Reliable Spectroradiometry

Reliable spectroradiometry takes a lifetime of work and attention to master. Unreliable spectroradiometry is much easier to accomplish.1 All it takes is missing a single critical point, like the interdependence of source time variation and detection system bandwidth or the source spectral bandwidth and instrument chromatic properties. Consider the task of measuring the spectral content of output irradiance from a solar simulator. The key elements of a typical solar simulator include a xenon high-pressure arc lamp, an elliptical reflector, and an optical integrator, which, jointly with the collimating lens, produces a uniform illumination area at the working plane of the device (see figure 1). To closely match various conditions of solar irradiance, spectral shaping of the light is usually accomplished utilizing dichroic and glass filters.2

Figure 1. A typical solar simulator incorporates a xenon high-pressure arc lamp,
an elliptical reflector, and an optical integrator.

The recipe for spectroradiometric measurements is simple enough: Couple the radiation of interest into a calibrated spectroradiometer and record the results of the measurement. In our case, the calibrated spectroradiometer consisted of a double monochromator equipped with a silicon detector. An automatic filter wheel carried sorting filters of the appropriate order necessary for measuring wide spectral ranges with a dispersive instrument.

We chose the double monochromator design, with its excellent stray light control, to accurately characterize the sharp UV edge of solar radiation. Spectral shape in this region is critical as it interacts with exponentially shaped action spectra in many applications (including UV degradation testing of materials, SPF testing of sunscreens, and erythemal skin response measurements). An off-axis parabolic reflector collected radiation from the working plane of the solar simulator and focused it, with appropriate f/#, onto the input slit of the double monochromator. The same reflector was used to couple the light from an irradiance standard (a 1-kW FEL lamp) into the monochromator to allow spectral response calibration of the measurement system. Our interest was mainly in the spectral signature of the system, not the absolute radiometric values of the irradiance; thus, distances and angular content did not concern us.

Signals were quickly collected and saved. However, being much-experienced contributors to the Journal of Irreproducible Results (at least in the preliminary attempts), making a second check is now a part of our experimental nature. Changing the relative spacing between the standard lamp, the reflector, and the monochromator did not affect the spectral shape of the readings; however, the same change when light from the solar simulator was being collected made a rather dramatic difference in the spectral signature (see figure 2). The rather monotonic behavior of the ratio of the two measurements suggested that a wavelength-dependent defocus might be at play.

ST-Vib Spectrum center
Figure 2. Spectroradiometer data for solar simulator shows signal curves collected at
two different focusing mirror positions. The ratio of the two signals should have
produced a horizontal line in a well-designed measurement setup, but here it exhibits
a fairly monotonic increase with wavelength.

A belated, but more careful, analysis of the experiment pointed to chromatic dispersion as the most likely culprit. Even though a spectrally neutral reflector was being used to couple radiation into the monochromator, its focusing performance was a function of the angular content of the collected light. The angular content was not equal for all the wavelengths due to refractive elements inside the solar simulator, specifically the optical integrator assembly and the collimating lens. The net result was unreliable spectroradiometry.

A quick changeover to an input optical design based on an integrating sphere solved the problem; thus, the Journal of Irreproducible Results lost this publication. The moral of the story is that the glib application of well-known facts (for example, reflectors don't suffer from chromatic dispersion like lenses do) can lead to useless results if the complete system in which they are used is not thoroughly analyzed.


1. Henry J. Kostkowski, "Reliable Spectroradiometry," Spectroradiometry Consulting, (1997).

2. e.g. ASTM E927-91 Standard Specification for Solar Simulation for Terrestrial Photovoltaic Testing.


Zbigniew "ZB" Drozdowicz, former chief technical officer at Thermo Oriel, Stratford, CT.

Related Products