Aditya Balasubramanianan's Notebook

WEEK 04 SUMMARY
Accomplishments: Tested window tint, finished the Design Constraint Analysis. Formulated a power supply system solution; just need to get it approved by the rest of the team.
Weekly Work Total: 16 hours
Project Work Total: 40 hours

Week 05

September 19, 2011 (8 hours)>:
Blaine provided a quick walkthrough of some of the code management software he had implemented, and a quick overview of the UART, ATD and pushbutton functionality he had implemented on the micro. Blaine and Aditya began to work together to get the GPS module to work. According the documentation, the GPS was capable of a run-and-start, where all that was necessary was to connect it to Tx, Rx, Vcc, and GND and it would work. As is the natural order of things, it was nowhere near that straightforward. The GPS TX/RX lines were connected to the serial port on the lab desktops. After matching the BaudRates, very little output was seen, maybe a few characters. A variety of different terminal programs were tried, and further consulation was made with the documentation. After adding some connections deemed "necessary minimum connections" we managed to get more sensible data - it was in the NMEA string format expected, but it was still garbage.

September 20, 2011 (8 hours)>:
It was hypothesised that the UART interface was not playing nice with the desktop's serial port, so it was decided to try a USB-to-Serial cable that Aditya had left over from ECE362. Unfortunately, it yielded the same results as the desktop. What was particularly confounding about the problem was that the nature of the garbage output suggested that the baud rates between the terminal and device were mismatched, which was not the case, as all settings used adhered to those specified in the documentation. Blaine had the idea of running the Tx/Rx lines from the GPS module, into the PIC32 eval board, through the uController and into the desktop's serial port. This worked like a charm, and the GPS module was talking to the Trimble GPS studio without any issues. The whole apparatus was then shifted to a door of the EE building, with the antenna just outside the doorway, and we were able to get over 5 satellite fixes! A battery powered apparatus was then jerry-rigged together, and Blaine and Aditya took it for a walk around the Engineering Fountain, using the Trimble Studio's datalogging function to make sure the GPS was functioning correctly. The image below shows the test walk, and the path between the EE and ME buildings was chosen to see if the module would lock between medium sized buildings. The rest of the day was spent making the schematic and PCB footprint for the PIC32 QFP-100 package in Eagle. The PCB footprint proved particularly challenging because of the custom pad sizes and extremely small pin spacing. It took several attempts to get the pin spacing just right, as some of the measurements on the PIC32 datasheet weren't very clear. Preliminary work was also started on the schematic and footprint for the voltage regulator package (LM2592).

September 21, 2011 (4 hours)>:
The final packaging for the device was decided upon - see picture below.
Due to the added weight of the battery, which is much larger than originally anticipated. This 7.4V 5000mAh battery weighs almost 300g. Since it would have to be located off-helmet, it seemed logical to take the Atom board and put it together with the battery and a PCB with voltage regulating circuitry. The remainder of the time was spent gaining familiarity with AutoDesk Inventor, with the objective of making a mockup of our off-helmet enclosure. AutoDesk proved extremely difficult to install, with installs failing on both Aditya's laptop and the Lab Desktop. Installation was finally sucessful on Aditya's home desktop. Meanwhile, a second attempt at the installation was executed on Aditya's laptop, using Autodesk 2011 instead of 2012. This succeeded. Initial mockups were made on the home desktop and appear below.

September 22, 2011 (5 hours)>:
After getting some feedback from the team, it was decided to revise the backpack unit. More specifically, to allow for easier access to the I/O ports, and perhaps some form of an active cooling system, since the airflow over the backpack unit would be of a much lower volume. The revised design is pictured below.

A drawing spec. sheet was also created, which can be viewed here: Final Drawing with Measurements.

September 24, 2011 (2 hours>:
Tried getting familiar with the Windows Embedded 7 toolkit and runtimes. Had a lot of trouble getting the runtime to install on a VirtualPC. Installation failed several times, where the OS installed would hang. The installer is being left overnight to see if any progress is made after several failed attempts.

September 25, 2011 (3 hours):
Updated the block diagram (seen below) to reflect some component choices and re-formatted it for HW5. Also completed PCB footprint for LM2592 switching regulator, and got familiar with the Sparkfun Eagle Library which is quite extensive and useful - it has schematics and footprints for the majority of parts they carry!

WEEK 05 SUMMARY
Accomplishments: Worked with Blaine to get the GPS working and tested it out. Finished the microcontroller footprint and schematic creation, and created CAD models for our off-helmet backpack unit. Began fiddling with Windows Embedded 7.

Weekly Work Total: 30 hours
Project Work Total: 70 hours

Week 06

September 26, 2011 (2 hours):
Discussed plans for a custom, integrated and functional harness solution between helmet and backpack unit - decided on DB25 (D-sub 25 pin) connector since 15-pin VGA pass through is needed. Began intial schematic work on backpack unit PCB. Also made some changes to the power circuitry. Decided to go with a 3A, 5V regulator at the Li-Po battery, and a 5V->3.3V linear regulator at the microcontroller board. This is because the microcontroller and sensors draw so little power, it renders the efficiency of the switching regulator moot. Since the linear regulator is so much cheaper, it was selected as the alternative. Further research on battery options was conducted.

September 27, 2011 (5 hours):
Sketched an intial schematic for the backpack PCB, seen here.

From Senior Design 2011 - The Incredible HUD

Spent some more time finalizing the power circuitry design, and specifics on the DB-25 connector to be used. Spent some time with Marcelo going over the optical challenges and brainstorming solutions. Without the aid of custom optics or a custom MEMS solution, it seems very likely that we may not be able to acheive collimation as originally planned. This means our image will not 'float' as was originally hoped, but the combiner glass still seems promising. Worked with Nikhil to finish the preliminary schematic work - Nikhil worked on the primary PCB schematic, and Aditya worked on the PCB schematic for the backpack unit.

Finished the TCSP presentation and majority of HW05.

September 28, 2011 (5 hours):
Gave the TCSP presentation, and received some good feedback regarding our schematic (presented schematics below). The main concern voiced was our lack of battery fuel gauge and battery recharging circuitry. Initially it was thought that we would simply charge the battery externally using an off the shelf charging station, but it would be awesome if it were incorporated into the design. The need for a fuel gauge though, was not something that could be avoided. It was needed for the following reasons: 1) to ensure the battery does not fall into deep depletion, where the cells are drained so low, that it causes damage to the batteries 2) to ensure that if a low-battery situation arises, the system has some way of knowing about it, so that it can safely power down and warn the user in advance 3) to monitor the battery health status during use, so that the user maybe alerted of any battery failure or fault conditions. Having not considered this at all thus far, very little information was known to the team at this point. However, Team 4 came to the rescue, and suggested a charging IC and fuel gauge solution which they were using in their project, that might have been useful to us. Work was started on re-writing the TCSP to accomodate the latest design changes to the schematic, to incorporate the battery management circuitry. It was intially decided to use the ICs suggested by Team 4.

The ICs suggested by Team 4 were : the MCP73213 for the charging controller, and the DS2782 for the fuel gauge (communicates to micro via I2C)

September 29, 2011 (8 hours):
Before lecture, it was discussed and decided that we should move all of our circuitry to the backpack unit, leaving only the temperature sensor, rear-view camera and display tech on the helmet. This is because our charging circuitry now needs 3.3V power sources, as well as to communicate with the microcontroller, so it really doesn't make sense to have so many data and power busses going between the helmet and backpack unit when they could all be located together. This was brought up with Prof. Meyer, and agreed with the team's concerns. Spent some time with Nikhil going over the latest plans for the power circuitry. While Nikhil worked on updating the PCB library to accomadate the new ICs, some time was spent going over the component selection sections of the datasheets for the battery management ICs. Finally managed to get WinEmbedded7 installed on the dev-netbook (donated by yours' truly). Installed the 'full' version which is basically the same as 'Windows Starter 7' but supposedly without the bells'n'whistles (which is odd, considering WE7 comes with Paint and Solitaire installed). Some deep research in how to manage the GUI was also conducted, with several possibilities being considered. The main problem with developing a VB app was the concern that it would be difficult to interface the webcam with it, on top of the writer's limited VB and .NET familiarity. The main thing to keep in mind, is that interfacing with serial is not thought to be a big deal, the main constraint imposed was that the method chosen would be able to display the webcam feed as well. The easiest way to get the webcam into a managable format is the use of a webcam server. YAWcam was chosen, since it has been used before by the writer, and corrals the webcam into a nice feed that's embeddable into any webpage. Hence, it would be preferable to have a GUI solution that fits in a webpage.
1) ASP.NET website development: It seems like interfacing with the serial port directly through ASP.NET is considered a security risk by Windows and raises complications
2) Silverlight webapps: Similar to ASP.NET, but has the capability to run as a stand-alone app
3) PHP: seems like a simple, easy solution. Some of the serial interfacing might be an issue due to external libraries that need to be used.
So as you can see, it turns out that serial interfacing might be just as much of a problem as the webcam feed. After further research, an interesting solution was stumbled upon that seems quite faesible was as follows (in PHP). Instead of bothering with the serial interfacing to a PHP website, the serial data should be logged to a file, and then the website can display the data logged in the file through a simple fopen command. Genius! This eases some concerns about the GUI aspect of the project. For now, the GUI effort will be put on hold, till the battery management issues can be sorted out, and finalized, as it was discovered that the MCP73213 is packaged in a form-factor that is impossible to work with without a reflow oven. So research was restarted to find a more suitable charging controller - starting with an IC related to the MCP73213, the MCP73842.

September 30, 2011 (4 hours):
Worked with Blaine for a little bit to figure out what was going on with the accelerometer - it wasn't as straightforward as we thought/hoped. It was initially thought that it was malfunctioning, but after further investigation it was discovered that the device fell within spec for both baseline values, and mV/g values. As evidenced by the data in Blaine's notebook, it is now apparent that we will need a suitable algorithm to calculate the "G's", and it will have to be slightly different for each axis. Finally put in the order for the battery to be used (GensACE 5000mAh 7.4V Hard Case Li-Po battery), as well as a charging station for testing purposes (till the charging IC situation is sorted out). Researched battery management ICs, and discovered that heat dissipation is now a real concern in the charging process. The majority of charging circuits considered thus far required additional passive components, most significantly a P-channel power MOSFET across which the majority of the heat dissipation occurs. Alternatives were considered, but the majority of the ICs considered were similar to the MCP73842, which is likely to be our final choice.

October 2, 2011 (4 hours):
More research into Li-Po charging was conducted. The complexity, and hazards of battery charging are numerous. Li-Po batteries, if improperly charged, can explode, catch fire, and emit fumes that are extremely toxic. But, Success! An IC from Texas Instruments was found, that is a single package solution, requiring very few external components. It is a part of their PowerPad series of ICs, and the neat thing about it is that it has a thermal pad built onto the bottom of the IC, which comes into contact with the PCB for efficient heat dissipation (since the Power MOSFET is integrated into the IC here). The IC in question is the bq24005. It is a 2-cell capable (8.2V/8.4V charger - 8.2V for coke anodes, 8.4V for graphite anodes.) with upto 1.2A charging current. It also has battery status indicators, a charge timer (3,4.5 or 6 hours) and a robust charging algorithm with cell preconditioning and fast-charge phases as well. It seems like this will be the best choice for the device. Also began work on putting together a BOM (with individual passives as well) for the device to facilitate PCB and Schematic completion. The BOM can be found here. Will begin to try and follow Blaine's footsteps regarding documentation using a camera - expect more pictures from now on!

WEEK 06 SUMMARY
Accomplishments: Installed WinEmb 7 on dev-netbook, finally. Explored Atom-board GUI options. Figured out accelerometer hardware niggles with Blaine. Settled on battery management IC solutions.

Weekly Work Total: 28 hours
Project Work Total: 98 hours

Week 07

October 4th, 2011 (3 hours):
Spent majority of the afternoon compiling the BOM for the PCB. Also began double-checking schematic work for any inconsistencies/errors. It was found that several of the microcontroller power/ground pins were incorrectly drawn - the microcontroller was in essence not receiving any Vdd. Fixed the majority of those errors

October 5th, 2011(3 hours):
Combiner glass arrived! Somewhat more reflective than we had originally thought - but it works great! The originally proposed angle of projection may have to be altered to reflect the image, instead of beam-splitting it. Cleaned up the schematic further, and made some additional footprints for capacitors and inductors that were not included previously. Also rectified any device-package-schematic inconsistencies. Finished cleaning up the schematic and handed it over to Nikhil to layout and route for the Friday homework. Also, added a nifty capability (should it work) to the schematic, wherby the device can either be powered via the battery, or the battery can be bypassed and the device powered by wall-wart DC should the need arise. This is just a backup - in case the device cannot be loaded and charged at the same time, or something goes awry with the battery/charging circuit. Battery pack and charger (not charging IC - just an off the shelf lipo charger) arrived as well. Spent some time setting it up and making sure the battery charges correctly.

October 6th, 2011(2 hours):
Attempted to charge the battery pack using the wall-charger. Battery charged correctly, but it also has a mysterious 3-pin connector, and its purpose is not immediately apparent. It seems to be some sort of jumper used to indicate to the charger what cell configuration the battery is. The test charging set-up is shown below.

From Senior Design 2011 - The Incredible HUD

Experimented with combiner glass somemore. The glass may be too reflective for complete tranparency, but it proves the concept well. See Blaine's notebook for further details. Nikhil and Marcelo will meet with a PhD student in Optics on Friday to get some feedback and ideas on the proposed design. The big Mouser order with the PCB components also arrived. Unfortunately, the order had a few errors in it (through no fault of Mouser's). The most notable error was the package the battery fuel guage arrived in - a QFN package was ordered instead of a TSSOP. Unable to prototype, this put a damper on the dev efforts for the fuel gauge. After talking to Chuck, he offered to work some of his magical wizardy, and managed to mount the fuel gauge on a TSSOP board.

October 7th, 20111(2 hours):
The Atom board arrived! Unfortunately it showed up without the correct RAM configuration or any flash storage (8GB of compactflash storage was configured). As a result installation of Windows Embedded 7 was not installed, and had to be put off till later. The unit did POST correctly, and USB functionality was verified by booting Atom unit with the WinEmb7 installation CD. (Shown below)

From Senior Design 2011 - The Incredible HUD

It was also noticed that the Atom unit shipped with a power supply specced at 5V and 5A. It is a little worrisome that it has a 5A capability - hopefully the Atom box does not actually draw 5A, even momentarily as the power circuitry designed thus far can handle a maximum of 3A. A method to measure the current draw accurately will have to devised. The layout of the battery, Atom box and fans was mocked up, as shown below. This shows roughly where the components are expected to be placed in the bigger backpack unit box.

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

October 9th, 2011 (5 hours):
Ran an ERC on the Eagle Schematic so far. Discovered several errors that wouldn't have been apparent. Apparently even though Eagle joins wires, it does not merge them into nets unless the wires are connected with junction dots. Also rearranged some elements on the schematic for readability in anticipation for the design review. Updated and checked all the packages associated with the schematic symbols. Put down an initial PCB layout for Nikhil to work with for the final design review PCB.

WEEK 07 SUMMARY
Accomplishments: Experimented with combiner glass layouts. Tested the battery and off-the-shelf charger. Mocked-up backpack unit component layout. Ran error checks on the schematic.

Weekly Work Total: 15 hours
Project Work Total: 113 hours

Week 08

October 11th, 2011 (3.5 hours):
Met with the team to start on the design review presentation. Worked with Blaine for sometime to find a workable solution for the Atom-current measuring problem. The lab was scoured for appropriate connectors to use with the Atom boards somewhat arcane connectors. After some research, it turns out the Atom board using a DIN-4 pin (trapezoidal) power connector. Digikey and mouser both did not seem to stock any connectors that could be used as 'breakouts' for the current drawing measuring effort. Put in a Digi-key order with all of the parts that were incorrectly ordered in the previous Mouser ordered. It was extremely difficult to find an inductor that would support the 3A voltage regulator. Went over the schematic and PCB layout changes with Nikhil.

October 12th, 2011 (7 hours):
Began re-doing the CAD work for the backpack unit in AutoDesk, in order to better reflect the design changes made up to this point. The updated models can be seen below. The Atom box was redone with a much higher level of detail now that real measurement were available.

The final CAD drawing sheet can be seen here.

October 13th, 2011 (3 hours):
Finalized the details for the Design Review tomorrow. Printed out the handouts and a little extra for the course staff and peer reviewers. The documents that were supplied at the design review are as follows: PCB, Schematic.pdf, and CAD Drawings.

October 16th, 2011 (3 hours):
Worked on the Voltage regulator circuit. The through-hole switching regulator version was the adjustable LM2596, so some configuration was required with resistors to get the desired 5V output. After some circuit debugging, 5V output was acheived. The next objective was to power the Atom with the circuit. The Li-Po battery was used to power the switching regulator, and in turn the Atom board (the LiPo battery was used sparingly for testing, as the battery management circuitry was not in place yet). Suchessfully managed to power the Atom using the Li-Po battery and through-hole LM2596-ADJ.

From Senior Design 2011 - The Incredible HUD

WEEK 08 SUMMARY
Accomplishments: Updated CAD drawings to latest revision. Completed design review - great success. Prototyped power circuit and successfully powered Atom board on battery power.

Weekly Work Total: 16.5 hours
Project Work Total: 129.5 hours

Week 09

October 17th, 2011 (3 hours):
Noticed an issue with the power circuit, where the entire setup would power off after the Atom had been on for about 10-40seconds. After much debugging, shortening of the wires used proved effective in halting this problem.

Chuck was also gracious enough to provide a breakout for the SMD version of the LM2596-5.0 (non-adjustable) that will ultimately be used in the final PCB. This brought a whole new slew of problems however. The LM2596-5.0 produces 5V, and powers LEDs, fans, etc, but as soon as the Atom board is hooked up to it, the output drops to 2.8V, which is obviously not enough for the Atom, which then refuses to power up. As soon as the Atom is removed from the load part of the circuit, the output goes back up to 5V. This is extremely strange to the team, and consultation with the Georges may be necessary.

October 18th, 2011 (6 hours):
Discovered the problem with the switching regulator circuit! It turns out the inductor is not a true 2-pin device as the datasheet might imply. The 4 pins of the inductor were all connected to the breadboard for prototyping purposes, and this was causing some sort of feedback that was causing abnormally large ripples in the DC output. Simply snipping the 2 extraneous leads off the inductor solved the problem. The working switching regulator circuit was then used to successfully power the Atom with a successful POST.

From Senior Design 2011 - The Incredible HUD

Now that the circuit was working, a protoboard with the components was made to facilitate use and free up the breadboard for prototyping the battery charging IC. The LM2596 switching regulator protoboard is pictured below.

From Senior Design 2011 - The Incredible HUD

Work commenced on protoyping the bq24005 battery charging IC. Several attempts were made with Marcelo to get it to work. The IC put out IMMENSE amounts of heat. So much heat that the flux began to vaporize. Not good.

The IC has two status LED outputs that can be used to verify operation and current operating state. These LEDs initially flickered when the circuit was powered on, but later remained off. After this, the circuit was left on for a few minutes to 'charge' the battery. This is when the flux vaporization happened, and after that the two status LEDs remained on no matter what was done. Even more worrying - when the battery was hooked up again to test charging, it seemed as though the battery was connected to the Vcc rail, when it should be isolated. Puzzled, no further headway was made by the team. Consultation with the Georges may be necessary.

October 19th, 2011 (8 hours):

After consultation with the Georges, the conclusion was made that whatever internal circuitry was isolated the input/output lines (Vcc and Battery) must have been damaged. It was decided to try another IC and see if the same behavior was observed. The circuit for this was set up but not debugged in detail in an bid to focus on the PCB effort.

With regards to Chuck's suggestion during the design review - he suggested we control the cooling fans with the microcontroller so we aren't wasting too much power all of the time. The team worked together to quickly design, test and implement the switching circuit (using a 500Ohm resistor and 2N3704) into the final design. The fans were successfully switched on using a single GPIO pin on the microcontroller. The changes were incorporated into the schematic and PCB immediately.

From Senior Design 2011 - The Incredible HUD

A significant portion of time was spent reviewing CAM procedures for the PCB fabrication. It was finally learned that Sparkfun provides their own CAM file which makes prep for fabrication easier. A test CAM job was run, with the results being sent to FreeDFM to verify that no show-stoppers were present. There were some difficulties with correct file extensions, particularly the drill file. One job succeeded without any show-stoppers, but the lack of certainty regarding the files/file extensions uploaded rendered the results moot.

Nikhil also continued making some last minutes tweaks and changes to the PCB layout, working towards finalizing the layout.

After some more debugging, and with the help of Marcelo, a new charging IC was tried - this time we also had Chuck drill a hole on the bottom of the PCB as is recommended on the layout. After re-setting up the circuit on a breadboard, the new IC was connected. In a dramatic turn of events - nothing happened. After spending some time debugging, it was discovered that the TSSOP-breakout board the IC was soldered to was slightly curved, and as a result majority of the pins were not touching the pad. After re-soldering the pins, it worked like a charm! The circuit functioned as expected, and the status LEDs verified the functionality.

From Senior Design 2011 - The Incredible HUD

October 20th, 2011 (8 hours):

The final PCB revision was completed by Nikhil, and the CAM job was run to generate the neccesary Gerber files. The list of specific file extensions to be used, provided by Chuck, helped clear up the issues with file extensions from the previous day. The BOM and proof of parts were also completed. The final PCB can be seen below, along with the proof of parts. The final Gerber files can be found here.
 

From Drop Box

From Senior Design 2011 - The Incredible HUD

Using the charging IC, battery and switching regulator, it was attempted to load the battery while charging it as well. The 'load' used was a simple red LED, and the charging IC was connected to the battery. While charging and loading was possible, it appears that it takes the battery much longer, and results in 'trickle' charging the Li-Po pack. Due to the nature of our battery, this is not recommended, as it could reduce the long-term performance of the battery pack. The next step will be to power the Atom board for an extended period of time using the battery pack, and experimenting with connecting the charging IC at the same time.

Some time was also spent getting the Charging IC circuit onto a protoboard to allow for more robust testing, as well as to free up the breadboard.

WEEK 09 SUMMARY
Accomplishments: Sucessfully protoyped switching regulator and battery charging circuits. Completed PCB design and submitted Gerber files for fabrication. Began testing simultaneous loading and charging capabilities of the circuit.

Weekly Work Total: 25 hours
Project Work Total: 154.5 hours

Week 10

October 24th, 2011 (4 hours)

Much of the day was spent finalising the proto board for the battery charging circuit. Some time was spent debugging the circuit - for some reason the LM7809 was producing 4.5V instead of 9V. The cause of this was found to be a floating ground wire on the LM7809. Once this was rectified, and the solder joints cleaned up, the circuit was tested without the battery. The circuit successfully powered up, and the status LEDs showed correct functioniong, as shown in the pictures below.

From Senior Design 2011 - The Incredible HUD

October 25th, 2011 (6 hours)

The battery charging circuit was tested with a battery load and functionality was sucessfully verified. By watching the current draw on the benchtop power supply used to power the charging circuit, the execution of the charging algorithm could be seen.

After consulting with the resident Visual Studio and C# expert, B. Bowman, the team came to the consensus that building the GUI for the atom board should be done in VS and C#, due to its native support of the serial protocol, and ability to interface easily with USB devices.

The remainder of the evening was spent trying to get a sample program working with the serial output of the microcontroller. Blaine reprogrammed the microcontroller to output a simple series of number (0-9, looping) so that a simple C# program could be devised to mirror the output of the serial port.

A simple application was written that opens the serial port, records the bytes until a 0 is encountered, and the outputs the numbers to a listbox. After much consternation in trying to intialize the serial port, it was discovered that a mere two lines of code were needed to initialize the serial port.

// Code start
SerialPort port = new SerialPort(
"COM3", 38400, Parity.None, 8, StopBits.One);
// Code End

From Drop Box

October 26th, 2011 (6 hours)

After successfully demonstrating the serial functionality in C#, the next step was to
interface the USB webcam with it. After searching on various forums and code
project websites, the best solution found was a library called AvCapWPF, which is based on the avicap32.dll library. Unfortunately, the libary had some niggles. The video feed was displayed upside down, and there were no rotation properties that could be manipulated at a code level. Everytime attempts were made to modify the code, it would break. After playing with it for a bit, debugging and implementation was put off till tomorrow.

October 27th, 2011 (5 hours)

More time was spent trying to get the webcam libary to function correctly, to no great avail. The problem of the upside down feature was temporarily fixed by simply opening the control in expression blend, and rotating it like a normal "object".

Even though the rotational issue was fixed, problems still persisted. Attempting to import the control into a different application, modifying the function calls for the library, and even adding elements caused the application to break.

After some more attempts, the final state of the sofware at the end of the evening was a stable, right-side-up video display.

October 28th, 2011 (4 hours)

Further consultation with the resident C# expert, B. Bowman, regarding the unstable behavior of the webcam libary revealed that it was not the control itself that was unstable. The cause of glitching was the 'preview' function of the webcam feed - for some reason the preview feed hijacks the webcam feed, preventing anything else from accessing it, including code that has been compiled and debugged. Disabling the preview of the webcam resulted in much better stability. A simple app in which a click button toggled the webcam display on and off was written and compiled. This turns out to be better than the initial approach, since the webcam is always 'on', just not shown, which will allow switching between the webcam and telemetry data to be more fluid and quick.

Marcelo mocked up a quick GUI with the various telemetry elements and the webcam toggle, which can be seen below.

WEEK 10 SUMMARY
Accomplishments: Finished prototyping battery charging circuit. Sucessfully interfaced with serial port and webcam using C#.
Weekly Work Total: 25 hours
Project Work Total: 179.5 hours

Week 11

October 31st, 2011 & Nov 1st, 2011

Had to have dental work done - was not in lab.

November 2nd, 2011 (3 hours)

Using Blaine's program on the Eval board which output packets of 1s and 0s, attempts were made to get to a test GUI to 'live' update the screen with data from the serial line. Implementation of a rudimentary polling loop failed, upon which Blaine and resident C# expert B. Bowman suggested implementing threading. Blaine attempted to implement threading, which was unsuccessful. Some useful examples on serial implementation were found online. Further implementation and debugging will be carried out next week.

November 3rd, 2011 (6 hours)

With Blaine initiating the PCB soldering efforts, the battery charging circuit was the first to be tested. It exhbited some strange behavior, whereby it's charging algorithm seemed to be turning off/on in a cyclical fashion, but after letting it run through a partial charge cycle using the battery pack, all behavior normalized.

While rooting around in the inner sanctum of the senior design lab, a USB driven LCD monitor was discovered. According to Chuck, its origins are unclear, but the team was given the go-ahead to experiment with it. Initial experimentation was promising - the screen was of a more manageable form-factor than the projector and its accompanying lenses are. The screen also outputs significantly less heat than the projector, and is more visible when viewed using the beam-splitter glass.


After hacking away at the screen and removing excess packaging, as well as hardwiring a USB cable to it for convenience, it was shown that the screen is slim, light and generally better than the projector tech. While a little large for the purpose, it will do. Smaller LCD screens were looked at, but initial results were not promising.

Attempts to install Win Em 7 on the Atom motherboard were made, to no avail. The compact flash card has a 'removeable' flag set, preventing the OS installer from recognizing it as a OS install target drive. It is proving extremely difficult to switch the flag from 'removeable' to 'fixed'. The card will be taken home and attempts to switch the flag will be made on the home computer, which has a CF card reader.

November 4th, 2011 (4 hours)

After sucessfully demonstrating the 5V regulator function on the PCB, a small hiccup was discovered with the 3.3V regulator. It was providing 4.3-4.5V, and even after multiple devices were tested, no change was shown. It was ultimately discovered that 2 config resistors are needed. Because of the nature of the gain function of the regulator, the values needed were small - ~240 Ohm and ~390 Ohm. This yielded a Vout of 3.28V which was deemed good enough for government work.

 

From Senior Design 2011 - The Incredible HUD

Online research yielded a unique idea on the Win Em. 7 installation front - install Linux (which isn't as finnicky about boot/removable/fixed disk flags) and then install Windows on top of that. This will be executed tomorrow.

November 5th, 2011 (3.5 hours)

After attempting to change the CF card flags using Gparted (off a Ubuntu liveCD), an WE7 install attempt was made again, to no avail.

Several flag/partition configuration attempts were made, and WE7 did not recognize any of them.

In an effort to get SOMETHING installed on the Atom box, an Ubuntu install attempt was made, which failed due to graphics driver incompatibility. After some googling, this was discovered to be a fairly common occurence with Ubuntu 11.04 and lower spec hardware, so attempts will be made to try a different Linux distro.

As a backup plan, the 2.5" harddrive on the dev netbook may have to be harvested for the Atom motherboard.

 WEEK 11 SUMMARY
Accomplishments: Began initial PCB population with Blaine, sorted out power issues. Major revamp of display tech occured, and CF cards refuse to makeboot. 
Weekly Work Total: 16.5 hours
Project Work Total: 196 hours

Week 12

November 7th, 2011 (6 hours)

Made further attempts to install windows on to the atom box. Chuck provided a SATA cable and PC power supply which were used to power the harddrive and connect it to the atom box. The idea was that Windows 7 could be installed on the harddrive, and then a utility run to convert the compact flash to a bootable fixed disk. However, after 3 attempts, windows failed to install on the harddrive.

Work was also started on preparing the helmet for the rearview webcam. Unfortunately it was also discovered that there's over two inches of foam between the inner part of the helmet and the shell that needs to be removed. Luckily a dremel tool was purchased over the weekend which greatly sped up the process.

While attempting to install the webcam, the adjustable focus began to act up - every time the webcam was placed inside the helmet, the focus would change due to the motion. Also, a small piece of glass was discovered on the CMOS sensor, which was obscuring the image partially. Attempts to remove this piece of glass resulted in breakage of the glass, ruining the webcam image. Work was halted for the day, while a new webcam was purchased. The new webcam purchased was the Microsoft VX - 800.



A cardboard mockup of the visor was also made, and traced out onto plexiglass, which was then given to Chuck to cut out.

November 8th, 2011 (6 hours)

Continued installing the webcam, with great success.


The picture above shows the webcam niche just before the final excision and placement.

The plexiglass pieces submitted for cutting yesterday arrived, and while in the process of mounting it to the helmet, Marcelo pointed out that the screen is not likely to fit with the current plexiglass design. This was a major flaw in my design. After further thought, and consultation with Chuck, it was decided to radically overhaul the plexiglass visor design and screen mounting solution. A mockup was made using cardboard. The second mockup was designed, with many more straight edges for ease of fabrication. The final mockup, along with the combiner glass was tested, and is pictured below.


Also, Windows 7 finally installed! The reason is not immediately clear why it worked today and not yesterday, but it is suspected that the changes made in the BIOS yesterday were somehow not recognized till today.


Now that windows 7 is on the harddrive, the next step is to get windows onto the compact flash card.

The structure of the helmet was also modified to better acommodate the new visor design.

November 9th, 2011 (2 hours)

The cardboard mockup of the new helmet visor design was transferred to the plexiglass sheet and submitted to Chuck for routing.

Further attempts were made to prime the CF card for installing an OS, but attempts failed again. For now, alteast Win Emb 7 is running stable on the Atom.

November 10th, 2011 (4 hours)

With Marcelo's help, holes were drilled into the helmet for mounting the side plates. Due to the use of foam as padding for the helmet, mounting any hardware inside is proving challenging. Superglue is out of the question as it dissolves the foam and provides no grip. For now, stuffing metal standoffs into the foam, with a screw as an anchor has proved effective. For future endeavors, perhaps self-drilling drywall anchors may be more useful. The top visor was also screwed in.

November 11th, 2011 (2 hours)

Began designing and constructing a mount for the actual USB LCD screen that will go in the helmet. Soldered LEDs onto the PCB for Blaine to use as diagnostic/heartbeat indicators.

After some kerfuffle with programming the micro, Blaine successfully wrote to the microcontroller - but the UART was misbehaving. Debugging to follow.

November 13th, 2011 (5 hours)

Met with Blaine to debug the UART issues. It appears that the micro is outputting serial data correctly, but the MAX 3232 line level translator was malfunctioning. Analysis of the schematic revealed that two polarized caps were placed incorrectly which was part of the problem. Fixing this did not solve the issue, until Blaine noticed that the TX/RX of the micro were connected incorrectly to the TX/RX of the MAX3232. After some fine flywiring, great success followed!

The helmet display unit was finally completed! After a lot of plexiglass bending, hole-drilling and brace-building, the results are shown below.

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

 

Furthermore, since the UART was working, the GPS, some more headers, and a few other miscellaneous components were soldered on.

 

From Senior Design 2011 - The Incredible HUD

WEEK 12 SUMMARY
Accomplishments: Finally installed something to Atom Board. PCB population chugging along. Helmet construction (webcam, display and visor mounting) complete, weatherproofing will come later.  
Weekly Work Total: 25 hours
Project Work Total: 221 hours

Week 13

November 14th, 2011 (5 hours)

Work was done to put the finishing touches on the helmet. Save for some silicone weatherproofing and the control buttons, the helmet unit is complete.

From Senior Design 2011 - The Incredible HUD

A new issue with the window 7 embedded also surfaced - it refuses to install any 3rd party drivers. After some research, it was decided that windows 7 starter would be a much better option, and with some help from B. Bowman, an image for Win 7 Starter will be obtained.

Work was done with the GUI, where the entire Serial port interface was re-written to use thread invoking to handle the SerialPort Data Received events, on the advice of resident C# expert, B. Bowman. There were some still issues where the thread would hang after a random period of time. The exact cause of this was not known, and further investigation is planned for tomorrow.

November 15th, 2011 (4 hours)

Brought Marcelo up to speed regarding the latest HUD mounting hardware, and the new GUI code. Changes were made to the GUI to accomadate the newest HUD changes, namely the new portrait orientation of the display, as shown below. Refer to Marcelo's notebook regarding the exact details of the test code implemented.

November 16th, 2011 (3 hours)

 The GUI code continued to be buggy, quitting randomly after display an exception. It was thought that the USB->RS232 adapter being used might be affecting the serial port code in some way, so the GUI was hooked up to the lab computer which had a real COM port. The GUI seemed to function better, but it still hung after some time. It is not sure if the Serial port code itself is malfunctioning, or it is the rapid switching between webcam visible/not visible that is causing the hanging.

November 17th, 2011 (4 hours)

Met with Blaine and Marcelo to discuss the finer points of the data packets being sent between the Atom and Microcontroller. Blaine has devised a 5 character lead 'word' for each specific packet. Marcelo notes that it will be trivial to get the Atom gui to process these packets due to the extensive string parsing libraries in Visual Basic. Because it is expected to be easy to parse the packets, more time was spent on trying to figure out why the Atom GUI was so unreliable.

Also worked with Marcelo to add buttons to the helmet. The historic color combination of RGB was chosen. The buttons were also ergonomically positioned to be within easy reach when operating a motor vehicle or some other activity.

From Drop Box

A major issue was also noticed with the Windows Embedded 7 being used currently - there is virtually no driver support. The webcam, USB display and peripherals refuse to install drivers allowing them to be used properly. The only drivers that installed correctly were generic HID drivers for the keyboard and mouse.

November 18th, 2011 (4 hours)

With the help of B. Bownman, a Windows 7 Starter image was obtained and installed to alleviate the driver issues. Unfortunately, a large number of updates were required before anything useful could be done, which took a long time to complete. There was also significant effort required to get the USB screen to run - there seemed to be a lot of conflict between the native VGA drivers and the DisplayLink drivers for the USB screen. After much consternation, great success!

November 20th, 2011 (4 hours)

Met with Marcelo to test the USB screen with the webcam and GUI on the Atom Board. The AtomGUI refused to cooperate with the USB screen and webcam combination. The GUI would either crash, freeze or open with multiple instances and then crash.

The webcam would start but not display, when tried in a separate webcam debugging application. This was rather strange, because the components all worked individually, but when combined, everything went awry.

All updates, drivers and software were in place, and the board itself should have had plenty of power to handle the GUI and webcam processes. It was noticed that the webcam worked when using a VGA display instead of the USB display, so in a last ditch effort, the PicoProjector was connected to the atom board, and lo-behold, everything worked!

The webcam/HUD switching still caused it to crash after a while, but it was FAR more stable than anything seen so far. It may just be the webcam/serial rapid switching that is causing the freeze.

Worked more with Marcelo to re-re-do the display packaging. After throwing around a few ideas, it was decided to model the new setup after rear projection TVs, with the projector projecting the image through the top of the helmet onto a substrate below for image formation. Work was done on the helmet to make room for the new display method while gutting the old. Preliminary tests showed promise, though experimentation will have to be carried out to determine the final alignment and best substrate.

Tinting of the plexiglass visor pieces was also attempted, as they are looking quite scuffed and scratched up. Unfortunately, the poor condition and lack of professional tinting skills and tools made this difficult. It was decided to look for some sort of spray paint-type clear coat tint.

WEEK 13 SUMMARY
Accomplishments: Lots of major milestones this week! GUI was developed to near completion - only file logging functionality remains. While the USB screen was working, it was decided that the projector is a FAR better option, so the team returned to that as the display tech. The helmet packaging also saw significant steps as far as the buttons and display mount hardware was concerened.
Weekly Work Total: 24 hours
Project Work Total: 245 hours

Week 14

November 21, 2011 (4 hours):

Met in lab to attend the Cornell Cup Webinar. Basically a general QnA session for participants.

Worked a little bit with Marcelo to get the packet parsing going, but after he completed the bulk of the parsing, great sucess! The GUI updates smoothly, though in the lab only the temperature and accelerometer work because the GPS unit has no line of sight with the sky.

Significant thought was given to how the projector should now be mounted. It was decided to continue this effort tomorrow, after obtaining some hardware from Walmart.

November 22, 2011 (4 hours)

Worked on getting the projector mounted. Using some L-brackets bent into shape, the projector was finally mounted - this was a much more arduous task than originally anticipated, mostly because it took a lot of time getting the angle of incidence just right, as well as balancing the weight of the projector along the medial axis of the helmet. A lot of time was spent trying to mount it without destroying or gluing any part of the actual projector. Finally it was decided to use the VGA adapter the projector originally came with.

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

From Senior Design 2011 - The Incredible HUD

November 25th, 2011 (3 hours)

Obtained some fresh acrylic and tint for the visor. With the help of Beering Enterprises, the new visor pieces were fabricated and finished. With the use of an awesome tinting kit from Lowes, the visor pieces were tinted in no time, and look fairly professional.

<image placeholder>

WEEK 14 SUMMARY
Accomplishments: Attached the new projector mount and made new tinted visor pieces.
Weekly Work Total: 9 hours
Project Work Total: 254 hours

Week 15 - The Home Stretch

November 28, 2011 (3.5 hours)

Installed the newly fabricated and tinted visors. The projector mount was also better aligned with the image projection substrate (a piece of frosted plexiglass) to allow the whole image to show. Acrlyic side panels for the backpack unit were laid out on a large piece of acrylic and submitted to Chuck for fabrication.

The industrial compact flash card was ordered through Chuck as well. It is hoped that this will solve the problems of installing the OS on solid-state flash storage that plagued the team earlier.

For current development purposes, a harddrive is being used. So, as a contingency plan, a harness was fabricated for the harddrive so that it can be used in the backpack unit should the second compact flash card also fail.

<image placeholder>

November 29, 2011 (3.5 hours)

The Atom board which was in a state of disassembly for development purposes, was put back together. Working with Blaine, the most appropriate layout for the Atom, PCB, battery and fans was explored on the backpack unit base-board.

As options were explored, Chuck showed up with the cut out pieces for the rest of the backpack unit. Some time was spent aligning and drilling the appropriate holes and assembling the backpack unit.

<image placeholder>

Work was done on the helmet in an attempt to seal it and align the visor pieces with each other and the helmet. Proposed adhesive solutions were not as effective as hoped, leading the team to resort to Loctite superglue.

A major problem was also discovered with the Atom GUI. While the GUI works well enough on the lab development computer, it continally crashes on the Atom board. The reasons for this are not immediately clear, and must be investigated further.

November 30th, 2011 (4.5 hours)

The day's efforts were primarly concerned with fixing the Atom GUI. Working with Marcelo, we managed to get GUI to a slightly more stable state, but it still crashes randomly.

After tweaking it over and over, the cause of the failure is not known. It is suspected that the large amount of 3rd party code used for the webcam view is causing some sort of problem. A new GUI was created, with all 3rd party code removed, and only team-generated code.

Right out of the gate with the new code, it ran for over an hour and a half without crashing. The code still crashes randomly, but is stable for several hours before it does so. Further investigation is needed.

The GUI was left running over night to test robustness.

December (!) 1st, 2011 (4 hours)

The GUI from last night crashed, but on restart, it ran for around 3 hours.

The industrial compact flash card arrived! Before the mocked-up backpack unit was torn down, worked with Blaine to take the unit outside in a bid to test a few PSSCs. The entire system functioned well on battery power, drawing almost 2.5A at peak - good for over 2 hours of battery life. The main motivation behind taking it outside, though, was confirming the GPS unit was able to get a fix in the system. The bearing function was sucessfully demonstrated (though the bearing calculations were slightly off). A few of the GPS parameters were parsed incorrectly, which is easy enough to fix. Overall a great sucesss!

The next step was to take the Atom out and install the industrial compact flash card. Before doing this, a quick field debugging station was put together so that Blaine could take the PCB outside and tweak the GPS unit, and polish the serial output from PCB.

Windows 7 starter installed sucessfully on the compact flash card. It was also discovered that if the battery was taken out of the projector, the current draw dropped by almost 1A, effectively giving us 3.5 hours of  battery life now.

Marcelo brought in some more mounting hardware for the backpack unit - and began implementing the idea of having a hinge on the large panel on the backpack unit for easy debugging access. Some excellent green spray paint was also purchased for the helmet - working with Blaine it was determined that it would work well with the surface of our helmet.

Regarding the stability of the GUI - Blaine came up with an Autohotkey script that would detect a program crash and restart the GUI if needed.

December 2nd, 2011 (4 hours)

Objective of the day was to have the helmet completely painted. Spent some time attempting to get lettering stencils cut out for the helmet using masking tape. This endeavous proved fruitless and frustrating. It would be much easier to get mailing labels and cut those into shape. For version 2, perhaps.

Thanks to the kindness of Dave Azpell in the 402 shop upstairs, use of the fume hood and paint cabinet was acquired. The visor was painted without much fuss, and the final result are shown below.

From Senior Design 2011 - The Incredible HUD

The next step was to paint the helmet. After sufficiently masking it off to protect the lining and internal components, it was primed as shown below.

From Senior Design 2011 - The Incredible HUD

Unfortunately, the brace supporting the helmet while being primed gave way, and the helmet fell on the wet primer against the paint cabinet leaving a nasty scar. The remainder of the time was spent rectifying this snafu, leaving the final painting for tomorrow.

December 3rd, 2011 (2.5 hours)

Paintwork on the helmet resumed. The primer on the helmet had dried and cured enough to be sanded down to reduced the scarring caused by yesterday's error. Unfortunately the area can still be seen, but it is not as bas as when it first happened. The drying helmet can be seen below.

From Senior Design 2011 - The Incredible HUD

The completed helmet can be seen below as well. Unfortunately, the paint chips quite easily, so some form of clear coat paint needs to be applied.

<image placeholder>

December 4th, 2011 (6 hours)

Worked on the final team poster and senior design report. The power switch on the atom was relocated and extended to allow for placement off-board. Additionally, a silicone sealing-ring was laid out on the backpack unit to help with dust and water resistance.

Some issues with the GUI were rectified, and Marcelo completed the data logging file generation code. There was a serious glitch with the code however, but Blaine discovered it was merely a case of improper file name generation that was causing an exception to be thrown.

The backpack unit was also wired up with all the necessary power connections, and laid out again.

Touchwood, at this point, all PSSCs seem to be ready for atleast intial demonstration.

- display change is working
- telemetry data is functional
- battery management is functional
- file writes are functional
- the HUD is up and running using projector technology.

WEEK 15 SUMMARY
Accomplishments: Redid the GUI code to exclude webcam functionality which will have to implemented in some other fashion. Data logging was mostly completed. Helmet was painted. 
Weekly Work Total: 28 hours
Project Work Total: 282 hours

Week 16

December 5th, 2011 (4 hours)

Met with Blaine in lab to finish packaging the helmet and backpack unit.

From Senior Design 2011 - The Incredible HUD

Demonstrated 3/5 PSSCS (initial demo) to G. Toh. Unfortunately our battery charging IC seems to keep shutting down periodically. Reasons for this are unknown, but luckily we have a back-up wall charger unit that G. Hadley suggested we use (at the beginning of the semester - so always listen to G. Hadley)

The GUI code was also modified to output .kml files instead of .txt files. The GUI was also generating data points ~15 times a second and storing it to file. This is redundant as the GPS data only gets updated once a second, so the updating loop was fixed to generate data points only if GPS data was updated.


Some issues were also noticed with the battery bar in the GUI and how it was not updating. Attempts were made to fix it to no avail, so further investigation is needed.

December 6th, 2011 (8 hours)


Spent some more time trying to debug the GUI and determine why the battery bar was not updating. Debugging showed that the battery update conditional statement wasn't even being entered, which was odd considering PuTTy showed that all the parsed output was normal, and the other input parsed by the GUI was working normally. Finally Blaine had the idea to use the visual studio 'output' window to see what the GUI really 'saw' on the serial line. It turned out that visual studio itself was reading input from the serial line compleetely wrongly. There were random new lines, missed carriage returns, and truncation of random strings. It was initially thought there was some sort of timing issue between the microcontroller and Atom serial interface, so several 'waits' were introduced into the microcontroller code by Blaine. The issues still persisted. The next possibility was with the buffer size on the GUI side, so the buffer size was quadrupled. This seemed to improve the stability of the serial line a little bit, but issues still persisted.

Blaine finally discovered some strange function call inserted into the code upon the suggestion of resident C# guru B. Bowman. It caused Brian's code to resolve issues he was having with the serial line, but at the same time seemed to be causing problems with the team's code. Upon removal of the faulty function call, all issues were resolved.

A new problem surfaced, however. The battery charging IC began to fail intermittently - it would charge for a few minutes, and then 'die' - but it would return to life when power cycled, for a few more minutes. After much consternation, it was abandoned, and the wall charger suggested by G. Hadley was used. (Always listen to G. Hadley).

Once the battery was charged and system primed, Marcelo was outfitted with the system for testing. Testing was conducted on the Engineering Mall, and by Northwestern Avenue. The GPS signal takes anywhere from 30second-5 minutes to acheieve a lock, but other than that the system performs flawlessly. The data logged during testing can be seen below.


 

From Drop Box

From HUD Mk I photos

December 7th, 2011 (6 hours)

Great success! All PSSCs were demoed and signed off on. The quick demo video created is embedded below, and was met with success in the PSSC demo session in FRNY, after which the team was invited to present at ECE 270 and ECE 362.

Work on the documentation continued.

Additionally, the Atom boards for the cornell cup arrived! Some time was spent playing with them - and they conveniently have a Linux distro already installed!

Quick Demo Video

December 8th, 2011 (8 hours)

Most of the day was spent recording video for the demos tomorrow and the final presentation next week. We rode the bus and took a car ride with the device, logging lots of data and garnering many strange looks in the process. The device turned out the be quite reliable - it did not crash even once during use, and it lasted just under two hours of continuous use on the battery.

The latter part of the day was spent creating the demo video for the 270 and 362 presentations, incoporating the new footage and data logged. 

December 9th, 2011 (4 hours)

Seeing as all the newly logged data was forgotten in lab the previous night, I had to come in and quickly incorporate the newly logged data and finish the video before the 362 presentation. Great success was acheived. However, due to the unlicensed music choices, the video was promptly banned from Youtube. The video was then uploaded to Vimeo which is more accomodating of project videos. The final video is embedded below.

WEEK 16 SUMMARY
Accomplishments: The Incredible HUD was completed and all PSSCs were successfully demonstrated! 
Weekly Work Total: 30 hours
Project Work Total: 312 hours