How do you measure something you can't touch? Physicist Erwin Schrodinger famously proved that merely observing an object (say, a cat) can change its state of being, at least in quantum physics.

Alcon Laboratories Inc. (Fort Worth, TX) recently faced this problem as they tried to regulate the vacuum level in their bottles of saline, used to flush eyeballs during surgery; they couldn't insert a sensor in the bottles without losing the vacuum. So Alcon turned to the senior design course in the engineering school at Texas Christian University, also located in Fort Worth.

The undergraduates' solution to this tricky problem was good-so good that it won Design News' second annual ANSYS College Design Engineering Award.

The class began by studying Alcon's existing methods. The BSS (buffered saline solution) is packed in 250 or 500 ml glass bottles, capped with rubber stoppers and aluminum caps, and sent along a conveyor belt at a rate of about 60 per minute. In the factory, employees would pick up about four bottles an hour, shake them, and strike them with a rubber mallet-producing a "pop" sound if the vacuum was at least 14 inches Hg, or a "thud" sound if the vacuum was too weak. If the employees tried to sample a higher percentage of bottles, they quickly ran into repetitive stress injuries.

So Alcon's specification for the TCU job demanded that the new, automated method test at least 65 bottles per minute, rejecting 100% of low-vacuum bottles, but rejecting no more than 0.1% of high-vacuum (satisfactory) bottles.

The student team first sampled a laser method of checking vacuum strength, but abandoned it in testing when they couldn't reach the required accuracy. From touring Alcon's production line, the students knew that knocking the bottles worked, so they tried to find out why. They analyzed the acoustics of bottle-whacking, trying to replicate (and regulate) the power of the human ear.

And they immediately found the problem was more complex than they'd imagined. At least seven variables affect the sound of a struck bottle:

Next, the team divided itself into three sections. The mechanical group designed, drew, and built the prototype and final product. The electrical group programmed the PLC and the analysis software, and wired the machine, sensors, and hardware. And the testing group experimented on the prototype and final machine to ensure it met the specs.

The students were: Matt Baldwin, Chad Bayer, Brian Best, Ryan Coles, Brian Erickson, Fabio Goncalves, Damon Hickman, Ryan Keeler, Robert Kinney, Sophie Penninck, Matthew Rodgers, Russ Sanders, Absera Stephenson, and Marie Stephens White.

They soon found that teamwork was another major challenge.

"One of the hardest things was to work with people you've gone to school with for four years, and now you're in a management role," says Erickson, who is about to begin a commission in the U.S. Air Force. "It's about learning how to work together; you learn a lot about each other."

They struggled through the team challenges and submitted the project on time.

Their solution works like this: A photoeye counts each bottle moving along the assembly line, and allows four to enter a movable frame. Scalloped clamps seize the four bottles with a linear thruster, holding them 0.25 inches apart, within just 0.05 inches of the clamp's center. Then the frame rotates to wet the stoppers, using a 40 to 45-degree turn actuator. As the frame returns to vertical, four pneumatic pistons with stainless steel ends hit the bottles sequentially, so a single microphone can distinguish four separate sounds. All four hammers and the turn actuator use Hall effect sensors to confirm their actions. Then LabView software, from National Instruments, analyzes the acoustic signals-a pneumatic cylinder ejects the bad bottles, while good bottles continue down the conveyor.

That entire cycle takes 3.5 seconds. And it works-this machine achieved a line speed of 56 bottles per minute (short of the specified 65, but greater than the human line's 48). It tests 100% of the bottles, as opposed to much lower percentage by human workers. And indeed, it passed 0% of low-vacuum bottles, although it false-failed 17% of bottles with 21 inches Hg.

"I think they did a great job," says Tim Habinak, a senior engineer at Alcon. "They really grasped the concept of what the problem entailed. I've actually been working on this particular problem for a couple of years, and came up with a couple solutions, but none of them worked as well."

Habinak was particularly impressed with the student team's structure, and how well they communicated with their Alcon contacts. In fact, the senior engineering course is designed to create precisely this type of real-world experience for the students.

"On the first day of the senior design course, I lay a spec on their desks," says Patrick Walter, who teaches the class. The students study the specification, then divide themselves into administrative and technical roles.

"We [professors] can consult, but if they don't ask, I don't tell them-it's just like the real world," he says. "There's the possibility of failure, but no class has not been successful."

TCU's engineering school was founded in 1992, and Walter arrived in 1995, after a 30-year career designing ballistic missiles at Sandia National Labs (Albuquerque, NM). His senior design classes have done projects for Endevco Corp. (1996), Bell Helicopter Textron (1997), RockBit International (1998), Bell Helicopter Textron (1999), Alcon Labs (2000), U.S. Army Waterways Experiment Station (2001), and Lockheed Martin (2002). Today there are 7,500 un-dergrads at TCU, including 130 in the engineering school.

As they began to test their ideas for the vacuum meter, the students set up a sample conveyor belt in another part of the TCU machine shop. The shop staff testify that some of the greatest challenges were not just making the system, but also the team work.

"There was the usual amount of disagreement, but you can't put that many engineers in one cage and not have that," says David Yale. "They find out it's tough to let some of their ideas go."

"They start out thinking it's a research project, and they find out it's a problem-solving project," agrees machinist Mike Murdock. "They have to solve the challenges of the project, and they have to solve the challenges of the group, working with each other."

The students found that another great challenge was the project's flexibility. "We had three, four, five, up to eight ways of solving some problems, and it was really tough to decide which to choose," says Goncalves, who is now employed at Corning Cable Systems.

And when the team first pitched a budget of $120,000 to Alcon, the company talked them down to just $23,000, says Kinney, whose administrative role on the team was budget manager. He's now an engineer at Lockheed Martin Aeronautics.

But in the end, the team found its focus by homing in on the single goal: "Grades were the last thing on our mind," says Sanders, also a Lockheed Martin employee. "It doesn't matter if one person gets an A and one gets a C. If you don't get the project done, you've all failed."