Design of a CMOS Led Print Head Driver With Compensation Circuits

pixel compensation. Besides, a method of a single-chip segment exposure is also proposed to make the power consumption more efficient. Based upon the device parameters of 0.5 m 1P2M CMOS technology with 3.3 V power supply, all the functions and performance of the proposed CMOS LED print head driver with compensation circuits are successfully tested and proven through measurements. The area including ESD I/O pads is 2000 2000 m2. The proposed chip is suitable for LED printers.

Index Terms—Compensation circuits, LED printer, LED print

head driver.

I. INTRODUCTION

N

OWADAYS, electrophotographic printers play a main role in the office and family. In the commercial printer market, both laser and ink printers are the most common printers. Although a laser printer has better printing quality than an ink printer, the cost of a laser printer is higher than an ink printer [1], [2]. To reduce the high cost and simplify the system complication, a technique of light emitting diode (LED) [3] applied on the printers is proposed. Although LED printers are still under researched, some achievements [4]–[10] are openly discussed. However, some achievements [4]–[7] mainly focus on bonding techniques of LED printers. For example, the bonding design of the chip-matrix LED print head and characteristics of the chip matrix LED are introduced in [4]. Three-dimensional epitaxial thin-film bonding is discussed in [7]. In this work, different to previous issues [4]–[7], circuit dis-cussions of a CMOS LED print head driver with compensation circuits are firstly described. We not only propose a method of a single-chip segment exposure to improve power consumption of an LED array, but also compensation circuits to perform the chip and pixel compensation. Readers can understand the whole design techniques from this work, and this is the main contribution of this work.

Manuscript received July 04, 2011; revised August 14, 2011; accepted August 18, 2011. Date of publication August 30, 2011; date of current version April 06, 2012. This work was supported by the National Science Council, Taiwan, under Contracts NSC-99-218-E-415-002. The associate editor coordinating the review of this manuscript and approving it for publication was Dr. M. Abedin.

C.-T. Chiang is with the Department of Electrical Engineering, National Chia-Yi University, Chiayi 600, Taiwan (e-mail: ctchiang23.ee90g@nctu. edu.tw)

C.-T. Kuo and Y.-C. Huang are with the National Chiao Tung University, Hsinchu 300, Taiwan.

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/JSEN.2011.2166115

Fig. 1. The printer procedures of electrophotographic printers.

In this paper, a CMOS LED print head driver with compen-sation circuits is newly proposed. Compared with traditional pulsewidth modulation (PWM) [9], [10], this work improves the traditional PWM into dynamic pulsewidth modulation (DPWM). The improved circuit is easily implemented, and can make pulsewidth more adjustable for applying on the LED printer head drivers. With the design of DPWM, a proposed method of a single-chip segment exposure can make the power consumption more efficient. Another issue is the compensation techniques. Due to the process and fabrication variations, the characteristics of each driver IC or an LED pixel are different. To solve this problem, two compensation circuits are designed in the proposed chip. One is to perform the chip compensation, and the other is to do the pixel compensation. In Section III, these two compensations will be discussed in more detail. All the functions and performance of the proposed CMOS LED print head driver with compensation circuits are successfully tested through measurements. The chip area including ESD I/O pads is 2000 2000 m . The proposed chip is suitable for LED printers.

In the Section II, the operations of electrophotography printers are addressed. Section III discusses system architecture and simulation results. Section IV demonstrates the measure-ments. Finally, conclusions and future works are described.

II. THEOPERATIONS OFELECTROPHOTOGRAPHYPRINTERS The printing procedures of electrophotography printers need seven steps as shown in Fig. 1. These steps are charge, expo-sure, development, transfer, fuse, clean, and erase. Among these 1530-437X/$26.00 © 2011 IEEE

CHIANG et al.: DESIGN OF A CMOS LED PRINT HEAD DRIVER 871

Fig. 2. Structure of the LED printer [8].

TABLE I

COMPARISONS OFLASERPRINTERHEAD ANDLED PRINTHEAD

steps, the exposure is the most important step. It can largely decide the printing quality. In electrophotography printers, the lighting source can be generated by a laser diode or an LED array.

A laser printer includes a laser diode, complicated lens ma-chinery, and a set of rotating multi-mirrors. This complicated system uses the principle of light refraction to perform the scan-ning operation and then to expose a column of printed data. The benefit is that laser does not have the scattering problem as gen-eral white light has. However, the disadvantage is that the cost of the whole system is higher, and the complicated systems de-signed to control the rotating multi-mirrors make the volume of printers larger. Sometimes, the stability of mechanical structure of laser printers is not good. The printing quality is thus affected. To solve the problems of larger volume, worse stability, and high cost, LED printers [8] are proposed as shown in Fig. 2. The complicated systems, such as lens machinery and rotating multi-mirrors, can be totally removed. The stability is raised and the volume is reduced. Besides, the printing speed can be faster than a laser printer due to its simple machinery of LED printers. However, the resolution is still limited by LED technology and its package. The problem of resolution can be improved by fol-lowing the industrial progress. The comparisons of these two kinds of printers are summarized in Table I.

LED print head consists of LED pixels and drivers. Both of them are connected by wire bonding [4]–[6]. The lighting performance is both controlled by LED and its driver. The characteristics of LED pixels are measured as shown in Fig. 3. Four LED pixels are measured. The relationship between

Fig. 3. Measured results of power versus inputting current of LED and error between 4 LEDs that were fabricated in the same material.

Fig. 4. Paper format used in the specification.

power and inputting current of LED is linear. The error be-tween LED pixels is derived as

% (1)

where and are the maximum and minimum power

of LED pixels under the same inputting current. Below the in-putting current of 2 mA, the error is not uniform within 6%. It is not uniform until the inputting current is larger than 2.5 mA. This result is used to specify the driving current of each LED pixel. On the other hand, LED print head drivers can largely decide the main performance of LED printers. In Section III, the proposed CMOS LED print head driver with compensation circuits will be discussed. In the specification of this work, the paper format is an A4 type as shown in Fig. 4. The resolution is 1200 dots per inch (DPI). The printing speed is 20 pages per minute (PPM). The driving current of each LED pixel is 2.5 mA.

Fig. 5. Circuit diagram of the proposed LED printer driver with compensation circuits.

III. SYSTEMARCHITECTURE ANDSIMULATIONRESULTS The proposed CMOS LED print head driver with compensa-tion circuits is shown in Fig. 5. The proposed circuits include control circuits that control three state transitions, shift registers and latches, compensation circuits, and output LED drivers.

A. Control Circuits of State Transitions

Fig. 6 shows the state transition graph. Three states are

CHIP DATA state which is to reset the chip and store the

com-pensation data, PIXEL DATA state which is to transfer and store the printed data, and EXPOSURE state. In the CHIP DATA state, all counters and registers are reset to zero. Some com-pensation data are stored into the comcom-pensation circuits. In the PIXEL DATA state, control circuits transfer printed data into shift registers and latches. Then, it waits for a permission signal which is defined as next to go into the next state. In the

EXPOSURE state, LED drivers will trigger an LED array to

perform the exposure procedure. Finally, the words or pictures are printed on the papers.

The timing diagram of control circuits is shown in Fig. 7. Control circuits generate signals to control state transition. Firstly, the reset and next signals are both generated by the

CLKI and STB signals. The STB signal stands for a next state

Fig. 6. State transition graph.

transition. Both of the CLKI and STB signals are generated by a field-programmable gate array (FPGA) or a microprocessor. By performing logic operations on the reset and next signals, controlling signals of three state transitions are thus obtained. When the reset signal is enabled, the proposed chip is in the

CHIANG et al.: DESIGN OF A CMOS LED PRINT HEAD DRIVER 873

Fig. 7. The timing diagram of control circuits.

logic high until the rest signal is asserted into logic low. At the same time, the first next signal is enabled. The present state of the proposed chip is changed from CHIP DATA state into PIXEL DATA state. Similarly, the pixel signal will be maintained in the logic high until the procedure of data transfer is finished. Then, the second next signal is enabled. The present state is changed from PIXEL DATA state into a short latch state. The latch state is to store the printed data in the latch. After latching these data, the EXPOSURE state begins its procedure. The exposure signal will be asserted into logic high until the exposure procedure is finished. Finally, all the procedures will be continuously performed until the words or pictures are totally printed.

B. Dynamic Pulse Width Modulation

Circuits and timing diagram demonstrated in Fig. 8 is to dis-play the operation of traditional pulsewidth modulation. It com-pares the n-bit data with n-bit counter. For example, if n is 4 and

DATA is 1100, the counter counts the number from 0 to 12. The out signal is always maintained in the logic high until it counts

to the number of 13. That means an LED pixel will light on the duration of the number 0 to 12. That implies that the power is al-ways consumed during this counting time. The power efficiency is not good. In order to make pulsewidth more adjustable, the pulsewidth modulation shown in Fig. 8 is improved into DPWM as demonstrated in Fig. 9. By performing the logic operations, the stb in signal is generated from the STB and CLKI signals. Now, the trigger signal of counter is changed into the stb in signal. The stb in signal is triggered high at the rising edge of the CLKI signal when the STB signal is high. Then the counter counts at the rising edge of the stb in signal. It counts upward one and holds until the next pulse of the stb in signal appears. For example, the time duration of the number 1 is two times than others numbers in Fig. 9. By changing the stb in signal,

Fig. 8. Circuit and timing diagram of traditional pulsewidth modulation.

Fig. 9. Timing diagram of dynamic pulsewidth modulation.

the pulsewidth can be more variable. Although the improved cir-cuit is simple, the adjustable time duration can make power con-sumption more efficient. The proposed method of a single-chip segment exposure will also use the technique of DPWM to im-prove the power consumption.

C. Chip Compensation and Pixel Compensation

After the chip fabrication is finished, the characteristic of each driver IC could not be absolutely unity. The pulse start-limited circuit is thus proposed to perform the chip compensation as demonstrated in Fig. 10. The pulse start-limited circuit uses an

Fig. 10. Circuit diagram of the chip compensation.

Fig. 11. Timing diagram of the chip compensation.

8-bit counter to compare with the compensation data. When the counting value is larger than the compensation data, the starting time of LED light is decided. For example, in Fig. 11, the com-pensation data is stored in the number of 4. After the operations of pulse start-limited circuit are finished, the pulsewidth of the signal and out is also generated. This signal will be used to con-trol the lighting time of an LED array.

By following the LED fabrication process, each LED has its individual difference. Thus, in the proposed pixel compensation, an output-current adjuster implemented by a 4-bit current-mode digital to analog converter is adopted. It can adjust the driving current of each LED according to the measured characteristic.

The compensation data to shown in Fig. 12 are stored in the CHIP DATA state, and used in the EXPOSURE state. The driving current is expressed as

(2) (3)

where , and are the current of MOS

, and , respectively. is the controllable bias voltage and is timing signal to turn on or off MOS and

CHIANG et al.: DESIGN OF A CMOS LED PRINT HEAD DRIVER 875

TABLE II

MEASUREDRESULTS OFDELAYTIME OFCONTROLLINGSIGNALS OFSTBANDCLK

TABLE III

MEASUREDRESULTS OFINTERVALTIME OFOUTPUTSIGNALS OFDATAOANDCLKO

Fig. 12. Circuit diagram of the pixel compensation.

Fig. 13. SPICE simulations of the pixel compensation.

. The SPICE simulations are demonstrated in the Fig. 13. The current of each LED pixel can thus be compensated by adaptively adjusting the codes of to .

D. A Single-Chip Segment Exposure

To solve the problem of large power consumption, a method of a single-chip segment exposure shown in Fig. 14 is also pro-posed. In the EXPOSURE state, if the current of each LED pixel is 2.5 mA and the effective resistance is 1 k , the resulting power consumption is equal to 6.25 mW. If there are 4962 LEDs within an LED array and are lighting together, the total power consumption will be 32 W. Such large power consumption will be a serious problem. The method of a single-chip segment ex-posure is to separate an LED array into two groups. Diodes within an array are numbered as odd and even diodes. The ex-posure time can be divided into two parts as shown in Fig. 15. The L1 and L2 LED diodes use a common output node together, which is O1. When the signal SEL0 is in the logic high, odd LED diodes such as L1 are turned on and even LED diodes such as L2 are turned off. Similarly, if the signal SEL1 is in the logic high, even LED diodes are turned on and odd LED diodes are turned off. The power consumption is thus reduced by 50%. This pro-posed method also can be made by dividing an LED array into more groups. However, the controlling signals are more com-plicated. The printing speed is affected due to the segment ex-posure. However, the printing speed is not a problem due to the fact that it can be easily compensated and adjusted by a faster clock frequency. In the next section, all the functions and perfor-mance of the proposed chip are proven through measurements.

IV. MEASUREMENTRESULTS

Fig. 16 demonstrates physical layout of the proposed CMOS LED print head driver with compensation circuits is

Fig. 14. Circuit diagram of the single-chip segment exposure.

Fig. 15. Timing diagram of the single-chip segment exposure.

2000 2000 m including ESD I/O pads. The CMOS 0.5 m 1P2M process with 3.3 V power supply is chosen due to its lower cost to fabricate this chip. The driving current of each LED pixel is 2.5 mA. The clock frequency fc is 20 MHz. Fig. 17(a) and (b) show the testing board and measurement setup, respectively. The data signals DATA [3:0] and control-ling signals of CLKIP, CLKIN and STB are generated by FPGA. The FPGA is programmed by Verilog codes. In order to avoid error sampling, the signals of DATA [3:0], CLKIP, CLKIN, and STB are passing through stage by stage. These signals can be correctly captured and processed within each stage.

Firstly, the transmitting function is tested. This test is sepa-rated into two parts. One is to test the chip in the CHIP DATA state. Fig. 18 shows the transmitting condition of the compen-sation data stored in the compencompen-sation circuits. The other test is

Fig. 16. Physical layout of the proposed CMOS LED print head driver with compensation circuits is 20002 2000 m including ESD I/O pads.

to verify the chip in the PIXEL DATA state. The measurement results are demonstrated in Fig. 19. As shown, the functions are correctly proven by the DOUT signals. The data can be captured in the rising and falling edges of the CLKIP signal. The DOUT signals are correctly delayed 2 and 4 clock cycles, respectively. Although the transmitting function is tested successfully, the delay and interval time should also be verified. The delay time is to check if it interferences the sampling functions of the chip or

CHIANG et al.: DESIGN OF A CMOS LED PRINT HEAD DRIVER 877

Fig. 17. (a) Testing board. (b) Measurement setup.

Fig. 18. Measurement results of state transition in the CHIP DATA state.

not. The interval time is to judge if the following chip can suc-cessfully receive the data from the preceding chip or not. Table II demonstrates the measured results of delay time of controlling signals of STB and CLK. Three chips are measured and the delay time is almost the same. Table III displays the measured results of interval time of output signals of DATAO and CLKO. By mea-suring the interval time of three chips, the time is all less than 1 ns. Measurements can prove that these signals can correctly be captured and processed within each stage.

Fig. 19. Measurement results of state transition in the PIXEL DATA state.

The measured currents in the pixel compensation are plotted in Fig. 20. Three chips are tested again. The difference of each step is about 0.03 to 0.04 mA. This implied that good linearity of the compensated current is obtained. Finally, Fig. 21 demonstrates the measured total current through LED pixels based on two methods. One is based on the traditional PWM method in Fig. 21(a). The other is followed by the proposed method of a single-chip segment exposure with DPWM in Fig. 21(b). Firstly, sixteen LED pixels are grouped into a set.

Fig. 20. The measuredI currents in the pixel compensation.

Fig. 21. (a) Timing diagram of the traditional PWM method. (b) Timing dia-gram of the proposed method of a single-chip segment exposure with DPWM. (c) Measured total current based on two methods.

The timing of the traditional PWM method is inputted to the proposed chip, and then the total current is measured. The maximum current is about 40 mA. The total current is different on the time variation. On the other hand, the total current based on the proposed method of a single-chip segment exposure with DPWM is measured. The total current is half of the maximum current of the traditional PWM method. Different

Fig. 22. (a) Timing diagram of the proposed method of a single-chip segment exposure with DPWM. (b) Measured total current.

TABLE IV

SUMMARY ON THECHARACTERISTICS OF THEPROPOSEDCMOS LED PRINTHEADDRIVERWITHCOMPENSATIONCIRCUITS

to the traditional PWM method, the current is uniformly av-eraged. The power consumption is saved. In Fig. 22, another measurement is to prove that better power efficiency can also be obtained under different condition of the proposed DPWM. The pulsewidth is smaller and within counting number 7. In Fig. 22(b), although the total current is different on the time variation, the maximum total current is 20 mA and also half of the maximum current of the traditional PWM method. That implies the proposed method of a single-chip segment exposure with DPWM has efficient power consumption compared with traditional PWM method. The characteristics of the proposed CMOS LED print head driver with compensation circuits are summarized in Table IV.

V. CONCLUSION

A CMOS LED print head driver with compensation circuits is newly proposed. Two compensation methods are designed to overcome the process and fabrication variation of each driver

CHIANG et al.: DESIGN OF A CMOS LED PRINT HEAD DRIVER 879

IC or an LED pixel. A method of single-chip segment expo-sure is also proposed to make the power consumption more effi-cient. The power consumption is reduced by 50%. Turning on all LED diodes, the measured power consumption of a single chip is 89 mW. Based upon the device parameters of 0.5 m 1P2M CMOS technology with 3.3 V power supply, all the functions and performance of the proposed CMOS LED print head driver with compensation circuits are successfully tested and proven through measurements. The area including ESD I/O pads is 2000 2000 m . In the future research, the proposed chip will be adaptively tested on the LED printers.

ACKNOWLEDGMENT

The authors acknowledge National Chip Implementation Center, Taiwan, for their support in chip fabrication.

REFERENCES

[1] M. Maier and E. Creutzmann, “Precision LED character generator,”

Proc. IEEE CompEuro, pp. 2/75–2/77, 1989.

[2] D. B. Carlin and Y. Tsunoda, “Diode lasers for mass market applica-tions: Optical recording and printing,” Proc. IEEE, vol. 82, no. 4, pp. 469–481, 1994.

[3] Y. Kusuda, N. Komaba, Y. Kuroda, S. Ohno, and S. Tanaka, “64-Bits integrated light-emitting device array with shift register,” in Proc. IEEE

Int. Electron Devices Meeting, 1992, pp. 427–430.

[4] W. R. Imler, T. Hildebrandt, S. Poalini, K. D. Scholz, M. Cobarru-vuaz, and V. K. Nagesh, “Design and fabrication of 600 DPI light-emit-ting diode print heads using flip-flop solder bump technology,” in Proc.

IEEE Electronic Manufacturing Technol. Symp., 1994, pp. 154–159.

[5] M. Koizumi, M. Nobori, H. Tohyama, M. Ogihara, and Y. Nakamura, “High-speed chip-matrix 1200 dpi LED printhead,” Proc. SPIE, vol. 4300, pp. 249–255, 2001.

[6] H. Tohyama, S. Ozawa, Y. Kitayama, T. Yamashita, and Y. Nakamura, “The fine pitch direct bonding technology for chip interconnection,” in

Proc. IEEE IEMT/IMC Symp., 1998, pp. 82–87.

[7] M. Ogihara, H. Fujiwara, M. Mutoh, T. Suzuki, T. Igari, T. Sagimori, H. Kurokawa, T. Kaneto, H. Furuta, I. Abiko, and M. Sakuta, “LED array integrated with Si driving circuits for LED printer head,”

Elec-tron. Lett., vol. 42, no. 15, Apr. 2006.

[8] Y. Nakamura, “The trends in LED Printheads,” OKI Techn. Rev., vol. 70, no. 2, pp. 8–11, Apr. 2003.

[9] D. J. Luman, “Electrophotographic Printer and Method of Control Light Emitting Element in an Electrophotographic Print Head,” U.S. Patent 6476845 B2, Nov. 5, 2002.

[10] A. Nagumo, “Driver Circuit and LED Head Incorporating the Drive Circuit,” U.S. Patent 6535235, Nov. 15, 2003.

Cheng-Ta Chiang (S’00–M’05) was born in Taiwan,

in 1977. He received the B.S. degree in electronics engineering from Chung Yuan Christian University, Jhongli, Taiwan, in 1999, the M.S. degree in biomed-ical engineering from the National Cheng Kung Uni-versity, Tainan, Taiwan, in 2001, and the Ph.D. degree in electronics engineering from the National Chiao Tung University, Hsinchu, Taiwan, in 2006.

He was a Visiting Scholar with the Department of Electrical and Computer Engineering, The Johns Hopkins University, Baltimore, MD, from October 1, 2004 until November 30, 2005. He was included in Marquis Who’s Who in

Science and Engineering 2006–2007 and Marquis Who’s Who in the World 2008. In 2007, he was a review committee member of the National Chip

Imple-mentation Center, Hsinchu. From 2006 to 2010, he was a Member of Technical Staff at Industrial Technology Research Institute, Hsinchu, and was responsible for Nyquist & Oversampled A/D converters, and MEMS circuits. Since 2010, he was with National Chia Yi University, Chiayi, Taiwan, and serviced as an Assistant Professor in the Department of Electrical Engineering. His main research interests include analog integrated circuits, biomedical electronics, image sensor circuits and systems, sensor signal conditioning and transducers, Nyquist A/D converters, and high-resolution delta-sigma modulator.

Dr. Chiang is a IEEE conference reviewer servicing for I2MTC 2008, ISIEA 2009–2011, PECON 2010, IAPEC 2011, ICEDSA 2011, ICBEIA 2011, PEOCO 2011, CSNT 2011, and a journal reviewer for the IEEE TRANSACTIONS ONINSTRUMENTATION ANDMEASUREMENT, IEEE INDUSTRIALELECTRONICS, IEEE SENSORS JOURNAL, Microelectronics Journal, EURASIP Journal on

Advanced in Signal Processing, and an editorial advisory board member for

the Sensors & Transducers Journal.

Chun-Ting Kuo was born in Taiwan, in 1979. He

received the B.S. degree in mechanical engineering from National Central University and the M.S. degree in electronics engineering from National Chiao-Tung University, Taiwan, in 2002 and 2004, respectively.

He currently works in My-Semi Inc., Zhubei, Taiwan, He was a product manager in Silicon Touch Technology Inc. from 2004 to 2008. His main research interests have been in analog integrated circuits, D/A converter, high accuracy constant current circuit, and LED driver IC.

Yu-Chung Huang received the M.S. degree in

elec-trical engineering and Ph.D. degree in process engi-neering from the Technology University of Berlin, Berlin, Germany, in 1982 and 1985, respectively.

Since 1985, he has been a Professor in the De-partment of Electronics, National Chiao-Tung Uni-versity, Hsinchu, Taiwan. His research interests are sensors and measuring technologies.

Prof. Huang is a member of the Committee of the Chinese Metrology Society and a member of the Micromechanical Science Institute.