Cn
put current level. The adjustable S-shape correction circuit was fabricated by using the 0.35µm TSMC CMOS process.
The die photograph is shown in Fig. 17. The supply voltage (V DD) is 3.3 V. Since the available instruments for the mea-surement are in a voltage-mode, we use HSPICE to measure the input voltage (Vin) for every input current in the design stage. Then the input voltage can be applied to the inputs of the fabricated chip. We measure and record the Vout for the chip. Then we can obtain the Ioutby comparing the Vout in the HSPICE. Figure 18 shows the measured results. In the experiment, we set the combination of (C1, C2, C3, C4, C5, C6)= (0, 1, 0, 1, 0, 0) as s1, and the combination of (1, 0, 1, 0, 1, 1) as s2, where 1 and 0 represent V DD and GND, respectively. We observe s1 is more condensed then s2, and s2 is less bended than s1, which prove the control functions of (C1, C2, C3, C4, C5, C6). Therefore, if we add more control points, we can flexibly match different S-shape from different manufactures by using bending, shifting, ro-tating, and condensing. Furthermore, we could overcome the errors due to process variations by adjusting some con-trol points. For conventional methods like [3]–[6], which use variable resistors to adjust the S-shape considering the functions of bending, shifting, and rotating but not includ-ing the condensinclud-ing function. Moreover, the mass resistors in conventional methods increase the chip size. The con-ventional methods use DACs to express the electro-optical transfer functions [3]–[6], which are not related to the input
Fig. 17 Die photograph.
dynamic range. The resolution of DAC represents the curve resolution of S-shape correction. In analog circuit design, however, it is important for S-shape correction with large in-put dynamic range which will increase the tuning resolution of S-shape correction curve. It is observed in Fig. 18, the input dynamic range of Iinis from 0µA to 220 µA, which is large in analog electro-optical transfer function for compar-ing to the voltage-mode range from 1.35 V to 1.75 V in [7].
For observing the frequency responses, we add a small signal 20sin2π f0t (denoted by Isinwith an unit ofµA) to the input of Iin. By changing f0from 1 KHz to 300 MHz, we could measure and record the AC output denoted by If out. As the amplitudes of If outand Isinare denoted by Af outand Asin, respectively, we can calculate each transfer function of 20 log(AAf out
sin). If the maximum value of 20 log(AAf out
sin) is set to
6
IEICE TRANS. ELECTRON., VOL.Exx–??, NO.xx XXXX 200x
Mi1
Iin
Vin
Iout
Vout Mi3
Mi2
Mx1 Mx3
Mx2 Mx4
Mb1
Mb2 Mb3 Mb4
Vn
Ms1
Ms2 Ms3 Ms4
Mc1 Mc2 Mc3 Mc4 C2
Ma2
Ma1 Ma3 Ma4
Mi5
Mi6 Mia1 Mi4
C3 C4
C5
C6 C1
Fig. 16 A trimmed adjustable S-shape correction circuit for fabrication.
0 50 100 150 200 250
−50 0 50 100 150 200 250
Iin (µA) Iout (µA)
s1 s2
Fig. 18 The measured results of the S-shape correction circuit. The shape s1 is the combination of (C1, C2, C3, C4, C5, C6)= (1, 0, 1, 0, 1, 1). The shape s2 is the combination of (C1, C2, C3, C4, C5, C6)= (0, 1, 0, 1, 0, 0).
be 0 dB, then the bandwidth measurements are plotted in Fig. 19, where the locations of -3 dB are at 97 MHz, 190 MHz, and 262 MHz for Iinbeing 25µA, 100 µA, and 175 µA, respectively. We observe that the -3 dB bandwidth is related to the value of Iin. Even though a lower value of Iinwill decrease the bandwidth, the frequency responses in lower current region are still good for -3 dB bandwidth near 100 MHz when comparing to [7] with -3dB bandwidth of 7 MHz and [12] with -3dB bandwidth of 0.8 MHz.
We utilize the measurement result of S-shape correc-tion curve s2 shown in Fig. 18 and MATLAB programs to demo an image correction. We map the output current range from 5µA to 195 µA to the grey levels from 0 to 255. Since each output current responses an input current, we can es-tablish a correction array, which records the mapping grey-levels for the input current. If the input current range from 0µA to 220 µA is divided into 256 segments, then the cor-rection array can be index by 256 grey levels instead of the input current. Consequently, a pixel grey level in the input image can indicate an output grey level through the correc-tion array. Therefore, the image with an S-shape characteris-tic shown in Fig. 20(a) will be corrected to the image shown in Fig. 20(b) by performing the mapping process. The cor-rected image obviously compress the pixel intensity value of the bright image to a darkness part from the observation of the image histogram, which are generated by MATLAB functions.
LIN et al.: A CMOS CURRENT-MODE S-SHAPE CORRECTION CIRCUIT WITH SHAPE-ADJUSTABLE CONTROL
7
103 104 105 106 107 108
−5
−4
−3
−2
−1 0 1
Hz
dB
Iin=25µ A Iin=100µ A Iin=175µ A
Fig. 19 Bandwidth measurement results for Iinbeing 25µA, 100 µA, and 175µA.
0 200 400 600 800
0 0.5 1
0 200 400 600 800
0 0.5 1
(a) (b)
Fig. 20 The measured results of S-shape correction. (a) uncorrected im-age and its histogram. (b) S-shape corrected imim-age and its histogram.
5. Conclusion
In this paper, we have proposed a novel S-shape correction curve generator based on a CMOS current-mode circuit. In particular, due to different S-shape LCD manufactures, one can adjust the shape form by switching the values of control points. The circuit design and simulation results have been presented step by step. We can observe that the shape form generation is very flexible. Furthermore, we have measured the test data and presented its performance from the fabri-cated chip. In the future, we could expect the reliability of generating a shape by setting more control points for com-pensating process variations.
Acknowledgments
The authors would like to thank the National Chip Imple-mentation Center (CIC) of the National Applied Research Laboratories of Taiwan for chip fabrication. The research is supported by the National Science Council of Taiwan, under Grant No. NSC100-2221-E-216-033.
References
[1] Sony Corporation, Colour Matching Technology: BVM-L Series LCD Master Monitor, Sony Corporation, 2009.
[2] A. Moln´ar and K. Samu, The Examinations of the LCD Monitor Pa-rameters in Case of the RGB Basic Colour-Stimuli, XI International PhD Workshop, pp. 190-192, 2009.
[3] P. M. Lee and H. Y. Chen, Adjustable Gamma Correction Circuit for TFT LCD, IEEE International Symposium on Circuits and Systems, vol. 1, pp. 780-783, 2005.
[4] M. Olivieri, R. Mancuso, and F. Riedel, A Reconfigurable, Low Power, Temperature Compensated IC for 8-Segment Gamma Cor-rection Curve in TFT, OLED and PDP Displays, IEEE Transactions on Consumer Electronics, vol. 53, no. 2, pp. 720-724, 2007.
[5] C. W. Park and J. Y. Ryu, Development of a New Automatic Gamma Control System for Mobile LCD Applications, Displays, vol. 29, pp.
393-400, 2008.
[6] D. J. R. Cristaldi, S. Pennisi, and F. Pulvirenti, Liquid Crystal Dis-play Drivers: Techniques and Circuits, Springer, 2009.
[7] Y. Li, C. Wang, X. Lai, and X. Li, Nonlinear Compensating Circuit for LCD Driver, 5th International Conference on ASIC, vol. 1, pp.
631-634, 2003.
[8] Y. Kwak and L. MacDonald, Characterisation of a Desktop LCD Projector, Displays, vol. 21, issue 5, pp. 179-194, 2000.
[9] T. Florin and N. E. Mastorakis, An Overview about Monitors Colors Rendering, WSEAS Transactions on Circuits and systems, vol. 9, issue 1, pp.32-41, 2010.
[10] K. J. Lin, C. J. Cheng, S. F. Chiu, and J. E Chen, CMOS Current-Mode Selectable S-Shape Correction Circuit, Proceedings of the 9th WSEAS Int. Conference on Instrumentation, Measurement, Circuits and Systems, China, pp. 70-75, 2010.
[11] A. Motamed, C. Hwang, and M. Ismail, CMOS Exponential Current-to-Voltage Converter, Electron. Lett., vol. 33, no. 12, pp.
998-1000, 1997.
[12] Texas Instruments, Multi-Channel LCD Gamma Correction Buffer-Check for Samples: BUF07703, BUF06703, BUF05703, SBOS269CMARCH-2003REVISED, April 2010.
8
IEICE TRANS. ELECTRON., VOL.Exx–??, NO.xx XXXX 200x
Kuo-Jen Lin was born in Ilan, Taiwan, in 1963. He received the B.S. degree in electronics engineering from Tamkang University, Taipei, Taiwan, in 1985 and the M.S. degree in electri-cal engineering from Chung Hua University, Hs-ingchu, Taiwan, in 1997. In 2004, he obtained the Ph.D. degree in Computer Science and Infor-mation Engineering from National Central Uni-versity, Taoyuan, Taiwan. From 1997 to 2004, he was an Assistant Researcher at Chung Shan Institute of Science and Technology, Taiwan. He is currently an Associate Professor in the Department of Electronics Engi-neering, Chung Hua University, Hsingchu, Taiwan. His research interests are CMOS current-mode circuit design, synthesis for current-mode circuit, and design in direct digital frequency synthesizers. Dr. Lin is a member of the Institute of Electronics, Information and Communication Engineers.
Chih-Jen Cheng was born in Hsinchu, Tai-wan, in 1984. He received the B.S. and M.S. de-grees from Chung Hua University in 2007 and 2009,respectively. He is currently working to-ward the Ph.D. degree at National Central Uni-versity of Electrical Engineering. His research interests include VLSI circuits design, and ana-log signal processing.
Hsin-Cheng Su was born in Taipei, Tai-wan, in 1986. He received the B.S. degree from Tamkang University in 2008 and M.S. de-gree from Chung Hua University in 2011. His research interests include CMOS current-mode circuit design and image processing.
Jwu-E Chen received the B.S., M.S., and Ph.D. degrees in electronic engineering from National Chiao-Tung University, Hsinchu, Tai-wan, R.O.C., in 1984, 1986, and 1990, respec-tively. Presently, he is an Associate Profes-sor in the Department of Electrical Engineering, National Central University, Chung-li, Taiwan, R.O.C. His research interests include multiple-valued logic, VLSI testing, synthesis for testa-bility, reliable computing, yield analysis, and test management. Dr. Chen is a member of the IEEE Computer Society.