Customer Handoff & Final Project Documentation
Table of Contents
Team VisionThroughout the Final Demo Phase, Team BASE has successfully produced:
- 3 prototypes of varying sizes (S, M, L), consisting of 15 total parts each (excluding electronics) and 3D printed out of Nylon X;
- An Android mobile application, using BLE 4.0 capabilities, to display data to the user;
- Final test results for all engineering requirements; and
- All documentation.
Team BASE was able to create a working prototype that measures environmental temperature, humidity and pressure as well as the users acceleration and blood oxygen (SpO2). Furthermore, an Android mobile application was developed, which required BLE 4.0 capabilities, in order to display data to the user wirelessly. Overall, nearly all customer and engineering requirements were met as will be described in more detail below.
The final BOM and all drawings can be found here.
The final prototype was 3D printed in the CIAS Fab Lab out of Nylon X Carbon Fiber, which has higher tensile strength than ABS. The final print consisted of 15 parts in total and took approximately 20 hours to complete (see Figure above). Furthermore, three sizes of the prototype were designed (S, M, L) to accomodate varying heading sizes. This also helps to satisfy the custmer requirements "adjustable".
All electronic assemblies were manually assembled using a soldering iron and xray machines available within the CEMA lab at RIT. Although off-the-shelf sensors were utilized, custom PCBs were designed for the environmental sensor (dimensions: 7.75 x 7.2mm) and PPG sensor (dimensions: 11 x 11.5mm) in order to allow for a smaller design. An off-the-shelf microcontroller was used due to time time constraints and a 250mAh, Bose-supplied was integrated into the design. The overall power consumption of the design was measured to be 3.9 mAh, which allows for a battery life of 64 hours (excluding audio components). If time had permitted, the assemblies would have been converted to flex-rigid circuits to allow for reduced assembly time and increased robustness.
FirmwareThe main program utilizes an event-queue architecture (see Figure below) to allow tests to occur as needed, as opposed to a superloop. This increases efficiency and allows sensor and Bluetooth operations to occur simultaneously.
Mobile ApplicationAn Android application, using the most recent Ice Cream sandwich API, was developed in order to display data to the user. The app consists of a navigation page, a Bluetooth connection page and a data display page that the user is able to toggle between (see Figure below). Other characteristics include:
- Sensor poll rate: once per second
- Exportable Data: User can output data to txt or csv file
- Compatible: Allow connection to third-party applications
- Secure: Service names not advertised over Bluetooth
Performance vs. Requirements
All test plan documentation can be found here. The target value versus the actual value for each engineering requirement is shown in the table below.
Overall, all general design characteristics were met, such as, weight, power consumption, battery life, polling rate, and data retention. The highest risk, and most difficult to achieve, engineering requirements were those involving sensor accuracy. While environmental humidity, pressure and acceleration accuracies were all able to be met, temperature and blood oxygen were not. More details about each of these signals can be found below.
TemperatureTemperature sensor readings are inflated by 6 degrees F due to heat transfer from the head (see Figure below); this is improved from previous measurements which had an error of 11 degrees F when the sensor was placed at the back of the head. Therefore, although still out of spec (ER6.2), the temperature measurement has been improved throughout the design process. Better manufacturing techniques and material option could further insulate the sensor and prevent heat transfer. Raw data can be found here.
Temperature testing data shows an increase in 6 degrees Fahrenheit when the prototype is placed on the head
At the beginning of MSD I, we identified the SpO2 measurement to be one the most difficult metrics to achieve. An off-the-shelf sensor and and pre-existing SpO2 algorithm were obtained from Maxim to simplify the design process. Although the sensor advertised an accuracy of +/- 2%, this was for when the sensor was measuring SpO2 form the finger. We, therefore, needed to test the accuracy of the sensor when measuring behind the ear. We had two major questions that needed answered:
1. Is the sensor measuring good reflected waveforms (red and infrared)?
The figure below demonstrates a typical PPG waveform versus the signal that we were receiving when measuring from behind the ear. Although the signal intensity was smaller when measuring on the head, the signal quality did not appear to be compromised.
Expected PPG waveforms (left). Waveforms when measuring SpO2 from the fimger (middle) and waveforms when measuring SpO2 from behind the ear (right)
2. Is the SpO2 value accurate?
Twelve trials were performed in which participants were asked to sit completely still and the amount of ambient light was minimized. The volunteer also wore a finger pulse oximeter which was used as the gold standard. In total, the measured SpO2 had a 6.9% error, which is out of specification (ER9.2). The figure below demonstrates the percent error for each of the 12 trials. Raw data can be found here.
Although the SpO2 sensor was unable to meet specifications, there were still many successes throughout the design process. These included:
- Developing a custom PCB for the PPG sensor which allowed for a smaller design;
- Demonstrating proof-of-concept of SpO2 measurement behind the ear; and
- Identifying limitations of measuring from behind the ear.
Recommendations for future work include:
- Housing should better occlude ambient light; and
- Use accelerometer data to determine movement and filter bad PPG data.
Risk Management & Problem TrackingWe did not need to update our Risk Management plan during this phase. Our problem tracking document can be found here. At this point, all major problems have been resolved.
Final Project Documentation
Design Review Materials
Plans for Wrap-up
Over the next week, Team BASE will prepare for Imagine RIT. At this event, we will have three prototypes of varying sizes (see above) for people to try on as well as the working application to display data. Furthermore, a pair of Bose SoundSport Pulse headphones will be given away. Finally, Team BASE will continue collecting user data at this event. Testing includes:
- Reliability Testing: When/how does the product break?
- User Testing: Under what conditions is the SpO2 data accurate?
- User Feedback: Is the prototype comfortable? Are people interested in the product?
Individual ContributionsTo prepare for this event, individual team members will perform the following tasks:
- Andrew: Complete building prototypes
- Terry: Finish UI of Android App
- Alex: Finish BLE services for the Android App
- Htet: Complete soldering all assemblies
- Tegan: Ensure all documentation is complete and develop test plan for Imagine RIT