EE444 Embedded Systems Design - Student Projects 2012: Harvey, Montz and Stribrny

Integrated Footstrike Sensing System

Justin Harvey, Ben Montz, Joseph Stribrny

Abstract – The Integrated Foot Strike Sensing System (IFSS) was designed to aid medical professionals in rehabilitation and sports injury prevention applications. This system operates in two modes: in the real-time mode, force readings are sampled every half-second and transmitted to a computer running a stand-alone application that displays the data. In the off-line mode, the average force per foot strike, the average foot strike duration and frequency are recorded for later analysis.

Background

The IFSS was designed with one particular purpose: to give medical professionals and athletes a quantitative means of measuring the force, or the impact that their feet are exposed to while running. The capabilities of the IFSS have been designed with the particular intent of appealing to a military consumer. Overtraining injuries in the military are so prevalent, they account for 40% - 50% of all outpatient clinical visits. With the IFSS, a consumer may be capable of diagnosing and adjusting their gait to prevent problems such as shin splints or overpronation.

What Does it Do?

Force Sensing:

The IFSS system utilizes force sensitive resistors (FSRs) mounted on the insole of a shoe to gather data related to the force experienced by different areas of the foot during running, walking, or even remaining stationary. IFSS V 1.0 utilizes three FSRs placed in areas of particular interest as shown in Figure 2. Sensor placement on the heel and on the ball of the foot is the most necessary data, as both are considered the areas exposed to the most force during normal running and walking. Future versions will implement the fourth sensor under the arch in order to test for fallen arches (flat feet) in patients.

Fig. 2: Sensor Placement on Insole

Wireless Communication:

Fig. 3: Nordic nRF24AP1 ANT Transceiver

In order to analyze the data that is acquired with the FSR, the lower power, highly versatile ANT protocol was implemented. In the real-time mode, force data from the sensors is transmitted live to a receiving station that then analyzes the data. The nRF24AP1 ANT Transceiver with a trace antenna was attached to the prototype board using a simple header. The Nordic USB ANT stick was used with virtual com port drivers to connect to the PC.

User Interface:

The IFSS user interface provides a means of receiving and interpreting data transmitted wirelessly from the sensor unit. The user interface interprets data by plotting the most recently received data as colored circles on a footprint plot using blue to red gradient color mapping where blue represents no force applied to a sensor, and red represents a large force applied to the sensor. The received data is also plotted in separate Force VS Time plots for each sensor. The option of showing plots of the entire data set or just the 50 most recent data points is given. The pressure applied to each sensor is displayed under each plot. Recorded data can be saved to .csv files to be analyzed further using other software, or to be loaded by the IFSS software at a later point in time. The option is given to clear the plots and the stored data in case the need to restart a test occurs.

Fig. 4: Graphical User Interface

The Design Process

Hardware:

The first task in the hardware design for the system was determining a functional biasing network to use in conjunction with the FSRs. The basic design of the biasing network is shown in Figure 5. The physical circuit designed for the empirical tests of the sensor hardware is shown in Figure 6.

Fig. 5: Sensor Biasing Circuit Schematic

Fig. 6: Force Sensitive Resistor Testing Unit

After establishing the functionality of the FSRs, a second prototype was developed, this one implementing a 3 V voltage regulator, and five FSRs connected in a parallel network. Figure 1 shows the schematic representing the complete sensor biasing network. The subsequent step in the design of this first prototype was to build a PCB layout and to mill a board. The layout is shown in Figure 8, while the first milled prototype is shown in Figure 9. Testing of the completed board indicated the existence of a short, which the group was unable to isolate. Lessons were learned in the implementation of this prototype and were applied to the final protoype.

Fig. 7: Schematic for Complete Biasing Network

Fig. 8: PCB Layout for Complete Biasing Network

Fig. 9: Milled Prototype of Complete Biasing Network

Some of the improvements implemented on the final circuit schematic included utilizing a lower voltage dropout regulator to supply the wireless transceiver, as well as a separate regulator to supply the MSP430 microcontroller. The schematic for the final prototype is shown in Figure 10. The PCB layout for the final prototype was approached with more care for the last iteration in order to avoid the complications experienced with the previous board. The minimum trace spacing was increased to help prevent shorting, and pins were added to allow the attachment of a ribbon cable for communication with the microcontroller. To optimize space efficiency, the 20 mm battery holder was soldered directly to the board.

Fig. 10: Final Prototype Schematic

Fig. 11: Final Prototype PCB Layout

Fig. 12: Final Prototype PCB

Software:

The software design consisted of several stages. The first stage of fully successful software implementation was in programming the MSP430 microcontroller to send messages wirelessly using the ANT protocol through the Nordic nRF24AP1. Testing for successful implementation was one of the greater challenges early in the project. Without the wireless ANT stick software developed for the user interface side, there was no way to verify successful transmission of messages. A network analyzer with an RF antenna was used to verify the existence of a wireless signal operating in the 2.4 GHz band while the transceiver was set to transmit on broadcast mode. A mode that requires acknowledgement of received data was later developed but was unnecessary for real-time mode operation.

With verification that a signal was successfully being transmitted, the receiving side software application started development. A MATLAB interface was chosen for receiving data and interacting with the ANT radio for ease of implementation, both for the development of graphical user interfaces (GUI), as well as for the built in serial communication capabilities. Eventually both of these functions were successfully consolidated into one GUI, as shown in Figure 4.

The next programmable software involved the setup and control of the FSRs within the MSP430. The sensor biasing network was connected in such a way as to provide an input signal to GPIO pins on the MSP430 that would then be processed by the 12-bit analog to digital converter (ADC12). A timer was set up to trigger an interrupt at least every half second that would call a function written for the operation of the ANT transceiver to broadcast a message. The data in the message was the raw data from the ADC12 registers correlating to particular FSRs. In order to make logical sense of the raw data that was being transmitted, a means of calibrating the sensors to some tangible relation was required. The calibration test bed in Figure 12 was built in order to empirically determine an equation for calibration. A gram scale was used to measure the total weight added to the test bed before the data from the sensors was read. As weight was added, measurements were made for each sensor and plotted, as shown in Figure 13. The resulting equations were utilized on the receiving software side to adjust the displayed results to match the empirical measurements.

Fig. 13: Calibration Test Results

Fig. 12: Calibration Test Bed

The last software segment that was developed was the design for the off-line mode. The final implementation was capable of adjusting the sampling rate to correspond to the duration of the foot strike. It cumulates and averages the data that is being sampled. One of the advantages is its capability to sample higher forces than the real-time mode due to a programmed force cutoff designed to throw out suspect data.

Final Product

The final prototype is shown in Figure 14, where the placement of the sensors in the tested product is indicated on the cardboard cutout. Data collected from the operation of this product in real-time mode is plotted in Figure 15. The data needed for this plot utilized the user interface capability of exporting a collection of accumulated data to a .csv file.

Figure 15: Plotted Real-Time Mode Operation

Figure 14: Final Prototype Indicating Sensor Placement

Including the cost of the MSP430 Experimenter Board, the cost of prototyping this product is indicated in Table 1. The MSP430 Experimenter Board by itself is $149, but with bulk orders, and with the MSP430 microcontroller and nRF24AP1 transceiver integrated onto a manufacturable PCB, cost per unit of a finished product would drop significantly from the prototyping cost. The cost of the unit to a consumer would likely be within the $100 range with the indicated adjustments.

Table 1: Prototyping Component Costs

Future Improvements

    • Investigate using transistors in place of R1 as a variable resistor controlled by the DAC module

    • Connect the ADC pins to the Port 1 pins in the PCB design to access the port 1 interrupt vectors

    • Integrate a suitable MSP430 directly on the PCB to eliminate a ribbon cable

    • Order and solder new SMD switches

    • Make the device smaller by:

      • Using a lower profile battery holder or solder power leads directly to the battery

      • Using a flexible printed circuit board

      • Using SMD resistors

    • Use sensor array to counter high part-to-part repeatability error or develop better calibration techniques.

    • Further explore different operating modes in the ANT protocol

    • Utilize flash memory storage for the dynamic operation mode in order to prevent data loss or corruption from loss of power.

    • Use separate voltage regulator to supply the sensor biasing network.

Conclusion

This project was taken on by three senior undergraduate student: Justin Harvey (Cp.E), Ben Montz (E.E.) and Joseph Stribrny (E.E.). We have all been at UAF for multiple years, taken classes together, and worked well as a team. Our time together before starting this project gave us a great advantage and we all enjoyed working with each other. We invested many hours in initial project specifications and scoping to determine the most efficient means of dividing the work load. In the early stages of research and development, the majority of the work was concurrent and collaborative. Later on in the project, areas of responsibility were more precisely defined. Justin took the lead in developing the power options including the power source and voltage regulation. He also completed the implementation of the real-time mode software. Ben led the charge tackling the ANT protocol and developing the wireless software and hardware implementations, as well as developing the graphic user interface. Joe worked on developing the hardware prototypes and programming the off-line mode software. Given the initial scope of the project, and the tight deadline that had to be met, we considered the project successful.

© Copyright 2012 Justin Harvey, Ben Montz, Joseph Stribrny