During the 2024 autumn meeting of the Dutch national astrophotography society Vereniging 'Werkgroep Astrofotografie', Hugo Batema presented his experiences with PySpectrometer2, a 2022 project by Les Wright that he started as PySpectrometer in 2021. Hugo subsequently published his presentation in Astro Bulletin 72.¹ A healthy dose of professional deformation led me to develop PySpectrometer3 in the months that followed, consisting of improved hardware and software, some of which had already been announced by Hugo.² Here the test bench is explained, the software has its own page.

In his article Hugo already mentioned a few improvements in hardware that came from the contact we had about this: a better light source (as calibration lamp), a collimator for parallel light, a stable filter holder with angle adjustment and a higher order spectroscope. Except for the latter (the standard Paton Hawksley Benchtop Spectroscope (PHBS) still suffices), these improvements have indeed been implemented in the current set-up.

The original version of PySpectrometer was developed to use compact fluorescent lamps. These have 22 usable spectral lines, of which five depend on the type of lamp and only four are well defined. It would be better to use a calibration lamp that is better defined and smaller in size, such as a Neon (Ne) or Neon-Argon (Ne-Ar) lamp. The disadvantage of the Neon lamp is that it only has spectral lines in the range of 585nm to 744nm. In itself fine when it comes to observations in, for example, the H-alpha region but in order to be able to test all standard astrophotography filters, we would like to be able to calibrate our spectroscope for the entire range from 400nm to 700nm. And since compact fluorescent lamps are no longer produced, it is a good idea to look for an alternative.

A suitable alternative is the Neon-Argon lamp. These are available from spectroscope manufacturer Shelyak (model SE0148). In addition to their use in spectroscopy, these lamps are also used in starters for fluorescent tubes. Although fluorescent tubes have been banned in the EU since August 25, 2023, the existing stock may still be used up, and so starters will still be available for the time being.

The following starters have already been tested by the Swiss amateur astronomer Richard Walker: the Relco SC480 (contains noble gases Ne, Ar and He), the Osram ST111 (Ar and H) and the Philips S10 (Ne and Xe).³ The first is an Italian product that is difficult to obtain here, the Osram has a large 'gap' between 500nm and 660nm and is therefore less interesting, but the Philips S10 has a large number of spectral lines and is widely available here. The larger flux of the Philips S10 compared to the Shelyak SE0148 also makes calibration a lot easier.

The starters consist of a housing with a capacitor and a gas discharge lamp (see figure 2), of which we only need the latter. These lamps consist of a glass tube, filled with (a mixture of) noble gases, and a pair of bimetallic contacts that lead to gas discharge at a voltage of more than 200V. At these and higher voltages, the bimetallic contacts become warm and bend towards each other, causing a short circuit and stopping the gas discharge. This can be prevented by placing a resistor of approximately 24kOhm in series with the lamp.⁴

Although the lamp can be connected directly to 230V mains voltage in this way, it is preferable to use a low-power 12V DC to 230V AC step-up transformer in an IP65 housing as a power source for outdoor use. The gas discharge lamp itself can be connected to the spectroscope in a light-tight manner using a 3D-printed or milled housing (two can be seen in figure figure 2).

In addition to the calibration lamp, a light source is also needed to test the filters, preferably one with a continuous spectrum. For this purpose I used a tungsten lamp, which is nothing more than a 10W 12V DC halogen lamp. The lamp is placed in an aluminium enclosure with a 1.5mm diameter opening, which is equipped with a glass diffuser (see G in figure figure 3). The diffuser is cut from a piece of window glass and ground to a matte finish with standard grinding powders from 50 to 9 microns.

By far the most important improvement of the optical bench is the collimation of the tungsten light source. If we really want to know how a filter performs, the light must pass through the filter perpendicularly. Astrophotography filters work by means of extinction in a series of vapor-deposited layers on the glass substrate. The thickness of these layers determines which wavelengths are transmitted and which are blocked. If the light does not pass through the filter perpendicularly, but at an angle, the wavelength at which the filter is transparent shifts. This last point is interesting for us, because it has an effect on astrophotography. If we want to test what happens when we use a certain type of filter in combination with a very fast telescope, we first need to know what the filter does with right-angled transmission. Hugo already showed the effect of oblique transmission in his presentation and article, and this has also been discussed extensively elsewhere.⁵ When a flashlight is used as a light source, part of the light will not pass through the filter at right angles due to its diameter. Of course, we can make the flashlight a point source, but then too little light may reach the spectroscope. By adding a collimator to the light path, both can be avoided.

The collimator consists of two doublets (D1 and D2 in figure 4, note their orientations!) from a Bresser 26mm Plössl eyepiece. They are positioned so that the front of the glass diffuser (G) is exactly in the focus of doublet D1 and the slit of the spectroscope is in the focus of D2. Since the light between the two doublets is collimated (parallel), it does not matter how far apart the doublets are (in my setup this is about 23cm). Correct positioning of the doublets with respect to the diffuser and the slit is important and can be done simply by placing a camera with a lens between the doublets. First, the camera is set to infinity by focusing it on an object at least a few hundred meters away. If the doublets are at the correct distance from the diffuser or the slit, they will be shown focussed. The area between the doublets is where filters (F) can be tested.

In order to be able to place the filter exactly in the middle of the optical light path, it is useful to use a Self-Centering Jaw Clamp (SCJC) (see figure 4). These are available ready-made from reputable optical companies, but I chose to make it myself. I mounted the SCJC on a small turntable in such a way that the clamped filter is above the pivot point. Attached to the turntable of the SCJC is a pointer which indicates on a curved scale the f-number with which the rotated position of the filter corresponds. A simple (paper) diaphragm can be used to block the light path if the filter is smaller in diameter than the doublets (this is to prevent light from 'leaking' past the filter).

The original version of PySpectrometer used RaspberryPi cameras. With PySpectrometer3 I chose to work with ZWO ASI cameras (see figure 5). I tested the setup and software with the ASI290MM and ASI290MC. The mono version has the advantage of having a larger bit depth (12bit), while in the beginning the colour version had the advantage of making it easier to learn to recognise the spectrum. The camera is equipped with a CCTV C/CS-mount lens with a fixed focal length of 16mm (f/1.4 - f/16) from the company CGL. In combination with the ASI290 and the PHBS, the range of the recorded spectrum is approximately 440nm. Because the camera is also on a rotatable platform, it is possible to choose at which wavelength this range starts. A special adapter plate (connected to the lens with a screw thread) ensures a virtually light-tight connection to the spectroscope.

This ring is 6mm thick: 3mm is threaded and screwed into the lens, leaving 3mm that protrudes. The central 26mm is recessed by 3mm and within that is a 20mm diameter aperture. The front of this aperture is therefore exactly flush with the front of the lens-housing. The spectroscope is mounted against this central aperture. This adapter ring has several functions:

- ensuring that the lens is at a repeatable distance from the spectroscope;

- preventing the spectroscope from touching the lens and damaging it;

- sealing the connection between lens and spectroscope without having to use tape or something similar, even if the camera is mounted at an angle to the spectroscope.

The pivoting point between camera and spectroscope is approximately 5mm in front of the front of the adapter (so approximately 8mm in front of the front of the lens-housing). If the platform on which the camera sits hinges around that point, the spectroscope remains neatly centred in the adapter. Light can only reach the lens through the adapter at an angle of 45 degrees, but since that angle is much larger than the field of view of the lens, this hardly poses a problem (only a very small percentage makes it to the image chip).

I created the ring on my lathe from aluminium, but it can also be easily 3D-printed, while the thread is not absolutely necessary.

The spectroscopic filter test bench works with PySpectrometer3.