Because both mirrors of an RC are adjustable, there is a chance that collimating only the secondary mirror will move it off the optical axis of the secondary (assuming it was not adjusted correctly, see figure 1). This has consequences for the wavefront and therefore for the quality of the image. The same happens if the two mirrors are too close or too far apart. This also has consequences for the wave front and causes the telescope to be over or under corrected. Four steps are required to properly collimate an RC:

1) Aligning the optical axis of the primary mirror with the optical/mechanical axis of the telescope;

2) Roughly collimating the secondary mirror;

3) Checking the mutual distance of the mirrors;

4) Fine collimation of the secondary mirror;

5) The final check.

For collimation a special holographic laser attachment for the laser collimator has been made by Howie-Glatter, but since its use does not automatically lead to a good result, it is not used in this procedure. However, the final collimation has been checked with this laser collimator.

In the discussion above, it was reported by David Davies that a polystyrene disc with a properly centred peephole mounted in the primary mirror hole greatly simplifies alignment of the primary mirror's optical axis with the telescope and coarse collimation of the secondary mirror. A similar solution has been used for Theo's RC8. Instead of a polystyrene disc I made a 2″ adapter from aluminium with a one and a half millimetre diameter hole in its centre (see figure 2). I gave the surface a matte white finish, so that the hole is clearly visible with the coarse collimation.

At the back, this hole has been widened to 1.25″, so that tools such as a light source or camera can be inserted there. In principle, this adapter fits right into the focuser, but then there is the risk that, when the focuser is tilted, the hole does not represent the centre of the mirror. Fortunately, an adapter had already been made at the observatory to convert the visual back from the RC8 directly to a 2″ connection and the attachment I made fit in again without any significant play. So the focuser was removed and replaced with this adapter containing my attachment.

Of course there is a chance that the mirror is not 100% centred in the tube (and therefore not 100% concentric with the hole), but at the same time we are not sure whether the hole in the primary mirror is exactly centred against its parabolic shape and both options are expected to be equally reliable.

The secondary mirror was removed from the front of the telescope. first the three adjusting screws were loosened exactly a quarter of a turn, after which the central bolt was removed. Before doing this, the central bolt is first marked so that the original distance from the secondary mirror is known. In retrospect, we should have also marked the secondary mirror. Although the impressions of the adjusting screws were clearly visible, that still gave three options for replacing the mirror. Fortunately, the three adjusting bolts had been loosened exactly a quarter of a turn, so that the correct position of the mirror could be restored when it was replaced. The secondary mirror is removed so that the attachment in the visual back is visible through the mounting hole for the secondary. Illuminating the hole in the attachment from behind makes it easy to see and visually centre it in the mounting hole.

To make centring even easier I made a second tool that fits exactly into the mounting hole for the secondary mirror (see figure 3). This second attachment has a central hole of 7 millimetres and is therefore 1 millimetre larger than the hole for the bolt of the secondary mirror. The front of the second attachment is ground so that it reflects evenly. When making this attachment, it was deliberately chosen not to finish the hole with a bevelled edge, since that will lead to perception errors.¹

With the attachments mounted, the telescope was placed on an alt-azimuth table. Opposite it at about 75% of the focal length is a ZWO ASI290MC camera with ZWO zoom lens mounted on a tripod. By moving the tripod and varying its height (or by adjusting the alt-azimuth table), the illuminated hole of the visual back can now be easily centred in the hole of the secondary attachment (see figure 4). ZWO's ASICAP was used for this adjustment. This can also be done with other software such as FireCapture, but it is important that the digital cross-hair is always 'attached' to the image, which is the case with ASICAP, but not with FireCapture. With this cross-hair it is now possible to centre the visual-back in the image. Also the reflection of the camera lens in the primary mirror can be used in the adjustment.

Davies suggested centring the reflection of the central obstruction relative to its direct image. Now this works reasonably well, but due to the greater mutual distance between the primary mirror and the camera, centring is a lot more accurate when the lens is involved in the method. However, it is important that the visual back is centred on the camera image, so a fixed cross-hair is required for this.

Figure 5 shows that initially the RC8's main mirror was not aligned correctly. The reflection of the secondary mirror holder is not neatly symmetrical with respect to the mirror holder itself and the reflection of the camera lens is clearly not symmetrical either.

This image can be made completely symmetrical with the adjustment screws of the primary mirror, the end result of which can be seen in figure 6. Now that everything is neatly concentric, the primary mirror is perpendicular to the optical/mechanical axis through the visual back and the centre of the mirror holder.

If a tilt adapter is used to adjust the focuser, now is the time to install and adjust it. To this end, the visual back attachment is placed in the focuser with the focuser rotated maximally out of focus (extra-focal). The order of operation is then reversed: the secondary side attachment is centred in the reticle and the telescope is aligned with the camera by looking at the reflected image of the secondary mirror holder and the camera lens. With a skewed focuser, the result will be the same as in figure 5, but then the central light-spot of the attachment in the focuser is not concentric or even not visible at all.

The tilt adapter is used to get the light spot of the attachment in the focuser concentric again, whereby of course the primary mirror must also remain concentric. Finally, the focuser is rotated maximally intra-focally to see that the light source does not drift. If it displaces the focuser is not centred properly at all. This can be checked by turning the focuser in its entirety 180° and checking whether the deviation also rotates. If indeed the focuser is not well centred, and the solution is to make the intra-focal and extra-focal deviations equal.

After the primary mirror has been correctly adjusted, the secondary mirror can be roughly collimated, so that it can then be checked whether their mutual distance is correct. To this end, the secondary mirror is first re-mounted and roughly collimated, and the light source is removed from the centring attachment of the visual back. Then, through the hole in the attachment, the centre mark on the secondary mirror and the reflection from the attachment can be seen (see figure 7). It may be necessary to remove the baffle tube from the primary mirror at this step to allow enough light to reach the attachment. This baffle tube unscrews easily, with the RC8 I could just reach it by going in with my arm through the spider.

Now, by turning the secondary mirror collimation-screws the hole in the attachment on the side of the visual back can be centred within the centre mark of the secondary mirror (right hand side of figure 7). Once this is achieved, the secondary mirror is positioned well enough to check the distance to the primary mirror (but it is better to do this with a well collimated scope).

Checking the distance between the mirrors can be done with a Ronchi-test. To this end, a Ronchi-filter is placed in the focuser. If this is a photographic Ronchi-filter, then the camera must be equipped with a lens with focus at infinity. For the collimation of the RC8 a photographic Ronchi-filter from Gerd Neumann was used together with a ZWO ASI290MC and ZWO zoom lens.

The camera and the filter are held together with a self-made adapter. The Ronchi-lines become visible as soon as the Ronchi-filter is just off the focus point of the telescope. Checking the line pattern should be done with the filter intra-focal, that is, with the filter too close to the back of the telescope (i.e. the focuser is turned inwards). If the distance between the mirrors is correct, the lines of the Ronchi-filter will be straight (see figure 8). If the distance between the mirrors is too short, a pincushion effect will occur, while a barrel formation will occur if the distance is too great. If the filter is used extra-focally, the patterns are interchanged, figure 8 is therefore only valid for intra-focal use.²

Figure 9 shows the Ronchi-measurements before and after adjusting the mirror spacing. The Ronchi-pattern can be made visible by aiming the telescope with a ronchi-filter at a bright star. However, the seeing must be perfect, something that is a rarity here in the Netherlands. It is therefore easier to use an artificial star or a collimator. The images in figure 9 and figure 10 were taken using my 12″ collimator.

Correcting the mirror distance will affect the collimation and thus the telescope should be roughly collimated again. This can be done as described in step 2 or, if the setup is stable enough, by turning the adjusting screws of the secondary mirror until the ronchigram is visible again.

The last step in the collimation process is fine-collimation of the telescope. This was also done in front of the 12″ collimator. The alt-azimuth table is useful here, since when adjusting the secondary mirror, the doughnut quickly disappears from view. In figure 10 shows a properly collimated doughnut. The cross comes from CollimatorGrabber, a piece of software I wrote for this, and shows the lines along which the image is mirrored. The lower left and upper right sections are swapped to show if the central obstruction is actually central. A small deviation can still be seen along the vertical line (the left side is somewhat narrower than the right side). Measurements with CollimatorGrabber showed that the telescope is somewhat affected by mirror-flop (one and a half pixels, which corresponds to approximately 1.1″). With fine collimation this is averaged out by collimating the telescope in two positions (rotated 180° along its optical axis) and this deviation is therefore visible in the image.

Of course, after the work described above, which took about 4 hours, we were curious how things would look with the holographic Howie-Glatter laser. As can be seen in figure 11, it showed that the collimation was successful, even the spider's cams are concentric.

Of course, the real final test still has to follow: an astrophoto! But for that the weather has to cooperate…

At the beginning of this article I mentioned the focuser tilt-adapter. This adapter is only intended to align the focuser with the collimated telescope, not to collimate the telescope. A tilt adapter is necessary if the image shifts over the sensor when the focuser is turned in and out. With a skewed focuser, distorted stars will appear in one of the corners of the image. However, it is also possible that the tilt is between the camera and the focuser (when turning in and out, the image does not change, but the photo does show distorted or out-of-focus stars), in which case a camera tilt-adapter should be used to correct the tilt.

[1]: Hilster, N. de, Navigation on Wood, Wooden Navigational Instruments 1590-1731: An analysis of early modern western instruments for celestial navigation, their origins, mathematical concepts and accuracies , (Castricum, 2018), pp.491-493.

[2]: Lockwood, M., Cassegrain Formulas and Tips , section 3a, in: http://www.loptics.com/ATM/mirror_making/cass_info/cass_info.html (last accessed August 27, 2020). On October 3, this effect was once again determined experimentally, see below.

I checked Theo's RC8 again on October 3, 2020. The previous adjustment had already given a huge improvement, this new inspection was to see whether the mirror distance could be improved even further and whether the focuser is neatly aligned with the optical axis, something that we had not looked at last time.

The mirror-spacing has again been checked with the Ronchi-test, but this time with a slightly different procedure. First the collimation was checked, then the mirror distance was tested with the Ronchi-test, then the secondary mirror was adjusted and again fully collimated before repeating the Ronchi-test. Last time I did the collimation globally, but that has a detrimental effect on the Ronchi-test. The Ronchi-test is not very sensitive anyway, so it is important to have the collimation as good as possible.

Checking the focuser was done as in the first step above, but now with the light source (in this case an artificial star) in the focuser. Paul and Theo had made this artificial star after my example. First the alignment of the primary mirror is checked as shown above and then the attachment in the visual back is replaced with the focuser containing the attachment (or an artificial star). Then alignment of the camera is checked with the OTA by centring the image of the attachment in the secondary mirror holder and the reflected image of the camera lens. If the light source in the focuser is shown neatly centred in the focuser, the alignment is correct. The focuser is then shifted from one extreme position to the other. If the focuser is properly aligned, the position of the light source will not shift.

To make things easier, I made a new alt/azi table to aim the telescope. The previous two were still made of wood and served as prototypes for this (provisional) final design. It is made of solid aluminium beam (40x30mm) and aluminium strip (200x20mm). All in all, this table weighs approximately 10kg, which benefits stability.

A day after this second adjustment I received confirmation that it had led to good results. Theo sent me adjacent image of Vega. The spider's diffraction spikes are razor-sharp. The photo was taken without a flattener, which gives the opportunity to see whether the collimation is well balanced. Coma can be seen in the corners and they all point towards the centre of the image, which is a good sign.

Prior to this final adjustment, Theo had set the focal length at 1646mm, while the design size was 1624mm. Now, I had read somewhere that adjusting the secondary mirror would have a tenfold effect on the focal length (so 1mm change in mirror distance gives 10mm change in focal length). We then moved the secondary mirror out approximately 2.25mm (three turns with the M5 bolts used to adjust the secondary mirror). As a check, I measured this distance with a calliper and came to an actual difference of 2.6mm. This increased the mirror distance and should reduce the focal length. Astrometry from Vega's photo showed us that the new focal length had become 1617mm, a difference of 29mm in the right direction, bringing the focal length change factor to around 11x.