Abstract: This paper introduces the design of a high-precision semiconductor laser driver power supply system based on the DSP TMS320F2812 control module. The system uses a high-power Darlington transistor as the regulating transistor and a current negative feedback circuit to achieve constant current output. The integrated analog-to-digital converter within the DSP samples the output current, and after processing with a PI algorithm, controls the PWM output to achieve dynamic error adjustment, eliminating static errors in the circuit. To improve system stability, overcurrent, overvoltage protection, and delayed soft-start protection are added. Results show that within the output current range of 10–2500 mA, the absolute value of the output current change is less than 0.1% + 1 mA of the output current value, thus ensuring the reliability of the semiconductor laser operation.
Keywords: DSP; semiconductor laser; PI algorithm; PWM
0 Introduction
Semiconductor lasers (LDs) are solid-state light sources that have been widely used due to their advantages such as good monochromaticity, small size, light weight, low cost, and low power consumption. LDs are ideal electron-photon direct conversion devices with high quantum efficiency; however, even small changes in current and temperature can lead to significant variations in their output optical power. Therefore, LDs require very high driving current, necessitating a low-noise, highly stable constant current source, which is difficult to meet with conventional power supplies [1-4] . Furthermore, transient current or voltage spikes, as well as overcurrent and overvoltage, can damage semiconductor lasers. This paper will use the TI DSP chip TMS320F2812 as the control core to implement a dual-closed-loop high-precision semiconductor laser driving power supply system with multiple protections.
1 System Overall Design
Constant current sources are generally implemented using a voltage-controlled constant current source (VDC) method, which combines integrated operational amplifiers with discrete components and a microcontroller. Compared to constant current sources composed of purely analog components, this method significantly improves the accuracy and linearity of the constant current. However, in this method, the microcontroller is used to display and control the given voltage, but it does not detect and control the output current in real time, making it an open-loop control system, which affects the stability and accuracy of the constant current source. This system consists of a voltage-controlled constant current circuit, a signal sampling and conditioning circuit, a protection circuit, a keyboard, an LCD display, an RS232 communication interface, and a DSP processor. The system block diagram is shown in Figure 1.
The input is given via the keyboard and displayed on the LCD. Simultaneously, the F2812 processor processes the input and outputs a PWM signal with the corresponding duty cycle. The PWM signal is then filtered, amplified, and conditioned to achieve a D/A conversion, which serves as the control voltage for the "Voltage-Controlled Constant Current" (VI Constant Current) module. The F2812 samples the output current in real time. The sampled data is digitally filtered, analyzed, and compared with the given current value. The difference is used as the input to the PI control algorithm expression. The PI calculation yields the control quantity U, which adjusts the PWM output to achieve constant current.
2 System Hardware Design
2.1 Implementation of DC Power Supply Module
The DC power supply module mainly consists of a transformer, rectifier, filter, voltage regulator, and a current amplification circuit. The DC power supply module is shown in Figure 2. +15V is used to power the voltage-controlled constant current module and operational amplifier; -15V is used for the negative power supply of the operational amplifier; and +5V is used to power the CNC module. The +5V is regulated by a high-precision, high-stability voltage regulator chip before powering the DSP processor.
The "current amplification circuit" consists of resistor RP3 and a high-power Darlington transistor TIP147. Adjusting RP3 allows for a high current output of over 2A from the +15V current. To reduce ripple in the DC current, an RC-π type active filter method is used. Variable resistors RP1 , Q1 , and C3 , along with RP2 , Q2 , and C4, form two RC filter circuits to efficiently filter the +15V and -15V power supplies respectively. The transistor is an NPN type, and its current amplification effect indirectly increases the capacitance of the filter capacitor. If the amplification factors of Q1 and Q2 are β1 and β2 , then the base capacitors C3 and C4 of Q1 and Q2 are equivalent to the emitter capacitors, which are (1+ β1 )C3 and (1+ β2 ) C4 respectively . This achieves large-resistance, large-capacitor filtering while reducing the circuit size. In the diagram, D5 and D6 indicate power failures, while D7 and D8 protect the voltage regulators LM7815 and LM7915. When there is a load at the output, if the input of the LM7915 voltage regulator is open-circuited, the LM7915 will have no output, and +15V will be applied to the output of the LM7915 through the load, potentially damaging the LM7915. The protection principle for the LM7815 is the same.
2.2 Purchase of constant current source module
"Voltage-controlled constant current" controls the output current by controlling the change in input voltage. The principle of the constant current source circuit is shown in Figure 3. Closed-loop negative feedback, i.e., inner closed loop, is achieved through hardware circuitry. In Figure 3, resistors Rs , R4 , R5 , Rf and operational amplifier U5 constitute the feedback network. Assuming operational amplifier U4 is ideal, and the input voltage is V5 , the output voltage is U0 . From the "virtual short" property of the operational amplifier, we can obtain:
When Rs , R5 , and Rf remain constant, the input voltage and output current remain constant. The stability of the operational amplifiers U4 and U5 , and the resistors Rs , R5 , and Rf themselves, plays a decisive role in the stability of the constant current source. Therefore, U4 and U5 are selected from high-precision operational amplifiers OP-27, with a drift of only 0.2μV/℃ and a maximum noise voltage of 0.25μV. R5 and Rf are selected from resistors with low temperature drift coefficients and high precision. The sampling resistor Rs is selected from a 0.01Ω high-power manganese copper wire resistor with an accuracy of 1%. Q5 is a high-power Darlington transistor 2SD1559. Since integrated operational amplifiers generally operate in a low-current state, a small-power transistor Q4 (9014) is used to drive Q5 . C5 , C16 , D9 , and L1 form a low-pass filter to reduce the impact of high-order harmonics in the power supply on the LD. D5 acts as a current suppressor when Q5 is cut off.
2.3 Implementation of A/D and D/A Modules
The F2812 chip has a built-in 12-bit ADC (Analog-to-Digital Converter) input with a voltage range of 0–3V. The sampling resolution of the 12-bit ADC is (3.0V–0V)/2^ 12 = 0.73mV. The F2812 sets the A1A0 pins of the PGA103 according to preset current values ( A1A0 = 00, A1A0 = 01, A1A0 = 10 correspond to amplification factors of 1, 10, and 100, respectively). Signal conditioning is shown in Figure 3. The F2812 does not have a built-in DAC module; an external D/A converter chip is required to implement D/A functionality. The conversion accuracy is directly proportional to the price of the chip, which undoubtedly increases hardware costs.
The PWM signal provided by the F2812 chip is a pulse width modulation (PWM) signal with variable period and duty cycle, high level V<sub> H</sub> = 3.3V and low level V<sub> L</sub> = 0V. According to Fourier transform, a unipolar PWM signal symmetrical about the origin of the time axis can be expressed as:
In the formula: T is the signal period; n = ±1, ±2, ±3…; A <sub>n</sub> and B <sub>n</sub> are their respective independent Fourier coefficients.
From equation (3), it can be seen that as long as the high-frequency DC component An is filtered out, the output voltage can be obtained by changing the duty cycle q (q=0~1) of the PWM signal. Since the third-order low-pass filter has better performance than the first-order and second-order low-pass filters [5] . A third-order Butterworth feedback active low-pass filter is designed using the "normalization" method, as shown in Figure 4. The transfer function of the low-pass filter is expressed as:
In the formula: G,b <sub>n-1</sub> ,…,b <sub>0 </sub> are appropriately chosen constants. Figure 4 shows that a low-pass filter must meet the following conditions to satisfy equation (4):
In the formula:
By normalization, the cutoff frequency fc (Hz) and capacitance C21 are normalized, so the resistance coefficient is K=100/fcCˊ, where Cˊ is the value of C21 in μF. To make the gain G=2, we can see from Table 2-54 in reference [6] that the coefficients of resistors R6～R10 and capacitors C22～C23 (μF) corresponding to K＝1 are 2.491, 2.339, 0.692, 11.043, 11.043, C21, C21 respectively. When fc=1000 and Cˊ=0.01, R6～R10 and C21～C23 in Figure 4 are 24.491, 23.39, 6.92, 110.43, 110.43, 0.01, 0.01, 0.01 respectively. Simulation results using EWB software show that the third-order filter circuit achieves excellent filtering performance. The Butterworth filter exhibits no ripple within the passband, which ensures the accuracy of the PWM to D/A conversion. The simulation results are shown in Figure 5.
2.4 Keyboard and Display Implementation
The keyboard is used to input preset current values and allows for real-time modification. It has 16 keys: "0-9" and "·" keys for numeric input; "ENTER" and "CANCLE" keys for confirmation and cancellation; "↑" and "↓" keys for increasing and decreasing the step size; and the "NUM" key for step selection. The preset current step size is available in ±10mA and ±1mA, allowing input of current values within the range of 10-2500mA. The preset current is displayed on the LCD after pressing "ENTER". For data display, the commonly used LCD1602A is selected, which displays the preset output current value and the real-time sampled current value in two separate lines.
2.5 LD Protection Circuit
The PN junction of a semiconductor laser (LD) is very fragile and easily damaged. Sudden current surges can easily damage the end face cavity mirrors of the semiconductor laser, causing permanent damage [7-8] . Slow start (also known as soft start) refers to the process where, after the drive power is turned on, the control voltage Vs is not suddenly applied to the entire constant current circuit, but rather gradually rises from zero to Vs within a set time. Connecting several forward-conducting diodes in series with the laser L can effectively extend the lifespan of the LD, because when a large forward voltage occurs, these diodes conduct, and the current will not flow through the laser tube LCD, thus avoiding damage to the LD [9,10] . A small capacitor is connected in parallel across the LD, along with a reverse diode to prevent the LD from being damaged by excessive reverse voltage. To prevent overcurrent, software and hardware protection can be used. The sampled current value is processed and compared with the current limit value. If the current exceeds the limit value, the switching transistor Q6 is turned on, V4 is clamped to 0V, and the regulating transistor is cut off, achieving the purpose of current limiting.
3 System Software Design
The software is written in assembly language and allows real-time modification of the current setpoint via the keyboard. The LCD displays both the setpoint and measured values. For ease of debugging, the software employs a modular design, primarily including the main program, subroutines for setting the current, LCD display, and PI control. System initialization includes initialization of the DSP peripheral interface chip and the current setpoint, while keyboard scanning involves adjusting the setpoint and step size. The flowcharts of the main program and the outer loop adjustment program are shown in Figure 6.
4. Conclusion
This design employs a combination of hardware closed-loop negative feedback and digital closed-loop to construct a dual-loop constant current power supply. The hardware closed-loop negative feedback provides strong constant current characteristics and reduces the workload of the digital closed-loop. The digital closed-loop primarily plays a fine-tuning role, improving the system's constant current accuracy. Furthermore, fully utilizing the built-in resources of the F2812 simplifies the complexity of peripheral chip design. Simultaneously, the 16 ADC channels and PWM outputs can control and measure multiple constant current power supplies. Therefore, this system has broad application prospects in fiber optic sensing, fiber optic communication, and laser power supply for active current transformers, among other areas.
References:
[1]Tucker RS. High—speed Modulation of Semiconductor Laser[J].Light-wave Technology,1985,3(6):1180-1192
[2]Doerr C R.Direct Modulation of Long-cavity Semiconductor Lasers[J].Light-wave Technology,1996,14(9):2052-2061.
[3] Lü Yuxiang, Yuan Kuo, Zhang Zhiqiang, et al. Low-power, high-stability semiconductor laser power supply [J]. Modern Electronics Technology, 2008, 31(7): 105-106, 109.
[4] Tang Min, Han Hai. Semiconductor laser driving power supply based on NCP5662 [J]. Electronic Measurement Technology, 2008, 31(3): 47-50.
[5] Han Antai, Liu Zhifei. DSP Controller Principles and Their Applications in Motion Control Systems [M]. Beijing: Tsinghua University Press, 2003.
For details, please click: Design of High-Precision Semiconductor Laser Driver Power Supply System
