The Ninth International Conference on Electronic Measurement & Instruments
ICEMI’2009 ICEMI’2009
Design of a High Precision Current Source Yu Peng Yanchao Xia Shaojun Wang Harbin Institute of Technology NO.92 West Da-zhi Street Street Harbin , 150001, China Email:
[email protected]
Abstract – Current source is one of the key components in electrical measurement and control systems. However, the noise introduced by temperature drift and other interference makes it a significant source of systematic errors. To develop a high precision and stable current source, a Howland-based voltage to current conversion (V/I) circuit is presented in this paper. The current source is fully programmable using a DSP via RS-232 interface. Moreover, a sampling circuit is provided, using the sampling results, calibration is made to further reduce the output error of the current source. Under a constant 140 load, experiment results shows that current output ranges from -50mA to +50mA with error less than 3A, and features low temperature drift and distortion. This proves that such design can be an effective way to provide a precise and stable range output current. Keywords –current source; Howland; calibration. calibration.
I. INTRODUCTION With technology scaling, precision current source plays an increasing important role in automatic test, measurement and various applications [1,2]. However, meeting precision goals, keeping a stable output over temperature are becoming more and more difficult. Among all the applications, most of the current sources used in measurement field are voltage controlled current sources (VCCS), which attributes high precision and stable output. Based on ultra precision operational amplifier, VCCS attributes high precision and stable output, however, the output range of current source is often limited, and the error introduced by temperature and nonlinearity is also a challenge. Fig.1 shows a circuit, which is also known as the Howland model, uses resistor matching to accomplish the feedback loops giving a steady output across a load [3]. It uses an ultra precision operational amplifier OP177 to generate a precise and stable current output. Despite that OP177 features the highest precision performance of any other op amp, its output range is limited to -22mA~32mA [4].
To better solve this dilemma, a high precision, wide range and stable current source is proposed. Voltage is generated by a D/A converter, and then is converted to current. In the end, the current is measured, so as to calibrate the output. V/I circuit is of critical importance to the design, however, the circuit in Fig.1, using OP177 to provide a ultra precise and stable current output, can’t satisfy the wide output range of our requirement. requirement. In this case, an ‘Improved Howland’ model is proposed, which can enhance the output range of the traditional Howland model. To further eliminate errors, the close loop control structure is adapted. Compares to open loop control, this method uses a sampling circuit to complete a feedback loop giving a steady output. Using this control structure not only guarantees the stability and precision, but also cancels the temperature drift and aging of the current source. These methods on the improvement of the current source, however, need to be verified in the actual application. The practical current source, which is designed based on our methodology, is discussed in Section . Although these descriptions give a guideline, it’s necessary to introduce the software design in Section . Finally, Section and demonstrates the current source, based on the methodology this paper presents, exhibit a good quality in different kind of tests. II. HARDWARE DESIGN The current source discussed in this paper is one of the components of an automatic test system, which can provide a programmable current output. Therefore, D/A converter and V/I conversion circuit based on Howland model is only part of the design. Fig.2 shows the control structure of our design. As the figure shows, Stage 1 is an open loop current generator, the D/A converter’s output voltage which leads to V/I conversion circuit, is directly controlled by computer through a controller. Stage 2 on the contrary, provides a sampling circuit, which can be used to correct the error generated brought in afterwards.
Fig 1 Howland current source
_____________________________ ____________________ _________ 978-1-4244-3864-8/09/$25.00 ©2009 IEEE 1-1065
The Ninth International Conference on Electronic Measurement & Instruments Settings
Micro controller
D/A Converter
V/I Transition Circuit
A/D Converter
Amplifier &Filter
Sampling Resistor
Stage 1
Stage 2
Current
Fig 2 Control structure for the current source design
The design can be divided into three parts: control circuit, current generator and sampling circuit. Fig.3 shows the block diagram of the current source’s control unit. The control circuit receives message from a computer, then set the channel and value of the output. Besides, after receiving sampling value, the control circuit performs a feedback control on the output current. To fill these functions, the control circuit should not only have a capability for calculation, but also have a strong logic control ability to ensure that A/D and D/A converter works normally. Control Unit A/D converter Setting
DSP
FPGA
D/A converter
ICEMI’2009
A. Current Generator Current generator plays a critical important role in current source. It transforms the digital value to analog current output, using a D/A converter and V/I conversion circuit. In this circuit, a 16 bit D/A converter, DA7734 is selected. It features high precision and low distortion, as well as a low temperature drift. Relationship between digital input “N” and voltage output Vout can be described like this: VOUT
VREF L
(V REF H
VREF L ) N
2-1
65536
VREFH and VREFL is the high/low reference voltage of DA7734. In our design, these value are configured as +5V and -5V, respectively. Thus the output range of DA7734 is -5V to +5V, and the resolution is 0.103mV/LSB. The V/I conversion circuit is based on the Howland model, which has been introduced in chapter .The improved Howland model, without affecting the accuracy of the output current, increase the output range of the current source, and strengthen the ability to drive load. Figure 4 shows the Schematic diagram of this V/I conversion circuit.
Instrum Amp Switch
R 3
Fig.3 Diagram for control unit of current source
A3
However, it is difficult for an ordinary DSP to carry out the complex logic control, meanwhile, computing power of the field-programmable gate array (FPGA) is not well suited for calculation, use them alone will be difficult to meet the system requirements. In the end, Texas Instrument (TI)’s DSP TMS320LF2407 is chosen as the core of the control circuit, and the logic control circuits are based on Altera’s FPGA EP1C6Q240. Command and data communication between them are realized through 16-bit address and data bus. The current generator uses an improved Howland model to convert the input voltage, which is provided by a 16 bit D/A converter, to an output current. As has been discussed, the D/A converter is controlled by an FPGA, and its output value can be set by computer through DSP. Since error caused by temperature drift, nonlinearity and other noise are hard to avoid, the sampling circuit is provided so as to complete a feedback loop rectifying any error exists in the output current, keeping a favorable output. The sampling circuit is consisting of sampling resistance, instrumentation amplifier and A/D converter, these components are also controlled by FPGA, and the result is sent to DSP for further applications. The above introduction has explained the working principles of our design; then we’ll present the design in detail.
1-1066
Vin
R 1 A1
R 2
IO2
R S0 IO1
R S1
U3 U0
A2
Io Load
R 4
Fig.4 Improved V-I circuit based on Howland current source
As is shown in this diagram, Ui is the control voltage, while IO is the output current with load R L. A2 and A3 are both voltage followers. However, A3 is given in serial with another sampling resistor R S1, which is the same to R S0, this ensures that the output current IO1 and IO2 are equal and is half the value of the whole output current IO. This structure is based on this consideration: amplifiers such as OP177 are unable to output current higher enough. For example, OP177 provides a maximum output voltage of only 32mA, this of course is unable to meet the output range demand of our design. Assuming both amplifiers are ideal, the voltage between each node has a relationship given by: (1) U 3 U O U1
R2 R4 R2
U1 U 3
U 2
U I U1
(2)
(3) R3 R1 Assuming R 1=R 2= R 3=R 4, from (1) (2) (3), the following expressing for the load current can be easily derived:
The Ninth International Conference on Electronic Measurement & Instruments I 0 I 01 I 02
U 2 U 0 R s
2
2U I Rs 4
From this formula, we can see that the current output is independent of the load R L. However, the output voltage range of OP177 is only 13.5V(when the supply voltage is 15V), as a result, when the load is higher than 140 , output current cannot achieve the designed 50mA range. To solve this problem, we can modify the proportion coefficient between R1 and R3, because: U U 0 3 R1 U I 5 I 0 2 3 R s R3 Rs In this way, we can enhance the driving ability of the current source without affecting the current’s output range. B. Sampling Circuit Design Facing all kinds of noise, digital calibration can be used to help the whole system obtain a better performance that analog circuits cannot provide intrinsically. In order to calibrate the current source digitally, sampling circuit is designed to obtain the value of output current. It transfers the value of current to digital information that can be easily modified. In our design, although the improved Howland model features an output of high accuracy, this structure without feedback circuit, is hard to guarantee a stable current output [5]. Therefore, we present a high-precision sampling circuit, in accordance with the DSP, to digitally compensate the output current. The principle of designing sampling circuit is to ensure the high resolution, as well as low measurement error, without affecting the original current output circuit. Sampling circuit includes signal amplification and analog to digital convert circuits. Signal amplification circuit, which is based on low-pass filter and instrumentation amplifier, extracting small difference voltages between sample resistance, while rejecting common-mode voltage. Since this measurement is based on Ohm’s law, high precision and low temperature coefficient sample resistance becomes a must [6]. In this design, we select resistance with a static error of 0.1‰ and a temperature drift coefficient of 5ppm/. The analog to digital conversion circuit is based on a 24 bit A/D converter AD7734. The signal to be measured is voltage between sampling resister, so there is a need to reject any signals that are common to the two ends. In this occasion, instrumentation amplifier, which is a closed loop gain block that extracting differential signal while canceling common signals. In order to ensure the accuracy of the sampling circuit, PGA204 is selected for differential signal amplification. PGA 204 is a gain-programmable instrumentation amplifier, its high CMR and low imbalance error makes it ideal for our design.
ICEMI’2009
To suppress the common 50Hz frequency interference, here we present a low pass filter circuit as is shown in Figure 5.
Fig.5 low pass filter for I nstrumentation Amplifier
In this block, R a and Ca, R b and C b form a differential input low pass filter. The value of the resistor and capacitor should be exactly the same. Any dismatch in the value may cause a common-mode error. To avoid this effect, capacitor C2 is added to the circuit, whose value is 10 times to Ca. A/D converter is the core of the sampling circuit, and its parameters are directly related to the performance of the sampling circuit. Based on the consideration of precision and stability, the design uses the - type A/D converter AD7734. Its resolution is 24-bit, non-linear output is lower than 15ppm, and the temperature drift coefficient is lower than 5ppm/. All these features can fully satisfy the requirements of the sampling circuits. AD7734 chip includes mode register, communication register, data register and so on. Through these registers, we can set the input range, sampling rate, output mode of the chip. All the operations to the devices is carried out through the communication register of the AD7734. To operate the A/D converter, we developed the control logic circuit base on FPGA. The conversion process is described in Figure 6. Wait for Start
Receive Settings
Wait
Write Com register
Data output
Write mode register
Serial-parallel
Receive data
Start conversion
Fig.6 Flow chart of the conversion state of AD7734
1)Start A/D conversion, during the “Receive Settings” period, the control logic receives configuration data, which is used to set the conversion mode of the A/D converter; 2)After receiving the configuration data, control logic enters into the "written communication register" state. During this period, the control logic write the AD7734’s communication register through the serial interface (SCLK), consequently, indicates which register
1-1067
The Ninth International Conference on Electronic Measurement & Instruments is going to be operated, and whether the next operation is read or write; 3)Based on the data written in communication register, the procedure enters into the "write mode register" state. Control logic write the configuration data, which is received in procedure 1, to AD7734’s mode selection register, to make sure A/D conversion process is under specified mode; 4)After setting conversion mode, start A/D conversion. When conversion ends, the control logic come into the "receiving sampling data" and "output sampling data" state. Then, the conversion data is sent back to FPGA through the serial communication interface of AD7734. Later, FPGA transfers the sampling data to DSP for further calculation; 5)After the output of sample data, control logic returns to the "wait" state, waiting for the command of the next conversion. III. SOFTWARE DEVELOPMENT The control software for the design is based on digital signal processor TMS320LF2407, using C for programming. The software set the value of the output current, and use feedback algorithm for compensation. Therefore, software of the design includes current output program and feedback control program, the main procedure of the software is explained in figure 6:
ICEMI’2009
Subsequently, based on the information given by DSP, FPGA is used to configure the control and data register of D/A converter, consequently, control of the output is realized. B. Feedback Control Program In order to further eliminate the error caused by environment factors, feedback control structure is established. According to sampling results, we devise the feedback control software, which is run on DSP, to compensate the output current. Based on this idea, we can design the closed-loop control program using the follow methodology. 1)Set the gain of instrumentation amplifier, at the same time, select the mode of A/D converter; 2)Compare the result of sampling circuit with the set value, their difference is the output error; 3)According to the output error, calculate the amount of compensation, modify D/A converter’s input value so as to change the output voltage, thus rectify the output current; 4)Repeat procedure (2), when the output error of the current is small enough, stop this program, otherwise repeat procedure (3) until the output current can meet the precision requirements. IV. TEST AND CONCLUSION
A. Current Output Program The current output block mainly performs two functions. One is to receive messages from the computer, then translate the message to the control settings, and pass this message through FPGA to the D/A converter; another function is accept the sampling data from sampling circuit, then calibrate the output current. At the beginning, DSP obtains command and data from computer through the serial communication interface (SCI), then after decoding, transfer the message to FPGA through the 16 bit data and address bus. This procedure is controlled via “read/write” and “enable” signal from the DSP.
Fig.8 Flow chart of the control software for current source
1-1068
In order to further verify the design, we carried out various testing experiments for the current source. These experiments verified the output range, accuracy and time stability of the current source. Our tests involve three main steps: Step 1: At room temperature, put a high precision 50 resistor in series with the HP34401A multimeter into the current source circuit output loop. Set current output as 100uA, 1mA, 10mA, 50mA and the corresponding negative value, read the output current’s value shown in multimeter. Each test are repeated for ten times to get a intermediate value. Step 2: Maintain the output current unchanged, using the same way to test its error after 24 hours. Step 3: Execute feedback control program, then repeat Step 1, compare the results to when there’s no feedback control, verify the effectiveness of the feedback control structure. The overall accuracy, stability and temperature coefficient are obtained within feedback control.
The Ninth International Conference on Electronic Measurement & Instruments
ICEMI’2009
Table 1 Measured Result
Setting
Test results (without feedback)
Test results (with feedback)
Overall accuracy (% of full scale)
1-day stability (% of full scale)
100A 1mA 10mA 50mA -100A -1mA -10mA -50mA
102.8A 1.004mA 10.005mA 50.009 mA -96.7A -996.5A -9.994mA -49.992mA
101.2A 1.001mA 10.001mA 50.004 mA -98.7A -998.5A -9.997mA -49.995mA
0.012 0.010 0.010 0.040 0.013 0.015 0.030 0.050
0.015 0.020 0.030 0.050 0.017 0.020 0.020 0.060
Temperature coefficient (ppm of setting) 40 35 30 25 40 35 30 30
Finally, the result shows that the output range of the current source is -50mA to 50mA. After the feedback control module is put into use, the error of the output current reduced from 10A to 5A. In general, this design of current source effectively improves the output range and precision of the output current. I. CONCLUSION In this paper, design methods and practical results of a current source are presented. Improved Howland model circuit shows precise, stable current output, as well as more than 2X improvement in the output range of current source. The output current features better performance because of the digital calibration. Using the sampling value provided by sampling circuit, the DSP compares the real value of current with the expected one, and execute compensation on the current output, thus adjust the output. Over all, this design provides a method for current source design that evidently meets performance goal. Nevertheless, errors introduced by temperature drift can’t be totally cancelled, these issues should be taken into account in later designs. REFERENCES [1]
[2]
[3] [4] [5]
[6]
Paul Annus, et al, “Excitation Current Source for Bioimpedance Measurement Applications: Analysis and Design”. IEEE Instrum and Meas Technology Conference, pp. 848-853, 2008. Marcos M Pelicia, et al, “Fully Integrated Programmable Howland Current Source for Sensors Recitation”. Fourth IEEE Caracas Conference on Devices, Circuits and Systems.Volume C. pp. 208, April 17-19 2008. Charles Kitchin, “Lew Counts: A designer’s guide to Instrumentation Amplifier”. Analog Devices. 2006 Analog Devices: User’s Manual:Ultraprecision Operational Amplier OP177.1995. Luo Xiaoqiao Zhou Lili. Design of Intelligence Current Source with High Precision and Wide Range, Journal of EEE. Vol 26, NO 6, Dec 2004. Xu Keqiang. Design of digital control direct circuit current source in high accuracy and wide range, Journal of Chongqing Technology and Business Univ. Vol.23, No 2, 2006.
1-1069
