About:

– Polar Climate Weather Station (PCWS), Madison College
– Automatic Weather Station (AWS), University of Wisconsin-Madison

How did I become involved in PCWS?

As an undergrad in my 2-year EET program at Madison College (formerly Madison Area Technical College), I worked in the Electronics Department as Student Help. I organized labs for professors, participated in department events such as Outreach, and inventoried electrical components as I progressed through school. Eventually a few friends and I re-started the college’s Robotics Club and participated in the WITC Robotics Competition. Honestly, we got crushed as our robotics steering system couldn’t drive straight, the drive motors were way too powerful, and we ran into a problem with Arduino’s IDE overlapping timers which caused issues compiling the code operating the robot. Even so, the experience I gained from programming the robot was very valuable and a contributing factor to my interest in embedded software & hardware systems.

My work in Madison College’s “Development of a Modern Polar Climate and Weather Automated Observing System (PCWS)” grant began during my graduating semester. I was first hired on as a student hourly worker with the job of programming sensors, along with EET undergrads Forbes Filip and Ben Anderson. It was a steep learning curve as the level of programming went much lower (programming lower) than Arduino’s native language. Over the course of the first month, I learned to use Atmel Studio with an Atmel ICE interface to program & debug Atmel 328PB microcontrollers. I began to understand how analog and frequency sensor measurements are taken, as well as how basic serial communications occurred at the embedded software level. As things started to become more complex, Andy Kurth, a grant Principal Investigator and former professor of mine, was super supportive of my work and helped me get through tons of code & troubleshooting until I could program nearly independently. After making significant progress in software development, I was promoted to Laboratory Assistant casual employee for Madison College under the PCWS grant.

The Development of the PCWS

The main goal of the PCWS Grant is to create a custom embedded automatic weather observation system designed to work in extreme environments of polar regions for extended periods of time. This task required careful selection and testing of all circuit components, especially temperature compensated crystal oscillators. Other specialized components tested were the durability of plastic terminals, stability of RS232 communication ICs, and even the PCB board material itself. The Atmel-based system is designed with robust CMOS technology that through vigorous testing was found able to operate in temperatures as low as our freezer can go, around -86C, assuming a stable external clock source.

The power source of the PCWS presented another challenge – the datalogger had to operate on low-enough power to withstand the cold, dark, harsh winters of Antarctica. The system was designed with solar panels to recharge glasspack gel-cell photovoltaic deep-cycle batteries in summer, which could then theoretically store enough energy to power the device through winter. At 12V, the current revision of the datalogger consumes around 60mA at full power, and less than 30mA in idle mode. A combination of power relays, a charge controller, and software controlling the power states of the integrated circuitry on-board accomplishes this goal.

Early Testing & Board Rev 1.0

Rev 2.0

“Frankenboard” Rev 2.0, the latest revision of the PCWS Datalogger, modded

Design and testing each board revision was quite challenging. My job evolved from writing simple sensor interfacing code, to testing and designing hardware & software to meet the specifications of the WMO standard for scientifically acceptable measurements, to writing the fundamental operating system code for a continually changing hardware design. An example of development from a hardware perspective was in Rev 1.0, when I realized that some sensors would require analog signal processing in order to be correctly measured by the MCU’s timer inputs. For our sensor suites RM Young Wind Vane, which outputs a varying frequency and amplitude sine wave based on impeller speed, I designed a diode-clipping operational amplifier buffering circuit with passive filtering using breadboards and through-hole components. The purpose of this circuit was to modify the sine-wave into a TTL-level square wave. After several design overhauls, I finally created a design that worked in our -80C Freezer. This design was then successfully implemented into Rev 2.0’s PCB.

Software development was also a significant hurdle to overcome. I had to create timers and event-driven code to define when the datalogger should be taking measurements, how the three microprocessors communicate data between themselves, how data is stored in memory, how data should be formatted in a way that can be sent via Iridium Satellite Communications Systems, and an entire external-user command-line interface for field technicians to run diagnostics on. This is akin to writing a custom embedded operating system. A requirement of the code was to have the ability to run continuously for long durations – months or years – with multiple fail-safe algorithms. To accomplish this, the way memory was stored had to be carefully designed, as to not create inadvertent overflows, underflows, and avoid continuous unchecked loops. The code implements a watchdog timer system, where in certain parts of the code (such as taking measurements) execution is known to take a determinate amount of time. By setting a watchdog timer to a value slightly greater than this execution time, if the code hangs or does not update the watchdog before the timer runs out, the board will enter an interrupt, saving all of its states to EEPROM memory, resetting the operating system completely, and finally returning to its state before the hang occurred. All of this design required hundreds of hours reading Atmels’ Application Notes, troubleshooting code and hardware, and self-learning in-depth knowledge of the hardware capabilities and behaviors of the microprocessors plus board circuitry at the component-level.

On November 11th, 2019, I will be sent to McMurdo, Antarctica, to finalize the software of the PCWS Datalogger, install our dataloggers in five locations – including one named after my little sister Sarah – and aid in the maintenance & deployment of many AWS weather stations for the University of Wisconsin-Madison. I will be mostly working out of McMurdo Station to service locations near the Ross Island Ice shelf, however I could potentially spend two weeks working out of the West Antarctica Ice Shelf Divide (WAIS Divide) for access to more remote tower locations. During my 8-week mission, I could potentially visit 20 locations if everything goes as planned:

Alexander Tall Tower
Austin
Bear Island (Peninsula)
Cape Hallett
Elaine
Kathie
Kaminko-Slade
Linda
Lorne
Marble Point I/II
Margaret
Marilyn
Minna Bluff
Phoenix
Sarah
Schwerdtfeger
Thurston Island
White Island
Willie Field
Windless Bight

Raise New Tower
New Instruments / Raise
Troubleshoot
Troubleshoot
PCWS Install
Raise New Tower
Raise Tower, Replace Power System
Check-up
Check-up
Check-up
Raise New Tower
PCWS Install, Replace Top Tower Section
Fix Communications
Troubleshoot
PCWS Install
PCWS Install
Electrical Maintenance
PCWS Install
Check-up
Check-up

These are the stations on the roster for this season. Some of these locations, such as Phoenix, are easy to access. Phoenix is within driving distance to McMurdo and does not require any aerial support. Other locations, such as Cape Hallett or Margaret, are hundreds of miles away. These locations require fixed wing aircraft or helicopters to visit. Very remote locations, such as WAIS Divide, are in remote regions over 1000 miles west of McMurdo. We would be flying an LC-130 from McMurdo to WAIS, then riding snowmobiles to the tower locations.With highly variable Antarctic weather conditions, flying can be a daily coin toss. Most Antarctic weather prediction is based on satellite imagery with some weather station data integration – So even though the forecast from this information can look good enough for flying, it’s up to the pilot to make the determination to land or turn around on-site. 

I would like to thank some key people within the project, and to those who have made a significant contribution to my interest in this project and in embedded engineering:

Matthew Lazzara – PCWS Lead Grant Principal Investigator
Andy Kurth – PCWS Co-Investigator & Analog Techniques Professor
Forbes Filip – PCWS Laboratory Assistant, Hardware development
Amy Limberg-Dzekute – Lead Lab Coordinator in Physical Sciences
Lee Wellhouse – UW-Madison AWS Grant Principal Investigator
Alberto Rodriguez – PCWS Co-Principal Investigator, Embedded Programming & Microcontrollers Professor
Jacob Eapen – Networking & Interfacing Professor

Thank You!

%d bloggers like this: