Characterization of LED-Based Sun Photometers for Use as GLOBE Instruments
David R. Brooks, PD/PI for Science, Haze/Aerosol Monitoring Project, Drexel University
Forrest M. Mims III, Co-PI for Science, Haze/Aerosol Monitoring Project
Tran Nguyen, Research Assistant, Drexel University
Stephen Bannasch, Director of Technology, The Concord Consortium
GLOBE Third Annual Meeting, Snowmass, Colorado, August 3-7, 1998
The use of LEDs as spectrally selective detectors of light, for use in sun photometers, has been pioneered by Forrest Mims III (Mims, 1992). These devices are inexpensive. easy to make, and extremely stable compared to instruments based on optical filters (Mims, 1998, this conference). A sun photometer using a green LED that detects light at around 525 nm, can be used to measure aerosols and makes an effective "haze meter" because its peak response is near the peak sensitivity of the human eye.
The goal of sun photometry is to measure the intensity of a monochromatic narrow beam of light transmitted directly from the sun through the atmosphere. The extent to which molecules and other atmospheric constituents remove light from this beam, through scattering or absorption, is an important characteristic of the atmosphere. It is expressed in terms of the atmosphere's optical thickness; the larger this value, the more light is removed from the beam. Molecular scattering is an essentially fixed property of the "pure" atmosphere, but scattering due to aerosols is related to pollution and other natural and anthropogenic phenomena.
Sun photometry is a critical tool for environmental monitoring. During just the past year there have been massive biomass burnings in southeast Asia, extensive smoke transport over the southwestern United States from forest fires in Mexico, and several weeks of extensive forest fire activity in Florida. In none of these situations has adequate instrumentation been available to monitor the atmosphere on an appropriate temporal and spatial scale. GLOBE students and teachers can provide a high-density network for making the measurements necessary to respond to such events and can make significant contributions to improving knowledge about the global temporal and spatial distribution of atmospheric aerosols.
The inexpensive sun photometers we are implementing for GLOBE are deceptively simple
(as are more expensive versions) and present challenges when they are considered for serious
scientific use rather than just as pedagogical tools for teaching students about the atmosphere. Is it
worth addressing these challenges? Consider cost: a commercial sun photometer can cost $25,000.
Ours can be built for less than $25. However, even a $25 instrument needs to be justified, especially
when this investment, plus training costs, is multiplied by the hundreds or thousands of instruments
that could find their way into GLOBE schools. Thus, in our project, the measuring device itself will
continue to be a topic of ongoing investigation. One essential task is to characterize the performance
of the instrument in a quantitative way, with a rigor that would not be required in a purely
pedagogical context. In this paper we will describe briefly the sun photometer circuit and summarize
our investigations for two important topics: the spectral response of LEDs used as light detectors,
and the determination of an optimum field-of-view size for handheld sun photometers.
A prototype LED-based sun photometer, the VHS-1, is described in detail on the Concord Consortium's Web site (www.concord.org/haze); an abbreviated description appeared in Scientific American in 1997 (Carlson, 1997). A slightly modified circuit consists of just three electronic components: a green LED, an op amp that will operate with a single voltage supply, and a resistor, plus a 9V battery, an on/off switch, and output jacks for connecting a digital voltmeter. A circuit diagram is shown in Figure 1. The circuit gain is controlled by the feedback resistor. Outputs of 2-4 V are easily obtained with a resistance of a few megohms. The circuit is assembled in a closed box with the end of the LED pointing toward a small circular aperture at one end of the box. In use, the instrument is pointed at the sun and the maximum voltage is recorded. It may also be necessary to record the "dark voltage" output from the circuit when the aperture is blocked.
Unlike commercial sun
photometers used in research networks
(e.g., Holben et al., 1998), LED-based
photometers are not designed for
continuous exposure to sunlight. The
output of LEDs is temperature sensitive
and this simple circuit contains no
provisions for temperature
compensation. The most common LED
packages are clear or tinted epoxy
cylinders that have a dome-shaped lens
on the end to disperse light. In order to
minimize solar heating that would result
from the sun shining onto the LED's
active element through this lens, it is necessary to flatten the end of the epoxy case or to use LEDs
encapsulated in cases with flat rather than dome-shaped ends.
The Spectral Response of LEDs
The fundamental property of LEDs is their emission color. Although emission spectra are readily available, the response spectra of LEDs used as light detectors are not. To measure spectral response, we have used an Optometrics DMC1-02 monochromator with a tungsten lamp light source. We have built a high-gain amplifier and positioned this breadboard circuit so the end of the LED being tested can be placed directly at the output slit of the monochromator.
Table 1 gives some characteristics of several LEDs, including the green Radio Shack LED
used in the VHS-1 prototype, and Figure 2 shows some results from spectral response tests.
Table 1. Characteristics of selected LEDs
|Description||Peak emission, nm|
|Radio Shack green||569 (? not specified)|
|Radio Shack red||626 (? not specified)|
|Hewlett-Packard green (HLMP-3502)||569|
|Hewlett-Packard emerald green (HLMP-D600)||558|
|Hewlett-Packard high efficiency red (HLMP-3762)||626|
For sun photometry, narrow detector bandwidths are preferred as an approximation to monochromatic light. Optical filters used in commercial instruments have bandwidths of only a few nm, but they are expensive and have a well-documented history of degrading in unpredi ctable ways. LEDs are inexpensive and extremely stable, but have wider bandwidths. Hewlett-Packard emerald green (HLMP-D600) and high-efficiency red (HLMP-3762) LEDs have half-maximum widths of about 75 nm and 30 nm. Note the asymmetrical shape of some of the response curves, and for the emerald green LED, note the downward shift, from 558 nm to about 525 nm, from peak emission frequency to peak detection frequency. The relationship between an LED's detection spectrum and its emission spectrum is not understood well enough to predict one from the other, and one goal of our ongoing investigations is to gain a better understanding of this relationship.
A wide bandwidth affects most directly the calculation for the contribution of molecular
(Rayleigh) scattering to the total optical thickness. Assuming that the total optical thickness a
consists of contributions soley from aerosols and molecular scattering, ,
where p/po is an atmospheric pressure correction and mo is the sea-level relative air mass. The
Rayleigh coefficient R can be calculated theoretically using computerized models of the atmosphere.
Over the half-width of the emerald green LED, from about 475 nm to 550 nm (e.g., Bucholtz, 1995),
the Rayleigh coefficient varies by a significant amount, from about 0.18 to 0.11. We are currently
investigating how best to represent the effects of molecular scattering for a sun photometer with a
relatively wide spectral response.
Selecting a Field of View
As the purpose of a sun photometer is to measure the transmission of direct sunlight, instruments should have the smallest practical field of view (FOV) in order to minimize the contribution of scattered light. What we seek is the relationship between the angle subtended by an instrument's aperture, as viewed from the detector, and the measured FOV. One approach is to hold the instrument fixed and move a light source through the optical centerline of the instrument. The angle at which the measured signal falls to half its maximum value on either side is called the full-width-half-maximum (FWHM) FOV. Commercial sun photometers, which include active electronics to track and center the sun, have FWHM FOVs of as little as 1. For handheld devices, the practical tradeoff involves the ease and accuracy with which an instrument can be pointed at the sun; easier and more reliable pointing requires a wider field of view.
In order to measure FOV for a handheld sun photometer, an instrument mounted on a tripod is pointed at the sun on a clear day, adjusted to obtain the maximum voltage, and then locked in place. Voltage and time are subsequently recorded as the sun moves, until the signal drops to near 0. Then, solar position is calculated for all the recorded times, from which the solar angular position relative to the original optical center line can be found. This process of creating a "half-scan" takes a few minutes, as the sun moves its diameter (0.5) in approximately 2 minutes. An alternative, but more tedious approach is to make "full-scan" measurements that involve fixing the instrument initially at a point where the sun will cross a few minutes later.
We have fitted two instruments with adjustable apertures. Distances from the aperture to the LEDs are 89.5 and 88 mm. Scan results for these instruments are shown in Figures 3 and 4.
The LED in the full-scan instrument is mounted in a clear epoxy case with no internal reflector. The other LED has a more typical 5-mm-diameter tinted case with an internal reflector. The FOVs for the apertures shown are roughly equal to the geometrical FOVs (the arctangent of the aperture diameter divided by the aperture-LED distance), as shown on the figures. The second configuration gives the expected FWHM FOVs based on the physical aperture, but the larger effective detector area produces a less desirable bell-shaped response curve. Several studies (e.g., Shaw et al., 1973) have shown that the contribution of scattered radiation to the total signal is small (< 1-2%) in most atmospheric conditions for a FOV less than about 5. Our experience shows that an instrument with a FOV as small as 2 can be pointed reliably by hand. Thus an acceptably small FOV is readily obtained with these instruments.
Because the sun is not a point source (it subtends an angle of about 0.5), it is not the best
choice for determining an instrument's field of view. However, these measurements provide other
useful information about the performance of sun photometers. Consider the several minutes during
which the sun was completely within the FOV during the half-angle scans. In this case the output
voltage is essentially constant. The voltage "noise" from atmospheric fluctuations during this part of
the half-angle scan has a standard deviation of about 0.3%, so that essentially all measurements are
within 1% of the mean. Note that this calculation includes whatever effects might be associated with
heating of the LED or the rest of the instrument electronics. (The op amp used in this circuit has a
very low temperature coefficient compared to the LED itself.) Thus even though the temperature
sensitivity of semiconductors is a legitimate concern when using LEDs in this way, these results
demonstrate that heating of an LED from exposure to the sun does not present a significant problem.
With this paper, we have brought directly to the GLOBE community some results from our ongoing investigation of the performance of inexpensive LED-based sun photometers. Our experience shows that when students use electronic instruments, especially when those instruments have digital outputs, they tend to believe they are "perfect." It can be disconcerting to discover that this is an unwarranted assumption. However, imperfect measurements are an integral part of "real" science, and it is important to communicate the fact that we can and must deal with measurement constraints in all data collection programs. Our sun photometer involves tradeoffs in design and application, as do its more expensive counterparts. Based on our initial results, we believe we have demonstrated that:
1. even though LEDs are optimized for other uses, we can select an appropriate LED for use as a sunlight detector;
2. an LED-based sun photometer can be built with a field of view small enough to meet scientific
standards and, at the same time, be usable by GLOBE students.
Bucholtz, Anthony: Rayleigh-scattering calculations for the terrestrial atmosphere. Applied Optics 34:15, 2765-2773, 1995.
Carlson, S.: The Amateur Scientist, "When Hazy Skies Are Rising." Scientific American, May 1997.
Holben, B. N., T. F. Eck, I. Slutsker, T. Tanre, J. P. Buis, A. Setzer, E. Vermote, J. A. Reagan, Y. J. Kaufman, T. Nakajima, F. Lavenu, I. Jankowiak, A. Smirnov: AERONET - A Federated Instrument Network and Data Archive for Aerosol Characterization. http://spamer.gsfc.nasa.gov/valdesaire/, 1998.
Mims, Forrest M. III: Sun photometer with light-emitting diodes as spectrally selective detectors. Applied Optics, 31:33, 6965, 1992.
Shaw, G. E., J. A. Reagan, B. M. Herman: Investigation of atmospheric extinction using direct solar radiation measurements made with a multiple wavelength radiometer. Journal of Applied Meteorology 11, 374-380, 1973.