Chapter IV. Data Acquisition System
Phase 1 System
The data acquisition system for Phase 1 of this project consisted of a Campbell Scientific AM416 mechanical relay multiplexer to switch between the 28 strain gage channels of the single pile being tested, a Campbell Scientific 21X data logger to collect the strain gage voltages and convert them to digital data, and an IBM compatible 386 computer to program the data logger and to store test data. This system worked well for single pile testing, but with the necessity to collect data from six piles at a much higher sampling rate, a completely new system was needed for this phase of the project.
A considerable effort was put toward finding an adequate system that could be purchased off the shelf. Any system capable of reading so many strain gage channels at an approximate required rate of 10 Hertz was much more costly than the project budget allowed. An effort also was made to buy into a system being purchased by the Civil Engineering Structures Division, but again system requirements could not be met with the funds available. These findings led to the idea of designing and building a data acquisition system. The Electrical Engineering Department at USU was contacted to determine feasibility of such a venture. An undergraduate student volunteered to take on the task, with faculty assistance, of designing the data acquisition system, which also could be used as a senior design project.
System specifications called for designing a system that could sample 256 channels at a rate of 10 Hertz, and interface with a personal computer where the testing could be monitored and the data stored in a user-friendly format. The specification for 256 channels was made so that future pile group tests might be expanded to include nine piles with 28 strain gages each with several extra channels for other instruments.
The system designed was based around a commercial analog-to-digital (A/D) circuit board manufactured by National Instruments, which interfaced with the piles through custom designed printed circuit boards. National Instruments LabVIEW™ software programs were written to control the data acquisition processes.
The basic unit of the data acquisition system is the Wheatstone bridge circuit. Printed input circuit boards were designed to accommodate quarter-Wheatstone bridge circuits for every strain gage. Figure 8 shows a Wheatstone bridge as configured on the input circuit boards. One side of the bridge is made up of one 160-ohm resistor and the 120-ohm strain gage. The other side of the bridge contains a 160-ohm resistor, a 100-ohm resistor, and a 25-turn potentiometer that was included in the circuit to provide a means of balancing the two sides of the bridge to compensate for variance in the resistance of the resistors. The values of the resistors were chosen to minimmize induced noise.
Analog multiplexing was used to allow use of a single A/D converter. Voltage readings from each side of the bridge were routed to a separate multiplexer (mux) allowing simultaneous passing to a differential operational amplifier (op amp). The differential op amp receives small voltages from the dual muxes, which correspond to voltages on each side of the Wheatstone bridge, and takes the difference between them, multiplies the difference by desired gain, and provides a single output. The signal is then sent to a low pass filter with a cutoff frequency of 230 Hertz and a settling time of around 6 milliseconds. Figure 9 shows the flow sequence of a strain measurement. An attempt was made to filter out noise signals at lower frequencies, particularly in the 60 Hertz range, but the settling time would have been too great to allow the system to operate at the desired sampling rate of 10 Hertz. The speed of the filter drove the design of the system in order to run at the desired rate of 10 Hertz, all sampling for 16 lines in parallel had to be collected in 6.25 milliseconds. After filtering, only 0.25 milliseconds remained for all other settling and switching functions. All readings get sent to the A/D converter after filtering.
Sixteen Wheatstone bridges were placed on a single input circuit board along with the two muxes, the op amp, and the low pass filter. The whole data acquisition system has a maximum capacity of 16 of the input boards. Figure 10 shows the input boards' design with bridge circuit traces and silk screened component locators.
The A/D converter board purchased for this project was manufactured by National Instruments and was of the type PC-LPM-16, and was plugged into an IBM compatible personal computer. The PC-LPM-16 had 16 single-ended analog and eight digital input lines, along with digital output lines that were used to control muxes on each of the input boards. Each of the 16 single-ended analog input lines was dedicated to one input board and was connected by means of a printed interface bus circuit board with card edge connectors for the input boards and a ribbon cable that ran from the interface bus to the PC-LPM-16 board. LabVIEW software is a graphical programming language from National Instruments and interfaces automatically with the PC-LPM-16 A/D converter board and was used to control all functions of the data acquisition system.
Power for exciting the strain gages was provided by a Sola model SLS-05-120-1 linear, open frame, DC power supply. It is capable of providing 12 amps of current at +5 volts. A smaller Sola model SLS-05-030-1 power supply was used to power the analog multiplexers with a possible 3 amps of current at -5 volts.
Once the design of the input circuit boards was completed, the blueprints were sent to Quick Turn Circuits in Salt Lake City, Utah, where the circuit boards were actually fabricated in a matter of days. All connection holes were predrilled and all component positions were labeled with silk screening. Each of the 16 input boards was built up with the following: 48 resistors of two types for the Wheatstone bridges, 16 potentiometers, two 16-port screw-type terminal blocks for the strain gage wire connections, two 28-pin analog multiplexer sockets, one 14-pin amplifier chip socket, four amplifier resistors, one capacitor for the low pass filter, and one filter resistor. The multiplexer and amplifier chips then were pressed into their respective sockets, with the only requirement being that they were installed with a small detent in each chip facing to the left side of the input board.
The interface bus also was fabricated by Quick Turn Circuits and was assembled next. The components soldered to the interface bus included 16 20-contact card edge connectors for connecting to the input boards, one 50-pin socket-to-socket connector for the ribbon cable connection to the PC-LPM-16 A/D board, one four-position screw type connection terminal block for power supply hook up, and one 20-pin socket for the signal buffer chip, which amplifies control signals coming from the computer. To assure that the card edge connectors would not be damaged or get pulled off by the repeated plugging in and unplugging of the input boards, holes were drilled through the interface bus to match holes predrilled in the connectors. Machine screws were inserted through the holes and fastened on the other side with nuts. After installing the signal buffer chip, the interface bus assembly was complete.
All of the components had to be housed, therefore a decision was made to bring all components external to the personal computer together in one box, including input boards, interface bus, power supplies, fuses, and all components necessary to control the pile group loading system. This would allow for easier management of all of the testing systems. Figure 11 shows the layout of the data acquisition system box.
The box was constructed of 4.8 mm (0.2 in.) thick plywood and is 476 mm (18.75 in.) long, 422 mm (16.6 in.) wide, and 178 mm (7.0 in.) high. The inside surface of one of the long sides was slotted with 16 3 mm (0.13 in.) slots spaced at 20.3 mm (0.8 in.) intervals to be used as guides for holding the input boards when plugged into the interface bus. A cooling fan was added to the box to provide the 0.57 m3/min (20 ft3/min) of convective cooling air required by the large power supply. A circular hole was cut in one of the short sides to accommodate the fan. Two slots were cut in the wall opposite the fan. Through one of the slots the ribbon cable passes in going from the interface bus to the A/D board in the computer. The second slot was positioned so cooling air could be drawn into the box and pass across the heat-sink resistors on the underside of the large power supply. Other holes were drilled in the box for power cords, one for bringing in power and the other for sending power to the pile group loading system. The component positions were laid out and holes were drilled for mounting bolts, after which the box was painted inside and out. Brass bushings were used as standoffs for the power supplies, interface bus, and the loading system switching relays and logic circuit.
The loading system logic circuit and relays were wired so they could be controlled by the A/D board in the computer and the LabVIEW software. Wires for power and signal passing were soldered to the proper pins for these functions on the interface bus underside where the 50-pin socket connector pins protruded.
Once all of the other components had been mounted, a wiring scheme had to be established and all of the wiring done. Fourteen-gauge wire was used so there would be as little chance as possible of power being dissipated by the wires. Wiring connections were either soldered joints or forked-end connectors fastened in place with screws. Heat shrink tubing was used in an attempt to cover as many exposed and potentially hazardous connections as possible. The power supplies were grounded with wires connecting them to the ground wire from the supply cord. For grounding the piles, a short bolt with enough nuts to fasten each of the grounding wires from the piles was attached to the frame of the large power supply. Finally, all of the wires were connected to either hot, neutral, or ground wires with wire nuts and then tied in place with zip ties.
A two-piece lid for the box was constructed of plywood and fitted with handles for easier removal. Cabinet door hardware was installed to hold the lid in place. The lid was made in two pieces so that the inside of the box could be accessed easily without having to disturb wire bundles coming in from the piles, which are trapped between the lid sections to keep them from being jostled when the lid is in place. It is important to keep the lid on the box as much as possible when running the full system so that cooling air will be drawn across the large power supply through the air vent. Otherwise, it might overheat because air does not circulate through the box properly. If only one- or two-piles are hooked up it is not as important to keep the lid on as the large power supply does not heat up as much.