February 2007 Mechanical Engineering Departments: Tech Focus, Pt. 2

This section was edited by Associate Editor Jeffrey Winters.

Technology Focus part 2:
Instrumentation and Control

One-Pixel Wonder

Advertisements for digital cameras often stress one measure: the number of pixels the device can record. The more pixels, the higher the resolution of the image captured on the camera's charge coupled device. Indeed, even small cameras embedded in such devices as cell phones and personal digital assistants now brag of recording megapixel images.

This image, captured with a conventional digital camera, contains a fraction of the original data.

Turning the logic of pixel gigantism on its head, however, is a team of researchers working at Rice University in Houston. Instead of building a new camera with hundreds of millions of pixels, they created an image with just one pixel. While their new technology may never develop into a holiday stocking stuffer, it could have important applications in imaging extreme wavelengths of light.

In standard consumer cameras, the raw images captured by the CCD chip aren't kept for very long; the size of the data files (as large as hundreds of megabytes) precludes it. Instead, as the images are saved, they are digitally compressed into standard image formats such as jpeg. In the processes, as much as 99 percent of the original information is discarded.

That process is wasteful, says Richard Baraniuk, an engineering professor at Rice. Not only that, but it becomes very expensive as well when it comes to capturing images at very long or very short wavelengths. "We've all become enamored with digital cameras," Baraniuk said, "but to build a camera that can see in the terahertz range would be prohibitively expensive."

With a new image-processing technique, this image was made using one pixel sampled repeatedly.

During discussions between Baraniuk and his colleague, Kevin Kelly, the idea of developing a new imaging process took hold. Using a recent mathematical breakthrough, a processor might be able to compile an image from one piece of data—one pixel's worth of information—sampled several thousand times in quick succession.

To make it succeed, however, would take some ingenuity. For the process to work, each time the pixel was sampled, it would have to contain information from a random slice of the larger image. To accomplish this, Kelly hit upon the idea of adapting a digital micromirror device, a chip-based array of movable reflecting surfaces that form the heart of modern laser projectors. But instead of using the mirrors to distribute light from a few beams, Kelly devised a setup so that an image projected onto the array is reflected onto a single point.

"It's like a digital laser projector run backwards," Baraniuk said. A light-sensing diode records the accumulated light level.
Repeated readings from the diode, each taken with a different selection

of reflections, are recorded. Approximately 90,000 such samples are needed to produce an image on par with a megapixel chip. And though so many readings take several minutes to complete, the advantage comes in cost. Although CCD chipsets for regular light are now cheap, exotic radiation sources such as the far infrared or ultraviolet may require detectors that are hundreds of dollars or more per pixel. A megapixel detector of that sort would be prohibitively expensive. But a one-pixel detector combined with a million-micromirror MEMS chip could be built for a fraction of the cost.

"Though we're now just proving the concept," Baraniuk said, "the hope is that this can move into light wavelengths where today's digital cameras are blind."

Sense of Touch
by Harry Hutchinson

There are all kinds of things that people have a need to measure—the pressure of a wiper blade as it travels across a windshield, for instance, or a patient's grip on the handle of a therapeutic machine. A baby diaper must fit right, so it neither leaks nor binds.

As different as the applications are, they share the problem of finding a means to capture detailed data, often on a complex surface. A company in Los Angeles, Pressure Profile Systems Inc., markets a variety of capacitive sensors to accomplish exactly that.

The sensors consist of two electrodes separated by a compressible spacer layer. As applied pressure closes the distance between the electrodes, their capacitance changes, and the difference can be converted into an electric signal that can be captured electronically. Signal conditioning electronics process the data, which can be stored and visualized.

A Lycra membrane forms a stretchable system of sensors that will fit irregular surfaces.

The spacer is a proprietary material that acts like a spring so that, when pressure is removed, the device resumes its original dimensions to achieve repeatability. The sensors remain accurate "within a few percent each time," according to the company's chief technology officer, David Ables.

The sensors can be packaged inside more or less flexible membranes, even membranes that stretch.

The company was founded in 1996 based on research at Harvard University, but its products have been available commercially for less than three years, according to the director of business development, Ohad Zeira.

The company makes two lines of sensor systems: tactile arrays that measure distributed pressure and contact sensors that measure average pressure, or force.

Among the company's products are industrial pressure sensing systems called TactArray, which can measure pressure distribution levels to 2,000 psi, or 14,000 kilopascals, and are reliable

at temperatures between -40°C and 200°C, the company says. They are available in various sizes, up to an active area of about 30 by 45 cm (12x18 in.) that can contain as many as 10,240 elements, or individual sensors.

Zeira said that cost starts at around $16,000 for a sensor with as many as 256 elements, including software and conditioning electronics. He added, though, that the company rarely sells its sensors off the shelf. Most of its sales are solutions designed for specific applications.

In a promotional video, Ables demonstrates the use of a TactArray sensing device with a copper-clad Kapton membrane to capture the pressure pattern of a brake pad on a motorcycle wheel. According to Zeira, they can be used while machinery is operating, and customers are using them in live printing operations, during live braking tests, and in engine gasket pressure studies.

A stretchable array of capacitive tactile sensors reveals the pressure pattern of a rider on a motorcycle saddle.

The company has a number of brief case studies on its Web site, at pressureprofile.com. One involves using a stretchable version of a tactile array product, using a Lycra membrane, to fit over a motorcycle saddle.

According to PPS, its most popular product is a point sensor that it calls ConTacts. The point sensor, about a millimeter thick, is designed for the same conditions as the TactArray. Repeatability is rated within 2 percent.

Another product, FingerTPS, is molded to fit over the fingertips. It comes in three sizes for fingers and a fourth that fits the palm of the hand. The company says they are less than 2 mm thick, and have a force range up to 10 pounds with a sensitivity within a tenth of a pound. They have been used, for instance, in the study of surgical tools.

Most of the products generate analog signals. A newer product, DigiTacts, is a point sensor compatible with the Philips I²C digital bus. It is available in an evaluation and development kit at $995. OEM pricing starts at $10 for high-volume orders, Zeira said. More recently, the company has developed a DigiTacts Array sensor, as well.