The device described in this article is intended to satisfy an industrial need for continuous position sensing over a measurement range varying from centimeters to many meters. As a fiber-optic sensor, it has the well-known advantages of immunity to EMI and an inability to create sparks in a potentially explosive environment. Furthermore, it is noncontact. Although a laboratory proof-of-principle has been accomplished, this patent-pending sensor would have to be engineered to satisfy a particular application. Sandia National Laboratories welcomes collaboration with an industrial partner to achieve that end.
Principle of Operation
In its simplest form the sensor (see Figure 1) consists of an optical fiber that is uniformly doped with fluorescers and a small light source that excites or "pumps" the fiber, thereby inducing fluorescence. It is assumed that the pump emits light in the vicinity of wavelength λ1 and that the fluorescence spectrum is significant in the vicinity of λ2, where λ2 > λ1. The pump source, which could be the end of another optical fiber, would be attached to a moving object that travels along the fluorescing fiber, although the fiber could also be in motion. It is the relative position between the pump source and the fiber that this sensor detects. How it does so is suggested by the lower right-hand corner of the figure, in which pump light impinges on a fluorescer within the fiber. Some of this optical power passes through the fiber without interaction, while some is absorbed by the fluorescer and is reemitted at the longer wavelength. A fraction of this radiation near λ2 travels at too large an angle to the axis of the fiber to be guided by it, but the rest of the radiation is guided to either end. Were it not for absorption within the fiber the optical power emerging at the two ends would be equal because of the symmetry of fluorescence emission, regardless of the position x of the pump source. However, absorption naturally exists or can be designed into the fiber, thereby producing the desired position sensitivity. Thus, fluorescence followed by absorption is at the heart of this sensor. High absorption implies high spatial resolution and small range; low absorption implies low spatial resolution and high range.
Figure 1. A mobile source of light can create in a doped optical fiber fluorescence radiation that initially travels with equal strength toward both ends from the point of generation. As a result of partial absorption in transit, the signals detected at either end will be different and their ratio, S1/S2, yields the location of the source.
The basic equations governing the position sensitivity are shown in the lower left-hand corner of Figure 1, where a single extinction coefficient is assumed to be characteristic of the emission spectrum. We note that the logarithm of the ratio of the two signals S1 and S2 is linear in x and independent of the strength of the pump source. Thus, variations in source strength have no effect on sensor accuracy. In addition, variations in separation between the pump source and the fluorescent fiber have no effect on the ratio if the pump light is collimated, or (as can be shown) if it produces a distribution of power along the fiber that is symmetric at any separation. This latter limitation is minor, since most common optical sources produce symmetric distributions of power.
Although insensitivity to pump strength or coupling of pump light to the fluorescent fiber is a distinct advantage of this sensor, signal-to-noise problems will arise if the individual signals S1 and S2 are too low. Consequently, to bolster signal strength one would design the sensor to produce as many passes of pump light through the fiber as possible with, for example, a reflector. An optimally shaped bundle of fluorescent fibers whose individual signals are added will also increase signal strength.
Figure 2. A laboratory test of the long-line fiber-optic position sensor has been conducted. The mobile light source is the end of an optical fiber that is illuminated by an LED.
It may be possible to produce a similar sensor without fluorescent fibers, but with so-called "side-emitting" fiber bundles that are manufactured for illumination. The clear fibers that make up the bundle are woven together to form a kind of rope. As a result, the individual fibers contain numerous bends that cause light to be scattered out of their core and into space. If the fibers are side-emitting, one would expect them to be "side-receiving." That is, light impinging on them from the side should be scattered into their core and transmitted to either end of the bundle, in contrast to the case with straight, clear fibers. Position sensitivity is achieved because some of this "in-scattered" light will be "out-scattered" before it reaches either end, in a manner analogous to absorption in the fluorescent fiber. The potential advantage of using these bundles is that the wavelength of the source is largely irrelevant. They may, however, be cumbersome for certain applications and perhaps expensive as well. We have not experimented with them.
Description of the Experiment. Figure 2 is a schematic of a laboratory test of this position sensor using off-the-shelf plastic fluorescent fibers from Saint Gobain (www.saint-gobain.com). Their diameter was 1 mm. The pump source was a blue LED from Nichia (www.nichia.com) that emitted in the neighborhood of 470 nm (or, in one case, a green LED emitting around 520 nm) that was coupled to the fluorescent fiber through a clear fiber. This fiber was attached to a support that was manually positioned at various places along the fluorescent fiber, which, in turn, was placed on an aluminum reflecting surface to produce an additional pass of pump light through it. At each end was a photodetector consisting of a back-biased PIN diode and a series resistor, whose voltage was proportional to the optical power falling on the PIN diode. In one set of experiments, the pump light was steady and the fluorescent fiber was covered, except at the point of excitation, to shield it from ambient lighting. In another, the entire fiber was exposed to ambient lighting, but the pump light was modulated at ~11 kHz. The electrical output of either detector was filtered at that frequency, which was far removed from the frequencies of any ambient lighting. Modulation and filtering is the obvious way to overcome any influence of background lighting. The technique also provides a means of monitoring the position of several objects simultaneously using a single fluorescent fiber (or bundle) and a single pair of detectors: The pump light from each object would be modulated at a different frequency, and the multifrequency electrical signal emerging from either detector would be filtered at these frequencies. Mechanical interference among the objects would obviously have to be avoided.
Figure 3. The signal ratio vs. position of the light source for an amber fluorescent fiber that has been excited by steady light from a blue LED exhibits exponential behavior consistent with simple theory. Linear behavior is obtained through its logarithm. This fiber was physically shielded from ambient lighting.
Presentation of Data. Figure 3 shows the positional response of an amber fluorescent fiber to steady pumping from the blue LED. We note the excellent fit of the data to an exponential rise, as indicated by the continuous curve described by the equation. Equating the multiplicative factor, 0.0557 to exp(–AL) in Figure 1, we arrive at an alternative value for 2A of 0.03815, which is close to 0.0398 derived from the fit. Differences in detector gain and coupling of the fiber ends to the two detectors could explain this small discrepancy. Unless these are minimized, the values of k in Figure 1 would have to be different, but the basic behavior of the sensor is unchanged. Despite the scatter in the individual signals shown in Figure 4, their ratio is nearly free of it. The scatter arose because the relatively unsophisticated nature of the apparatus in Figure 1 did not allow for reproducible coupling between the pump and sensor fibers, but it does demonstrate the point made earlier.
Figure 4. The individual signals from the amber fluorescent fiber contain scatter that the ratio does not. When used ratiometrically, the sensor is therefore insensitive to variations in the strength of the source and its coupling efficiency to the fluorescent fiber.
Figure 5 is the same as Figure 3, except that the modulation and filtering technique was used on an exposed fiber. The results are similar, although the multiplicative factor deviates by ~13% from that in Figure 3. It is probable that the discrepancy in AC characteristics of the two detectors could explain this, since no attempt was made to match (or optimize) them. This fiber was also pumped steadily with the green LED, although its fluorescence efficiency is noticeably lower when excited by this spectral region than by the blue. Despite the smaller individual signals, the signal ratio in Figure 6 is essentially the same as that in Figure 3.
Figure 5. The signal ratio vs. position of the light source for an amber fluorescent fiber that has been excited by modulated light from a blue LED exhibits similar exponential behavior. Modulation allows the effects of ambient lighting to be filtered out without physical shielding.
A red fluorescent fiber was also studied using modulated pump light from the blue LED. Figure 7 displays its exponential positional response, though with a different absorption constant from the amber fiber. Figure 8 shows the scatter in the individual signals, not present in the ratio. Similar measurements were performed on a green fluorescent fiber using modulated light from a blue LED. Its signal ratio was exponential also, but with a much lower absorption coefficient of 0.0071/cm (as opposed to 0.0185/cm and 0.0407/cm).
Figure 6. The signal ratio for an amber fluorescent fiber excited by a modulated green LED is also exponential in nature. This behavior suggests that the optical spectrum of the fluorescence radiation is relatively insensitive to the spectrum of the source.
We have presented a laboratory proof-of-principle of a fluorescent long-line fiber-optic position sensor, suitable for measurement ranges varying from centimeters to many meters and for which extremely high resolution is not needed. We have also described a method by which the position of several objects can be measured simultaneously. Not discussed are the configurations that can be used to perform measurements, in addition to linear position. These include angle measurements, two dimensional position measurements with arrays of fluorescent fibers, and simultaneous high- and low-resolution measurements by combining fibers with high and low absorption.
Figure 7. The signal ratio for a red fluorescent fiber using a modulated blue LED is also exponential, as predicted by theory.
Related work has been reported in "Fluorescent Fiber Sensor," by Tanaka et al., Japanese patent # JP2201203. Apparently, however, no data were obtained by the authors.
This article was written by Sandia National Laboratories under Contract No. DE-AC04-94AL85000 with the U.S. Department of Energy. The U.S. government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for U.S. government purposes.
Figure 8. The individual signals from the red fluorescent fiber exhibit scatter that the ratio does not, as in the case of the amber fiber.