OSWALD

c. 2009

The Computer Science Platform For Learning (CSPFL) is a student-developed Ultra-Mobile PC (UMPC) called the Oregon State Wireless Active Learning Device (OSWALD), made for undergraduate students to explore CS concepts hands-on. To do this, the OSWALD is powerful, flexible, and incorporates some of the latest technology available, while keeping the cost to a minimum.

The concept of the project stemmed from my success in creating small low-power Linux devices for the RDRL project.  During the development of these systems for the RDRL, I built one into an ALTOIDS tin powered by a 9v battery so I could carry it around safely and work on it wherever I found time.  This concept of carrying around a little computer where it was possible to really explore the Linux kernel and write meaningful drivers (drivers that actually interacted with hardware like ADC) became the motivation for the design of the OSWALD.

The design began as the rough rapid prototype pictured at the top right.  The idea was to build a device that would require careful thought about program size and efficiency while allowing students to explore novel input methods and user interaction.

As project lead (and sole developer of the electronics) I was responsible for implementing these ideas and having working devices for students within 9 months. To reduce costs to students, I negotiated for hardware donations from Texas Instruments totaling in excess of $30,000 and worked with Tektronics to donate manufacturing time for 300 devices at their Beaverton, OR factory.  I even managed to convince Intel to pitch in $50,000 to help pay for the developers and molds for the cases.  The devices, more powerful than the Motorola Droid (and out first!), had a final cost to students of about $160.

The work did not end at the creation of the hardware. There did wasn’t a distribution of Linux that both worked on the ARM Cortex A8 architecture and provided a reasonable graphics interface optimized for touch screens.  The Android OS was just beginning and did not provide the freedom and flexibility needed to really let students explore.  Our only option at that point was to build our own (which we called  RADIX).  My role at this point was to develop any drivers needed to get all functionality working for the devices.  This monumental task included building completely new drivers for the USB (both for device and master modes), the audio codec, the power manager, and frame buffer.

By the end of my time with the project (when I left for graduate school), I had built 300 units with injection-molded clear polycarbonate cases and functioning hardware.  An example of the device in action playing
Doom II can be seen in the embedded youtube video.

 
3D printed case prototype.

3D printed case prototype.

In all it’s clear polycarb goodness. 1 speaker and a 5-way rocker on the left; 6 buttons and a capacitive touch pad on the right. Glorious TFT resistive touch screen in the center.

In all it’s clear polycarb goodness. 1 speaker and a 5-way rocker on the left; 6 buttons and a capacitive touch pad on the right. Glorious TFT resistive touch screen in the center.

Stress testing the hardware

 
All the bits except for fasteners and button nubs. It used 4 screws total.

All the bits except for fasteners and button nubs. It used 4 screws total.

The “Core” board. Headphones, 2 USB-A, mini USB-B, HDMI and a 5 V DC jack for inputs.

The “Core” board. Headphones, 2 USB-A, mini USB-B, HDMI and a 5 V DC jack for inputs.

Loaded into the X-ray inspection tool.

Loaded into the X-ray inspection tool.

X-ray of the processor to check for good solder joints.

X-ray of the processor to check for good solder joints.

Solder paste applied!

Solder paste applied!

Ready to be trimmed!

Ready to be trimmed!