Optical Sensors go where others can't tread

Baltimore--In the days of wooden sailing ships, when the rats came pouring up the hatchway, you knew the hold was flooding. We've moved a long way from rodents-as-liquid-level-sensors.

But, as we've developed new sensors, we've developed new sensor problems. The rats might not mind the EMI from the ubiquitous large motors in the modern Navy, for instance, but thermocouples and other sensors with electrical lines do. EMI can disable them, sometimes permanently. Sensors have trouble with the noise, heat, humidity, and vibrations in ships, as well as in machinery in steel mills and power plants.

Fiber-optic sensors might hold some answers. They don't corrode galvanically, they resist humidity and vibration, and, within their range, they resist heat. They'll report fire, flooding, pressure, or smoke, and keep reporting it during the crisis. Because they query with photons and not electricity, they are safer to use in explosive atmospheres. Another big advantage: fiber-optic sensors use photons in their queries, so EMI and radio-frequency interference (RFI) have almost no effect.

There are a number of types of optical sensors, and the spectrometric ones seem particularly interesting. Henry Whitesel, a research engineer at the Naval Surface Warfare Center (NSWC), explains: "The spectrometric sensors read wavelength shifts, or color shifts, if they are in the visible spectrum. Signal-processing systems may look at 'color' changes, may take ratios of the intensity at different wavelengths, or both. Either way the system generally resists electrical or optical power fluctuation, and the signal attenuation from couplings, microbends, and dirt that plague light modulation systems."

NSWC has used spectrometric ratioing equipment made by Photonetics of Wakefield, MA. Robert Blocksidge, Photonetics' sales engineer, explains the attenuation difference: "In the case of the Navy, I think we're running about 500 feet of cable. With an absolute intensity-based, single-wavelength system, you would have a problem running that length of cable. You just lose your signal."

Optrode inroads. There are a number of types of spectrometric optical sensors. A couple of interesting, effective ones simply show changes in color, while interferometers read the shifts in the intensity of wavelengths created in much the same manner as an oil slick causes iridescence.

The "optrode" of Geo-Centers Inc., Newton Centre, MA, one of a complex array of chemical sensors, is a sophisticated variation on litmus paper. Geo-Centers immobilizes indicators into polymer thin films, then coats them onto the sensor or onto a substrate, such as tiny beads. The querying light shines on the indicator, which has changed color upon contact with the chemical of interest, and that color is transmitted by a second fiber back to the signal processor. The optrode uses at least two light sources, one where the indicator shows the most change, and one where it shows the least. Again, the ratio of their intensities is not sensitive to spurious changes.

NASA has been using optrodes for more than a year to control the nutrient pH in the Controlled Environment Life Support System at Kennedy Space Center, a closed-loop plant growth system that must operate for months at a time without outside intervention for sensor recalibration or maintenance. NASA and the pharmaceutical industry are interested in ammonia optrodes to monitor the action in bioreactors in fields as diverse as waste recycling, feed production, and continuous-cell culture.

In a completely different spectral-shift technology, Luxtron, Santa Clara, CA, has developed the Fluoroptic system, a temperature sensor with a phosphor embedded in the tip. Bursts of light flash down the cable and excite the phosphor so that it glows back at longer wavelengths. The decay of the fluorescence varies precisely with temperature. The technology works in temperatures that range from -200 to 450C.

The system doesn't need calibrating, is accurate and durable, and the probe is tiny. It doesn't need any metal parts; therefore it can be used in high-voltage or RF/microwave fields, especially if they are associated with medical treatments, say Luxtron engineers. It also works on live electronic circuits.

Fiber-optic gratings, on the other hand, have no sensor heads; the fiber itself is the sensor. The gratings are embedded in the fiber at manufacture when a UV laser sidewrites tiny layers of slightly denser fiber. The distance between the dense layers determines which wavelengths are reflected or passed through. If the fiber expands thermally or is strained, the distance between the gratings changes and they affect a slightly different wavelength. Signal processing correlates strain or thermal expansion with wavelength shift. Companies are beginning to produce the gratings.

With multiplexing, a number of sensors, on different light wavelengths, can input to a single signal processor, sometimes through a single fiber. It is most useful if a wide variety of parameters can be measured, which requires that a variety of sensors must use the same operating principle.

Multiple measurements. Fabry-Perot interferometers do just that. With one operating principle they measure pressure, temperature, liquid level, smoke density, strain, index of refraction, and probably more. Not only can they be multiplexed into a single signal processor more easily than standard electrical sensors, but also more easily than other fiber-optic sensors, according to a National Institute of Science and Technology report on optical-fiber sensors.

Under the Fabry-Perot principle, light comes down the single optical fiber and strikes a partially reflecting surface. Some of the light is reflected, and some traverses an optical cavity, then is reflected. When it joins the first reflected light, the two interfere and form bright and dim interference fringes. Any change in the depth of the cavity produces a predictable change in the wavelengths at which the interference fringes occur.

For the shallow cavity sensor, with a depth on the order of 2 aem, the mechanical equivalent would be a resonant mechanical cavity. If you know the modulus and look at the resonance peaks, you can calculate the size of the cavity.

Photonetics' pressure sensors of this type have measured pressure at the bottom of flooding ship compartments for NSCW. At Virginia Commonwealth University, tiny gauges less than 1/8-inch in diameter provide information on optimal joint function when inserted into non-living pig knees, where they handle up to 3.5 MPa. Researcher Jennifer Wayne comments: "Some of the pressures on the knee are several times body weight." She likes the sensors' small size, robust range, and their absolute measurement.

The Navy wants durable pressure sensors to predict equipment life and for maintenance. Air compressors are used throughout the fleet, and they're maintenance-hungry. Navy engineer Chris Nemarich inserted a short pressure gauge into a high-temperature air compressor, using just a 1/8-inch bore. He tested it at 2,000 psi and 420F. "We can compare the pressure to the crank angle position," Nemarich says. "That's important, because if there is a leak, it occurs at a particular piston position. With this information, we can pinpoint it." The ultimate goal: to train a computer neural network to recognize the pressure/time signature of a healthy compressor, and to diagnose the problems in an unhealthy one.

The Navy also wants temperature sensors to indicate fire. When engineers tested a Photonetics-made short-cavity temperature sensor in a ship test fire on board the Ex-USS Shadwell, they combined it with one of Luxtron's Accufiber "black-body" sensors (the tip is coated to form a black body that emits infrared into the fiber). The Photonetics sensors worked well within their temperature range, but permanently stopped at 400C. The black-body probe kicked in at about 400C, and handily survived the 1,200C heat. The Navy needs the cool-down temperatures for safety. The report noted that the long-cavity Fabry-Perot sensors can be made of more heat-resistant materials and potentially could cover the entire range.

The long-cavity Fabry-Perot sensors can be made either of more heat-resistant materials, or more cheaply (but usually not both). They're on the order of 500 [lmu]m deep, which produces many interference fringes, so they generally measure only relative changes.

Fiber and Sensor Technologies, Blacksburg, VA, makes a long-cavity, extrinsic (the light travels outside of the fiber) Fabry-Perot interferometer that uses an air gap between two polished ends of optical fibers held in alignment by a glass tube. Stephen Poland, F&S' production manager, says that the signal-processing system that interrogates the sensor with a laser is fast and high resolution, but must be zeroed each time it is turned on. Another system uses a broadband light source to measure absolute cavity size and thus absolute strain conditions.

The NSWC in Philadelphia is using MTI/FISO Fabry-Perot strain sensors to test watertight ship door plates for hydrostatic, bending, and torsional loading. They provide "absolute" long-term measurements where it is wet, with no electrical ground, and they compensate thermally.

The smoke detector developed by Whitesel and John Overby of NSCW is not, strictly speaking, a Fabry-Perot sensor, but it is designed to plug into the same signal processor and use the same algorithms. Conventional optical smoke detectors react when the intensity of their light transmission is cut down by smoke. False alarms can be triggered by dirty lenses and low power. Ratios can only compensate so much.

New smoke detector. Whitesel's sensor element consists of an 850-nm LED source, which is reflected by a mirror across a gap of about two inches. When there is no smoke, the returning light has the same distribution as the incident light. But when there is smoke, the small particles tend to deflect the shorter-wavelength light, and the lightwave distribution moves toward wavelengths. The signal processor compares the ratios of the short-wavelength energy with the long-wavelength energy, so that the ratio is independent of optical power. "We've demonstrated models on the Shadwell, and the Navy is considering it," he added.

So, will spectrometric optical sensors become standard equipment?

"It's a matter of getting manufacturing and material costs down," says F&S' Poland, "and of a market maturing so it can realize its economies of scale. Many of the costs will be brought down by the telecommunications industry, because it uses the same elements as fiber-optic sensors, but in large quantity. The prices for optical sources, lasers, and LEDs have dramatically dropped over the past few months."