Casandra II Spectrograph

Construction of a spectrograph by reusing/recycling old elements, as well as with 3D printing pieces - 2021

The Casandra II solar spectrometer is a development based on the reuse of the optical box of the KONIK Instruments UVIS 200 absorption spectrophotometer for HPLC (liquid chromatography, or High-Performance Liquid Chromatography). These devices are 100% obsolete, as they are very old and their operation is semi-manual. It is common for many laboratories to get rid of them (dropping them into the electronic waste bin). Sometimes we are notified of this, so that we can check if it is possible to give a discarded device a second life. In this case, the essential part we are interested in recovering is the monochromator box. The monochromator has a broad spectrum visible light or ultraviolet lamp installed at the entrance. The light thus emitted enters into the box, where the parabolic diffraction mirror (monochromator) disperse the light in its different wavelengths towards the exit slit, where the absorption window and photodetector are installed. This box is sealed, so it does not usually show any internal damage. The KONIK UVIS 200 is a device designed to measure absorption. This means that it has an entrance to the box, where the light is dispersed and, depending on the angle of the diffraction mirror, different wavelengths are directed to the exit slit. The mirror acts as monochromator to select the wavelength for which the absorption is to be measured. The output side of the selected monochromatic light therefore has the chamber with a double optical window, through which the liquid of the HPLC column, whose absorption is to be measured, passes. The absorption is measured in terms of the decrease in the light that passes through the chamber, compared to the light that arrives directly. This is done by using a bundle of optical fibers that connects the light from the monochromator with two possible exits, one that goes to the liquid chamber, and another that goes out directly. The optical fibers are split into two bundles, making a "Y" shape path, to carry the light along two orthogonal ways, which end up, as we have said, in the optical window of the liquid chamber, behind which is the absorption photodetector, and at the direct output for the reference photodetector.

To reuse this spectrometer box as a solar spectrometer, the double-window chamber part is removed, since the light that is going to enter into the monochromator box arrives directly from the sun, so that the photometer will be placed directly in either of the two outputs of the "Y" to measure absolute intensities as a function of the wavelength. In this case, no reference signal is required, and the second photodetector is no longer used. Instead, a calibration of the instrument with a flat light source should be done, in order to correctly analyze the spectra. Thus, recycling the KONIK UVIS 200 consists of disassembling everything, removing the monochromator box, and reconfiguring it as a standard spectrometer for the incoming light.

It is interesting to mention here that this monochromator box is of the Rowland type, so the only internal element is the parabolic diffraction mirror. This simplicity allows the box to be manipulated without too much fear of misaligning anything.

As a side note, some interesting electronic elements can also be recycled from the KONIK UVIS 200, such as the photodiode amplification electronics and the digital display controller, which is based on the INTERSIL ICM7216 chip. The latter is considered obsolete by the manufacturer today, but can still be used without problems in many applications. The selection of the wavelength in this KONIK device (rotation of the diffraction grating to place the desired wavelength in the position of the optical fibers that guide the light to the detectors) is done by means of a manual dial, coupled to gears with a numbering drum. This dial moves the arm (angle self-compensator) that rotates the axis on which the mirror is mounted. All this, logically, means that the conversion into an automated spectrometer necessarily involves replacing these mechanisms. In the picture of the link to this article, the spectrometer box still appears with the mechanical coupling for the movement of the arm, although it is now stripped of all the original electronics. Figure 1 shows the first version already adapted for solar spectrometry.

In diagram 1, the KONIK Instrument UVIS 200 "Rowland type" absorption spectrometer is depicted, with its different parts. Note that here, the monochromator box acts just as a wavelength selector for the light traversing the optical windows of the fluid chamber. Comparision of the light traversing this chamber, and the light going directly to the reference photodetector allows to measure absorption of light by the liquid inside the optical chamber.

Diagram 1 – Rowland type absorption spectrograph as the KONIK UVIS 200.

Figure 1 – Aspect of the first version of Cassandra II, after the KONIK IVIS 200 housing was adapted for use as a solar spectroscope. In this image, the apparatus is provisionally mounted on a tripod ball head. The motorized rotation of the diffraction mirror by means of a stepper motor is already visible in this image. The entire front element of the telescope soon became obsolete in favor of a simpler system, as will be seen later.

In the diagram 2 it is shown how the KONIK UVIS 200 monochromator has been reconfigured into a spectrofotometer for the light entering from the entrance slit. the drawing corresponds to the second version of Casandra II, based on a guider camera as fotometer (explained later).

Diagram 2 – Rowland type Solar Spectrometer built from the monochromator box of a KONIK UVIS 200. Rowland geometry has been preserved in the construction.

For the construction of the Casandra II spectrometer, the entire optical box of the KONIK UVIS 200 was cleared of all other elements. The mirror movement was motorized by a stepper motor, coupled to the shaft with a pair of gears with a reduction factor of 3:1, and a support system for optical elements was prepared at the light entrance to the box, where the entrance slit was also placed, manufactured by hand. Below it is explained how it was done.

Adaptation of the KONIK's monochromator box

At the light entrance to the KONIK spectrophotometer box, threaded holes were made at the vertices of a square drawn concentrically to the circular light entrance window, a 30 mm square designed to be compatible with ThorLabs Cage System. Thus, in these holes the threaded rods necessary for the assembly of the front entrance elements, that is, the entrance slit (figure 2) can be mounted. A 2-inch-1.25-inch eyepiece adapter, from the Baader brand, was also prepared with four holes (figure 3) at the vertices of a 30 mm square. In this way, the monochromator box is able to support standard optical elements (from the market), as well as ThorLabs Cage System 30mm construction elements, which provide all the versatility necessary to configure the spectrometer entrance.

Figure 2 – Views of the mounting of the threaded feedthroughs in the entrance of the KONIK monochromator box. In the top image, the holes from the inside can be seen, with the screws to be used as rods. In the middle image, an exterior view of the aforementioned rods screws. In the bottom image, a Cage System 30mm mounting piece with one of the slots already in place (see below).

Figure 3 – Baader eyepiece adapter, 2 inches to 1.25 inches, with holes for attaching to the threaded rods.

Set up of the entrance slit

For entrance slit preparation, 25 µm-tight plates (Figure 4) were mounted on a 1-inch aluminum disk from a disassembled focuser, and this was then attached to a ThorLabs 30 mm Cage System holder. The slit plates were prepared from two aluminum plates from the chopper of a disassembled old spectrometer, the edges of which were carefully chamfered using a Proxon microdrill on an XY table (Figure 5).

Figure 4 – Entrance slits. Above, one of the first plates assemblies, based on sharp, trapezoidal plates of anodized aluminum. These plates were recycled from a "chopper" of another spectrophotometer (not described here). Below left, Cage System 30 mm holder with one-inch window. Below right, perforated aluminum cylinder, with a slit based on a pair of hand-formed silicon plates.

Figure 5 – Chamfering process of the plates for the entrance slits shown in Figure 4.

Optical assembly for the entrance to the monocromator

The objective lens in the first version of Casandra II consisted of an achromatic doublet from old binoculars (achromatic doublet of 50 mm diameter and 250 mm focal length). The aluminum tube on which it is mounted comes from another obsolete application from which it was recovered for use here. This tube is mounted on a GSO focuser, although in the second version of Casandra II, this whole system would be much more simplified (see below). The GSO tube is assembled to the ThorLabs mounting system using the 2"-1.25" adapter, described in the previous section (figure 3). In this way, the slit is mounted against the entrance of the monochromator box, with a ThorLabs Cage System 30mm plate, and on this, using the rods described in the previous section, the 2”-1.25” adapter is mounted, with the 2” part of the adapter outward. This allows the coupling of the GSO focuser right there, since it houses the 2" tube of the modified adapter. The assembly can be seen in figures 1 and 8. The focuser allows the image of the sun to be focused directly on the entrance slit.

Mirror motorization

For the rotation of the diffraction grating, the entire original manual mechanism of the KONIK UVIS 200 was removed, and a recycled stepper motor from an old printer was attached in its place (figure 6). The motor is off-axis from the mirror, and is connected to it with a standard 3:1 belt-type gear reducer assembly like those used in consumer 3D printers. The motor was screwed directly onto holes made for this purpose in the monochromator box itself. The mirror had to be aligned a little, so that the Rowland circle would pass correctly through the output slit, which is a conduit made with a bundle of optical fibers that split into two "Y" paths, to exit to two different photodiodes. In figure 7 above you can see the light scattered by the mirror, projected onto the input to the bundle of optical fibers, before its alignment (as you can see, it is shifted downwards). In this same figure below, you can see the approximate distance spanned by the visible spectral region at the distance of the Rowland circle, which is where the parabolic diffraction mirror focuses. The distance spanned, in millimeters, divided by the width of the exit slit, will give the bandpass width of the light that goes to the photodetector, and therefore, the wavelength resolution of the spectrometer. Since the beam diameter is approximately 1 mm, the resolution for a spectrum width (distance from red to violet) of about 7 cm would mean a maximum resolution of 300 nm/70 mm = 4.2 nm. This is a clear indication that the exit slit must be reduced (discussed a little later).

Figure 6 – Mirror rotation motor set up.

Figure 7 – Above, realignment of the mirror so that the spectrum falls within the output slit (small green illuminated hole at the top of the spectral band). In this figure, the color band is shifted, and must be move upwards a bit, by adjusting the mirror tilt. Below, reference in centimeters that gives an idea of the width of the spectral band in the Rowland circle.

Photodetector's electronics

A transimpedance amplifier (TIA) was built to amplify the photodetector signal, based on OP27 with a 100 K feedback resistor, visible next to the symmetrical power supply in figure 8. This was done because recycling the original electronics of the KONIK UVIS 200 was complicated, being an integrated electronics with a programmable microprocessor. In figure 8 on the left, the spectrometer assembly can be seen in its first version, with the TIA visible at its side. Later, this TIA was replaced by a commercial chipset, connected to an ADC that digitizes the signal to give it to an Arduino board with which the electronics communicate with the computer. All the electronics was encased in a special electronics box with the power supply included.

Figure 8 – Left - Assembly for testing of the first version of Casandra II. On the left, the assembly of the KONIK UVIS 200 monochromator box, with the motorization and all the entrance optics. The output photodetector is mounted, without the double-window chamber (unnecessary here). The homemade PCB of the OP27-based TIA can be seen on the left of the set up. This TIA would be the one used in the first tests. Right - The photodetection electronics box can be seen here, once built with a commercial TIA, an ADC and an Arduino UNO board for control. All of this, in turn, would be replaced by a CMOS camera in the second version, as will be seen later.

The first tests gave a poor result, because the photodetector is mounted at the output of a bundle of optical fibers that covers too much width of the diffracted beam (almost a millimeter), so the resolution drops a lot. An output slit should be mounted, or the output fiber bundle should be limited only to the central fiber (the fibers are less than 200 microns in diameter). The latter was somewhat complicated, so a different strategy was chosen: to build a "synthetic" slit. To do this, instead of acquiring the signal through a photodetector, the signal is acquired through the image of the optical fiber bundle at its output. The image is captured by a conventional guider camera. In this way, the integration of the signal can be carried out in a limited way, by using an ROI* on the captured image. In addition to this, the input optics would later be simplified as well, since it was not really necessary to have such a large aperture or focal length. Figure 9 shows the spectrograph on the assembly bench, with the new, much smaller entrance optical tube, based on an achromatic doublet with a diameter of one inch and a focal length of 100 mm.

* ROI comes from “Region Of Interest”, which is the term generally used to denote a small box within an image on which we will perform the operations of interest.

Figure 9 – Cassandra II on the bench, preparing the new input optics. Here, tests are being carried out with various focal lengths. The final doublet selected was the 100 mm focal length.

The camera was mounted on a conventional set of 1.25” to M48 adapters, inside which the lens that focuses the output beam on the camera focal plane was placed. Figure 10 shows the finished assembly, but without the camera.

Figure 10 – Cassandra II on the bench, preparing the new input optics. Here, tests are being carried out with various focal lengths. The final doublet selected was the 100 mm focal length.

The finished assembly was mounted on the modified Celestron Ultima PEC mount (refurbished, and with Markarian Mkee's on-purpose developed software, see other section). A small tube was placed on the spectrometer to perform a "sun guiding" using the MkStarGuider self-developed software. A ZWO ASI290MM Mini was used for the image of the output optical fiber bundle in the tests, while the USB3 version of the same camera was used for the guiding tube. Figure 11 shows the assembly set up for the first definitive tests, although in this image the guiding camera is not yet mounted. The precision of the Mkee-reconditioned Celestron proved to be sufficient to perform the first tests without using guiding.

Figure 11 – Cassandra II on Celestron Ultima PEC/Mkee, ready for taking the the first spectra.

For the first spectra tests, the entire fibers were integrated, so as to verify that the result was the same as that obtained with the homemade photodetector and TIA amplifier. The result was 100% reproducible, although it suffered from the same problems with the width of the output slit.

Figure 12 shows the image captured by the camera for the integration of the central region. Figure 13 shows the spectrum obtained, integrating the whole fiber bundle. Calibration lamps and some filters have been used to ensure that the wavelength scale in Figure 13 is correct, although instrument response has not been calibrated here. With this, many of the peaks of the solar spectrum can be identified, although some are less deep than they should be, or are slightly masked by the mixing of wavelengths due to the excessive width of the fiber bundle.

Figure 12 – Image of the optical fiber bundle as seen by the integration camera, ASI290 MM Mini. As can be seen, the total width of the optical fiber bundle is large. This limits the output resolution, mixing or even making some of the peaks of the solar spectrum go unnoticed, as seen in Figure 13 below.

Figure 13 – Solar spectrum from Cassandra II test. The spectrum shows some of the known features of solar spectrum, but has many of the peaks masked or distorted because the resolution is limited by the excessive integration width at the exit slit. The acquired spectrum corresponds to the black dots and line. The peaks that appear in pink correspond to a Ne calibration lamp. In light grey appears the spectrum of a white LED, in green the bandpass of the Baader Solar Continuum filter at 540 nm, and in red the H-alpha filter at 656.3 nm.

Some technical notes

The resolution of the device is given by the bandpass over the output slit. As we have already observed, the width of the visible spectra band scattered by the diffracting mirror covers a distance of the order of 70 mm at the exit slit position. Strictly speaking, these calculations should not be made based on "distance" but on the subtended angle, but for a first approximation it is sufficient. The visible band has a spectral width of approximately 300 nm . So if the output slit is one-millimeter-wide, that millimeter represents a bandpass of approximately 4.5 nm. Therefore, with this output slit bandpass, any peak with a smaller width will be mixed (convoluted) with the surrounding signal and, eventually, could be embedded, being hence unnoticed. To have a reasonable resolution, it is necessary to reduce the output slit. Thus, an output slit of 0.2 mm would lead to a resolution of 0.85 nm. On the other hand, we must consider the step of the motor that rotates the mirror. If the mirror rotates in too large steps, we will lose many peaks, which will be not measured. For this reason, the motor works in microsteps of 6400 per revolution, with a reduction gear of 3:1, which is equivalent to 19,200 steps per revolution. The angular coverage of the visible spectral region, which we previously indicated covered 70 mm at the height of the output slit, is about 15 degrees, so our resolution for the motor steps is 800 steps / 300 nm. In other words, each step covers 0.37 nm. So the limitation in resolution remains the output slit, with the step at this point being sufficient, although it can be improved (for example, microsteps drivers of 25600 per revolution can be used, allowing a mirror advance resolution of 0.0935 nanometers). A 50-micron output slit would be adequate for this step advance. Many modifications in this regard remain pending. At the moment, the first modification that has been carried out is the use of a synthetic slit, based on the selection of one of the optical fibers as integration area on the captured image itself. This is pending further testing, but there has not yet been an opportunity due to bad weather and other priorities.

In the following diagram 3, the calculation of the resolution is shown. Accordingly to previous explanations, the width of the optical fiber bunch at the exit of the spectrometer determines the resolution in nanometers. In diagram 4, the concept of "synthetic" slit is shown, with a real image of Casandra II Spectrometer software. the exit slit is determined by the ROI used to integrate the counts in the focalized image of the fiber bunch at its exit towards the detector (in this case, a guider camera). Although the software is ready and tested, we are still waiting for good conditions to do real tests with the sun.

Diagram 3 – Calculation of the resolution when all the fibers in the exit bunch are integrated.

Diagram 4 – Real image of Casandra II Spectrometer Software with two possible options for the ROI to be integrated, constituting a "synthetic" exit slit.

PENDING IN THIS ARTICLE:

The results obtained with a ROI “sinthetic” slit are pending, as well as the information regarding the software developed for this purpose, the Casandra II ee Spectrometer. This software is based on the Markarian MkStarGuider software (described elsewhere).