- Home
- Products
- Download
- Applications
- Contact
ARIEL SPECTROMETER DIFFERENCE
WHAT IS ARIEL SPECTROMETER? Astigmatism corrected fiberoptics spectrometer using unfolded, fixed/rugged Czerny-Turner optical bench with 80 mm focal length. Detector: 2048 pixels CMOS array. USB2.0 and 1Gb LAN communication interface. All data correction/conditioning is performed in the firmware. Visible (400nm-1000nm) and UVVis(200-1000nm) wavelength ranges. More details in ArielSpectrometer webpage
The important trade-offs are in two areas: optical configuration design and data correction/calibration.
Table 1. OPTICAL DESIGN CHOICES FOR FIBEROPTICS SPECTROMETERS
Table 1 Shows typical optical design choices. Blue squares (diagonal) are choices, top header are disadvantages and left header advantages of these choices. SPEC1,2,3 (red rectangles) are spectrometers from largest manufactures and green rectangle is ARIEL spectrometer. It is clear that different engineering trade-offs were made… Let’s find out more details.
Cross Czerny design vs. Unfolded Czerny design.
Unfolded/Symmetrical Czerny design has optical paths nicely separated and light scattering greatly reduced. But at the expense of a little larger footprint as compared to x-czerny design.
Cross Czerny optical has optical paths crossed. Proximity of optical elements and optical geometry increases the light scattering effect. But mechanical design can be more compact
Fixed Optical Bench vs. Flexible Optical Bench. Flexible optical bench has adjustable angular position of the diffraction grating and angular/linear position of the mirror. This allows to configure the bench for different wavelength ranges and resolutions, resulting in cost saving and flexibility in mass production. It also descreases stability of the optical system. Typically, annual wavelength re-calibration is recommended. Using such a system in industrial environment with vibrations, etc. maybe challenging. Fixed Optical bench has all components in fixed positions, set on epoxy. It is very rugged. However, it can be used only for specified wavelength range and resolution. Diifferent bench design is needed for alternative wavelength ranges.
40mm vs. 80mm Focal Length Bench. Typical small fiberoptics spectrometer has either 40mm or 80mm FL optical bench. To achieve the same wavelength resolution, small (40mm) spectrometer needs to have a grating with 2x higher dispersion (600 g/mm vs. 300 g/mm). Higher dispersion grating causes higher angle of incidence on the mirror. This increases astigmatism and coma of the image.Coma is an off-axis optical aberration that causes image to blur out and have a tail resembling comet. Coma is inversely proportional to the F-number and increases with off-axis angle. In spectrometer coma is frequently manifests as a curvature of the line.
Astigmatism correction: cylindrical lens vs. Torroid mirror. Off-axis light reflected from the spherical mirror has different foci in 2 orthogonal planes (saggital and meridianal). As a result, a point in the entrance slit becomes a line in the imaging plane – this is astigmatism. Astigmatism can cause a loss of ~70% of light – it is just not getting to the sensitive area of the detector. Correcting astigmatism can increase light collection dramatically. Either cylindrical lense or torroid mirror can be used to correct astigmatism. Cylindrical lens is an additional element that needs to be places in front of the sensor. Torroid mirror is just replacing the spherical mirror – no additional element. As we discussed earlier, in mini spectrometers (40mm bench) astimatism is coupled with coma. This means that correcting astigmatism decreases resolution (we are focusing curved line in the spot). This is a perfect storm – astigmatism is increased but it cannot be corrected without a penalty (decrease of wavelength resolution).
Torroid mirror, actually, is doing a little bit more than correcting astigmatism – it also allows to optimize design (we do not need to worry about astigmatism anymore) and decrease coma. The result is clean, symmetrical peaks in the spectrum. More details in ArielSpectrometer webpage
Astigmatism Example (Image of 0.5 mm slit in sensor plane)
A. Ariel spectrometer astigmatism corrected image (400 g/mm grating)
B. Same optical system as (A) but using spherical mirror
C. Same optical system as (B) but using 600 g/mm grating
DATA CALIBRATION AND CORRECTION
Spectrometer calibration is frequently understood as “wavelength calibration”. This is, indeed, absolutely essentual and every spectrometer has to have it. But there is much more that needs to be done: dark signal correction, non-linearity correction, fixed-pattern noise (FPN) correction and various data processing options, like boxcar averaging.
This is especially important for active CMOS detector arrays (like Hamamatsu S11639) where each pixel behaves as an independent detector. Mass produced spectrometers, typically, do bare minimum correction – enough for the basic use.
Signal conversion path for each pixel. Active CMOS detector composed of independent pinned Si detectors(pixels) each having its own capacitor (Charge to Voltage conversion) and Source Follower. Pinned Si detectors themselves are highly linear and have the same dark noise. However, the presence of independent capacitors and source follows introduce the difference in dark noise, non-linearity and cause fixed-pattern noise. All of which can be corrected, ideally, on a pixel by pixel basis.
Signal Correction Diagram. Signal acquired from the detector needs dark signal correction, correction for non-linearity and, optionally, Fixed pattern noise correction.
DARK SIGNAL CORRECTION. Dark signal depends on the integration time and the temperature. CCD detectors, typically, have “dark” pixels that are used as a reference for dark signal. The signal of the dark pixels is simply subtracted from the measured signal of active pixels. This works well in CCD detectors because dark current is, basically, the same in all pixels. Similar option exists for the CMOS detector as well (e.g. Hamamatsu S11639-11). However, in CMOS detector this strategy is much less effective because each pixel has different dark signal and temperature/ integration time dependance is also different. The fall back option is to take a “dark” spectrum measurement before signal measurement. Both of these approaches are unacceptable for fast and precise spectroscopic measurements.
ARIEL spectrometer has dark current calibrated for different temperatures and integration times. The regression coefficients are stored in spectrometer and are loaded based on the detector temperature and integration time used. This allows fast and accurate correction of dark signal in the firmware.
Dark Signal dependance on integration time is non-linear. At integration times < 10ms dark signal changes very little, for integration time > 100ms it is, practically, linear. It also happens that, at short integration time, dark signal practically not changing with the temperature while at longer integration time it is increases exponentially.Both cases are calibrated separately to reduce the amount of calibration data. Correction is done automatically in the firmware based on measurement conditions.
Non-linearity correction. In CMOS detector nonlinearity is very low at small signal (< 50% of the range) and very high near the top of the range. Typical nonlinearity of the S11639 detector is < 0.5% at low signals and up to 20% near saturation limit. In addition, each pixel has a slightly different nonlinearity. In some mass produced spectrometers high signal amplification is used to compensate for astigmatism light losses – this further increases nonlinearity.
ARIEL spectrometer has residual non-linearity < 0.4% (99.6% linearity) with global nonlinearity correction. Per pixels non-linearity correction can give ~ 99.8% linearity.
Fixed Pattern Noise (FPN). FPN in CMOS detectors is always present due to the fact that each pixel is an independent detector with its own electronics(charge-to-voltage and source follower). In Hamamatsu S11639 detector FPN is usually ~ 2% for most pixels and up to 10% for some pixels (detector specification is, as always, quite liberal and allows up to +-10% nonuniformity). This FPN noise is partially corrected in actual reflectance or transmittance measurement due to baseline calibration. But the spectrum maybe not perfectly smooth. It is acceptable in most case. In some cases, however, higher quality of the spectrum is needed. ARIEL spectrometer has optional FPN calibration to improve the spectrum.
STAY IN THE KNOW - Sign for our monthly Newsletter.
Learn about latest development, applications and promotions
We take your privacy very seriously and never share your information with anybody. The information is kept encrypted and secure. You will only receive our monthly newsletter and you can unsubscribe at any time. If you have any questions or concerns - please contact us.
Thin film measurement system. Practically, any transluscent film can be measured quickly and reliably.
Everybody is an expert with MProbe
