Optical Testing – Star Testing / Diffraction Testing

18 Using a CCD camera, alter the primary mirror tilt so an outside-of-focus star image, in the center of the field, is an evenly lit circular symmetrical annulus.

19 Similarly, alter the secondary tilt to ensure any residual aberrations in the outer field are radially symmetrical (balanced), or diffraction mask spikes of a near-focused star image, in the center of the field, have perfectly intersecting lines.

20 Repeat 18–19 to converge on the two mirror positions.

21 Focus the image, plate-solve and use the image scale to calculate the effective focal length and compare with the telescope specification.

22 Adjust the mirror separation, if necessary, assuming a 10:1 ratio (a 1-mm increase in mirror separation effectively reduces the focal length by ~10 mm) either by adjusting the secondary mirror position (fig.15), or for small changes, moving the three adjusters on the primary mirror to the same degree (a M6 bolt conveniently has a 1-mm pitch) (fig.14).

23 Confirm alignment with another star test (steps 18–22).

Preliminary Checks (1–9)

Everything has a tolerance and a RCT is a sensitive beast. Any collimation device that inserts into an eyepiece tube is required to be perfectly centered. Putting aside the vagaries of the eyepiece clamp for a moment, the quickest way to verify device centering is to rotate it in a V-block. In its crudest form, a V-block is constructed from four three-inch nails, hammered into a piece of wood to form two V-shaped cradles. Lay the collimation scope, Cheshire eyepiece, laser or sighting tube body in the cradle and check its beam or image center is stationary as it is rotated. The greater the distance from the device to the convenient wall, the more obvious any error. The best lasers offer some means of adjustment, normally by three opposing grub screws. If yours does not and is not accurately centered, send it back. In the case of a precision level, the best devices have a bubble level with 0.03° markings which, with care, enable measurements to 0.01°, 10x more resolution than its digital readout. This is sensitive enough to detect a thin piece of paper placed under one end. To calibrate the level, place it on a smooth level surface, note the bubble position and ensure it is consistent when the device is turned around to face the other way. On the unit in fig.10 there are two tiny screws that tilt the phial. Armed with your calibrated devices, it is time to start checking and adjusting the mechanical system as much as possible.

The focuser system has a difficult job to remain orthogonal to its mounting. The nature of its construction translates microscopic errors in the draw tube into angular movement and, this is without swinging it around with a heavy camera on its end. All good models offer some form of tension adjustment that, at the same time, remove some flexure between the sliding parts. The better ones, like the large FeatherTouch and Moonlight Telescope Accessory models have collimating adjusters. In the absence of a laser, a precision level can be used to ensure the camera and focuser mounting flanges are parallel at all angles (figs.10, 11). This is most conveniently adjusted by placing the focuser assembly telescope-end down onto a flat horizontal surface. Place the level on the surface and adjust its level so that the bubble lies between the end-stops (within ±0.1° from horizontal) in both a conceptual E–W and N–S direction. Note the exact position in both instances. Then, place the focuser on the end of the focus draw-tube. Nominally assign one of the collimating screws to “North” and adjust the other two first, to achieve the same E—W level as the reading from the plate. Then, adjust the third one for N–S calibration. In this orientation, facing downwards, flexure is at a minimum and this adjustment represents the average position. If all is well and the focuser has a rotation feature, it will be consistent at all angles.

If the back-plate of your RCT is a flat aluminum panel, rather than a complex casting, it is easy to go further and confirm the focusing coupling plate is parallel to the panel. In this case, with the telescope pointing downwards, mount the (calibrated) focuser onto the mounting plate and confirm the panel and the camera mounting flange are parallel in N–S and E–W directions (fig.11). It is very useful to have a collimation coupling plate on the back of the telescope, especially if there are no centering adjustments on the secondary spider. On my 10-inch truss model, I decided to square the primary mirror, or more correctly, its housing. I carefully rested the scope on its front face (or you can mount it and point it downwards). I removed the smaller push-screws on my truss model and measured the depth of the hole to the outside housing. For this I used the depth gauge end of a Vernier caliper. They were in the range of 10–10.5 mm. I adjusted the pull-screws until the distance to the back of the mirror housing was exactly 10 mm (the back-plate is 8-mm deep). This is not essential but a useful reference if things go wrong. In my case, the black push-screws were rather short and only engaged the back-plate through half its depth. I also discovered that their flat ends would “corkscrew” and displace the mirror laterally. With the pull-bolts in place, I rested the RCT on its front face and replaced these grub screws with a longer pointed stainless-steel version. Not only are these easier to spot in the dark, but the longer thread engagement is more stable between the soft aluminium and stainless steel. Usefully, each point creates a small conical indentation in the softer aluminum and minimizes lateral movement during adjustment.

fig.13 The secondary and baffle, showing the central donut. The donut is not silvered and appears black. Initial collimation relies upon the fact that the manufacturer, after polishing a glass surface to 100 nm, is able to locate the center within 1,000,000 nm (more likely than relying on the baffle and mirror being accurately centered).

fig.14 On the back are the three primary push- and pull-bolts A, B & C. You can also see two focus-plate adjusters D & Eand the focuser collimating adjusters F & G. I changed my primary push grub screws to pointed stainless steel versions.

fig.15 The secondary mirror flange A is fixed to the spider tilt-mechanism. The mirror and baffle assembly C can be unscrewed to set the mirror distance and is locked in place by the knurled ring B. The pitch of the lock ring is about 0.75 mm.

High-end RCTs often have adjustable spiders. The current popular GS() derivatives have an assembly with no obvious method of centering the secondary mirror other than disassembly and experimentation. In my case, I used a Vernier caliper to confirm that the mounting boss was in the physical center of the front truss ring. It was within 0.1 mm of the physical center, but unfortunately one cannot infer that the front truss ring is aligned to the optical center of the primary mirror. If there is no easy way to adjust the secondary mirror position using the supporting spiders, it modifies the subsequent process used to the

Lastly, if your RCT has its mirror separation set up in the factory, tape over one of the primary mirror adjusters to prevent any accidental change to the mirror separation. Some larger RCTs have their primaries bolted down tight for transit and in these cases, follow the manufacturer’s instructions to set the initial primary position, normally by unscrewing the pull-bolts bolts by one or two turns. Similarly, if your secondary mirror is only attached with three pairs of opposing bolts, tape over one set. The lower-cost RCTs have a secured central fixing bolt that is used to set the distance of the secondary mirror base, rather than three sets of opposing bolts. In these designs, you need to use all three tilt adjusters, by easing and tightening in pairs, in that order, to rock around the central sprung bolt. It is not immediately apparent but the GSO-based RCTs have a very useful precision mirror separation adjustment (fig.15). The black knurled ring and secondary baffle unscrew, leaving the bolted back-plate untouched. This thread on my RCT has a pitch of 0.75 mm, enabling precision adjustment. As it happens, my RCT required a mirror separation reduction of about 2.5 mm, to increase the focal length to 2,000 mm, accomplished by unscrewing the baffle by ~3.3 turns and screwing up the knurled ring to lock it into position. (The secondary mirror appeared to remain perfectly centered but I checked its collimation after the adjustment with a laser and fine-tuned it with a star test to be sure.)

Precision Focuser (10)

To improve on the focuser assembly collimation requires a laser. (Please remember to observe the safety instructions that accompany a laser.) With the focuser fully assembled to the telescope, make adjustments to the focuser’s collimation (if it has that facility) so the laser dot remains stationary on the secondary mirror, as the focuser is rotated. (If your secondary mirror has a lens cap, one can make a simple target by making a reference dot on a small piece of masking tape to assist the assessment.) After removing the cap, if the laser is not incident on the middle of the donut, it implies the focuser axis is not aligned to the secondary. Something has to move; if there is no obvious spider centering method, center the beam using the focus-tube coupling-plate adjusters. This is a compromise that is discussed later in more detail. (If you are unable to clearly see the laser on the mirror surface, place a piece of clean polyethylene bag on the mirror surface, as in fig.6. It scatters the laser beam, making it visible but at the same time, you can still see the donut too.) At this point, do not use the secondary tilt adjustments to try and center the laser on the donut! For enclosed RCTs, it is necessary to make the equivalent of a dentist’s mirror and peak back at the secondary mirror.

To assess whether focuser sag is going to be an issue, push the focuser tube in different directions and notice if the beam moves about. If there is excessive play, the focuser mechanism may need a small adjustment, or more drastically, upgraded with a more robust unit.

Initial Secondary /Focuser Alignment (11–13)

The aim is to place the center of the secondary mirror on the main optical axis and set its tilt to align its optical axis using a laser in the focus tube. Once the laser beam is perfectly centered, initial alignment is complete when the beam reflects back on itself (figs.7–9).

That discussion on compromise is required here; the ideal solution is to align the focuser axis independently with the primary mirror’s optical axis and shift the secondary mirror laterally to center the beam on the donut. Without that centering facility, the alternative is to angle the focuser assembly to aim the laser at the secondary donut. This tilt moves the focuser axis with respect to both mirror optical centers. Although this is a compromise, since the secondary mirror is about 30x further away from the focuser adjuster’s tilt axis than the primary mirror, any de-centering with the primary mirror is minimal and the final alignment of both mirrors during optical testing reduces the error to a small (sub-millimeter) image displacement on the sensor.

In practice, carefully place the laser in the eyepiece tube, so it sits square, center the beam on the donut, either by adjusting one or more spiders (if your RCT has that facility) or by tilting the entire focuser assembly with its back-plate coupling. If your laser unit tips within the eyepiece coupling ring when the locking screws are tightened (a common issue with those units a single clamp or without a brass compression ring) point the telescope vertically downwards and let the laser unit simply rest on the eyepiece tube flange (assuming the flange is square). I achieve good alignment consistency by using a light touch on the clamp screws with metal-to-metal shoulder contact.

The next step is to adjust the secondary mirror tilt to reflect the beam back onto itself. (This involves minute adjustments. As delivered, the secondary mirror tilt-adjuster bolts on my RCT were stiff and jerky, making small adjustments impossible. I removed mine, one at a time, lubricated and replaced them before bench collimating.) On an RCT, this can be done by carefully peeking down the central baffle. (Since the laser is aimed at the secondary mirror, there is no risk of a direct incidence on your eye.) The outgoing and reflected beams can be quite bright and fuzzy and difficult to distinguish. I use a pair of polarizing sunglasses, and tilt my head to eliminate the glare, or reduce the power of the laser. This makes an accurate assessment considerably easier. Alternatively use a laser with an exposed target, like those from Baader Planetarium, or use the Tak in place of the laser and center the dot and donut, as in fig.6. These last two methods also work for those RCT derivatives, such as modified Dall-Kirkhams, that have refractive correction optics within the baffle. In the case of the RCT, the faint diffraction halo of the laser illuminates a considerable portion of the secondary mirror and is reflected back to the primary. If you look carefully at the white face of the laser, you will see a faint donut shadow on the laser’s target surface. When the mirror is properly centered, the laser beam, reflected beam and donut shadow are concentric (see figs.7–9). Since a change in mirror tilt has a minor effect on the donut position, repeat the mirror centering and tilt adjustment one more time (if required). Lasers are wonderful things and the donut shadow on the laser face is something I have not seen mentioned before. It is the equivalent of the view through the Tak and is a viable alternative.

The Fifth Dimension

The proof of bench testing is that, after doing several extended star tests, the optics still pass the bench test. In my case, this was not always so. In a few instances, the final alignment was indistinguishable from the bench setup and notably different in others. Bizarrely though, in every case bench alignment always needed a reasonable adjustment during star testing to achieve collimation. At the same time I would sometimes hear a creaking noise during adjustments during star testing. The cause was the primary mirror cell shifting laterally on its three mounting-bolts during adjustment. This is a common issue in some of the lighter designs. (It also accounts for some of the variations between user-experiences, with one method or another, and the reason that some of the more confident users insert a stiff elastomer to provide lateral support to the mirror cage.) I realized that during star testing near the Zenith, the act of adjustment was equally to do with re-centering the mirror. As a consequence, although I may assess the alignment of the RCT in a horizontal aspect, I always point to the zenith to make primary adjustments. My unit is light enough to rest on a table or up-end. Heavier units will require to be mounted and swung on the DEC axis. The bottom line; a laser-based adjustment is only as good as the primary mirror centering with the focuser and sensor axis.

fig.16 A classic Cheshire eyepiece, this one has a white rear face, typically used during Newtonian collimation process. This one is beautifully made and does not have an internal crosshair to obscure the view. Its color even matches the Paramount! A Takahashi scope extends the eyepiece outwards so that it can detect the thin gap between the mirror reflections. This eyepiece can do this too, if it is similarly extended with focuser extension tubes (providing the 200 mm or so extension does not introduce focuser tube sag).

fig.17 The view through the Cheshire, showing a marginal error, indicated by the spider clamp showing (A) and the slightly larger gap (B). As the eyepiece is moved further out, the gap (B) between the mirrors increases and it becomes considerably easier to perform the alignment. This image is taken at the normal focus position, but moving out another 200 mm with extension tubes and the focuser rack, makes it easier to see and equalize the thin annulus.

Initial Primary Alignment (14) (sighting tube)

In this process I prefer to use an original Cheshire eyepiece (fig.16) or a simple viewing hole. I aim the RCT at an illuminated white wall (or you can place a diffuser over the end) to evenly illuminate the mirrors. I then look through the Cheshire and adjust the primary mirror so that the mirror and its reflections are concentric (fig.17).

fig.18 The view from the rear, through a camera, showing the concentric mirror outlines and aligned spiders. The camera is precisely centered on the collimated secondary mirror and the primary mirror is adjusted to align the spiders. This should be confirmed by concentric mirror reflections too, from their outside edges rather than their baffles. It is often confusing to work out what you are seeing in the reflections. A is the outer edge of the secondary mirror, 8 is the reflection down the focus tube, C shows the outer part of the spider vane and its double reflection (aligned) and D is the outer edge of the primary mirror baffle.

In particular, I look at the tiny gap between the inner and outer mirror reflections and at the same time, I also note the symmetry of the outer field. When my RCT primary is not aligned, I can just see the bulge of a spider’s outer support bracket at the outer edge. When it is aligned, all four spider brackets are hidden, unless I move my eye about and view obliquely through the eyepiece.

Initial Primary Alignment (14) (camera)

This process is a variation of the above one, except for enhanced accuracy, I use a camera, fitted with a telephoto lens and mounted on a tripod and aimed squarely at the secondary mirror, so the center of the camera lens is seen reflected in a mirror. Any misalignment is more obvious at longer viewing distances and I typically do this from about 5 m (15 ft). The spider and its reflection can be seen in the viewfinder (fig.18) and usefully, my Fuji X-T1 camera has an option to magnify the focus point when the manual focus ring is moved. By making small adjustments to the primary mirror, I align all four spiders with their reflection. This works on the assumption that if the two mirrors are not aligned, one or more of the vane images and their reflections will be disjointed. At the same time the correct primary setting is confirmed by concentric images of the two mirrors’ outer edges (not the baffles). In both processes, if one has already established the mirror separation for the right focal length, only use two of the three primary adjusters.

Initial Primary Alignment (14) (laser)

Throwing technology at the problem introduces other interesting possibilities; for example using SCT laser alignment tools on a RCT. One can either shine a laser onto the primary (from the front, as the Hotech advanced CT collimator) or from the rear onto the secondary (as the Howie Glatter holographic projection method). Both products are principally marketed for SCT and Newtonian users but can be used for RCTs too. This is because in their simplest deployment, they are used to adjust the secondary mirror and crucially, assume the primary mirror is fixed (aligned) and the focus tube is aligned on the optical axis of the primary mirror.

The Howie Glatter laser has a number of alternative attachments, one of which beams concentric rings (fig.19). One alignment technique relies upon the mirrors being filled with light rings (and concentric to the mirror edges) and beams them onto a nearby wall for closer examination. Alignment is tuned and confirmed by checking the concentricity of the central shadow. In practice, this test is very sensitive to the laser alignment onto the secondary and the secondary tilt. My holographic attachment, as supplied, projects an uneven pattern and central spot, hampering assessment. I remedied this by leaving the standard 1-mm aperture screwed in place and taping the holographic attachment back-to-back. Some on-line methods suggest to ensure the circular rings are concentric on both mirrors. In my old darkroom and even after a period of acclimatization, my pristine mirrors do not reveal the concentric rings incident on their surface. There is a degree of mix’n'match between methods; one method is to set up the secondary mirror to a collimated focuser with a simple laser and then using the holographic attachment to fine tune the result to centralize the projected rings.

As mentioned earlier, the laser must be precisely centered onto the secondary (a 1-mm error here equates to a quarter-turn on a primary mirror adjuster) and the secondary aiming squarely at the primary, or it adversely affects the outcome. As such, this test is also very useful as an independent method to confirm system alignment prior to optical testing and is most easily accomplished indoors, projected against a light colored wall.

In the case of the Hotech device, it uses the reflective properties of the two mirrors to ensure that three laser beams, parallel and equidistant to the primary mirror axis, are reflected back on paths that are equidistant from the optical axis. To do this, one first squares the primary mirror to the target and then adjusts the secondary. The reflections require an additional (semi-silvered) mirror to be inserted into the focus tube and the physics rely upon it being coplanar with the primary mirror (as assumed in the case of a SCT). Hotech have a unique 2-inch adaptor tube design with expanding rubber glands. These are designed to overcome de-centering issues in the simpler 2-inch eyepiece tubes. I am not enthusiastic about compliant rubber interfaces and in the case of the Moonlight focuser, with its close tolerance smooth bore, I achieved even better repeatability by pushing the attachment up against the metal collar and using minimum force on the three clamp screws.

fig.19 The projection from the RCT, fitted with a Howie Glatter laser and holographic attachment. Up close, the spacing between the ring atA and the central shadow is slightly than the spacing at B, indicating (assuming the laser is hitting the secondary square on and aligned to the optical center) that the secondary mirror is slightly misaligned. At the same time, the spacing between the outer ring and the edge of the diffuse background is equalized. It sometimes helps to attach a piece of paper to the wall and mark the positions, rather than judge by eye alone. As with other laser methods that rely upon the focuser alignment, if the primary mirror is not precisely centered with the focuser assembly, this will affect the accuracy of the final result.

The laser source and its circular target in effect ensure the mirrors are parallel using secondary adjustments (fig.20). With that accomplished, the rear target behind the semi-silvered mirror (inset) indicates if the primary mirror is tilted. When aligned, the three incident beams are symmetrically placed around the target crosshair. (As this target plane approaches the focus plane, the dots converge to a single bright dot.)

In fig.20, the Hotech unit is confirming RCT collimation, after classical star testing, back on the bench. In this case, it is very close to being fully aligned. The smallest change to any setting throws off this bench alignment, indicating its sensitivity is considerably higher than a simple single beam reflection method and arguably as accurate as classical star testing in typical seeing conditions. Indeed, the results from using this device to set up the secondary mirror correlate perfectly with the “hall of mirrors” test described below.

As mentioned before, all these alignments work best if both focuser tilt and centering errors have been minimized. That is not always mechanically possible, so the next best thing is to collimate the focuser assembly as best as one can, assuming an optical axis that runs through the secondary donut. If a device such as the Hotech is not available, the next best thing is to tilt the secondary to reflect a simple laser beam back on itself (as explained in the earlier sections) and then proceed with the primary mirror centering, viewing from the front or rear, to fine tune both mirrors.

Tuning Alignment (15–17) (Visual)

With one or both mirrors in its rough position, we use a sighting test from the front, suggested by Jared Wilson on the Cloudy Nights forum, dubbed the “hall of mirrors” test. This is best done by eye alone and is very sensitive to the relationship between the two mirrors. This can be used for setting either the secondary or primary as, ultimately, it is only the relationship between the two mirrors that is being evaluated. Crucially, it is independent of the focuser assembly but only confirms the mirrors are parallel. (They might not be necessarily on the same optical axis.) As such, it is best to adjust the mirror that is assumed to be the most out-of-alignment. In my case, after star testing my RCT and then returning the telescope to the optical bench, I found the secondary mirror still reflected an incident laser beam back on itself and used this visual test to modify the primary position.

fig.20 The Hotech Advanced CT laser is a very sensitive test due to the double reflections employed in its optical design. Excellent results are possible but only if close attention is paid to the initial setup; for instance, moving around on floorboards affects the laser trajectory. Conceptually, the front target confirms secondary alignment and the rear target confirms primary alignment.

This test is conceptually simple: View down the end of the RCT from about 0.5 meters away, close to the center axis, so you can see the repeating reflections between the two mirrors. When the mirrors are parallel on that axis, the spider, its reflection and the reflection of your pupil in the primary mirror are aligned. At the same time, the repeating and diminishing reflections of the secondary baffle are symmetrical and the subsequent spider reflections are aligned too. After checking one axis, repeat for one of the adjacent spiders and confirm the collimation is true on that axis (see fig.21).

The multiple reflections make this test very sensitive to any misalignment and it can easily detect the smallest turn on a secondary or primary adjustment screw and obviously, is unaffected by seeing conditions. Usefully, it can be easily performed with the RCT in situ and does not require any equipment, though I do make the primary adjustments with the mirror in a horizontal position.

It does require a few tries to figure the relationship between the error and necessary adjustment. Again, it helps to take notes in case you need to retrace your steps and to become familiar with the effect of a push- or pull-adjustment.

I use the following process to rationalise the diagonal spider vanes with the two primary adjusters set 120° apart: First, I tilt a mirror left/right using equal and opposite adjustments to the two adjusters in the 4 and 8 o’clock positions, until the reflections (warts and all) of the two upper (or lower) spider vanes are mirror-images of each other. I then know that any remaining error is caused by an up or down misalignment. To correct this and to ensure that I do not introduce a lateral tilt I use all three secondary adjusters (or both primary adjusters) moving the two bottom adjusters by the same amount. As mentioned earlier, this new mirror position ensures both mirrors are approximately square-on but not necessarily aligned on the optical centers.

So, there are two approaches using the same technique: If one is more confident on the initial primary mirror position, use this secondary tuning step a few times to converge on the optimum position, and align the primary by repeating the rear sight test. If the primary tilt is less likely to be correct (and assuming the focuser alignment is accurate) the order of these alignments is logically reversed; the secondary alignment is solely accomplished using a laser reflection and the primary mirror is setup using the hall of mirrors or a holographic projection.

So, a good collimation satisfies the hall of mirrors test (confirming parallel mirrors), a simple laser fired at the center of the secondary will reflect back on itself (confirming secondary alignment) and at the same time the Howie Glatter holographic projection should be symmetrical. There is no single answer here on which road to travel, since every experience may differ and, if the primary tilt is incorrect, the end result will likely not be optimum. Ultimately, however, all roads lead to Rome.

The good news is that with care, the combination of these visual techniques consistently achieve a good alignment and subsequent optical testing demands the smallest of corrections. The last step is to confirm the bench alignment holds true for different orientations; the easiest way being to rotate the RCT through different angles and repeat the hall of mirrors assessment.

A Novel Alternative for Mirror Collimation

As an interesting aside, a Harvard University Education paper 1969, by J Krugler, describes the collimation of a professional observatory RCT. This proposes a novel solution for aligning the optical axes, by introducing an independent reference point. In an open-truss construction, they stretched two thin wires across the truss to create a crosshair in-between the mirrors. All other things being equal, if the optical axes of the two mirrors are coincident, the crosshair reflection in the secondary coincides with the actual crosshair and the reflection of the secondary mirror in the primary is concentric. In the paper, they employ a theodolite to align the reflections but a tripod, digital camera and a telephoto lens, mounted on a sturdy tripod is a good alternative, with a little imagination.

In an attempt to recreate this, I fastened a small circular flat mirror, removed from a bicycle accessory, to the back-plate of the focuser tube with double-sided sticky tape. The mirror was prepared by finding the middle and drawing a cross on the glass surface to facilitate centering. In my case, I placed the telescope on a table and aimed an APS-C camera (fitted with a telephoto zoom lens and mounted on a sturdy tripod) at the mirror. I made small adjustments to the camera so that I had a perfectly-centered reflection of the lens in the mirror at the center-mark of the viewfinder. I then removed the 2-inch adaptor and knew I was looking down the middle of the focus tube. I stretched two thin enameled-copper wires across the truss joints, holding them in place with masking tape. Using the focus ring and confirming by taking photographs at f/22, I made small changes to the secondary so I had coincident cross-hairs. I chose to alter the primary so that the reflections of the spider were aligned, on the basis of human perception of the vernier effect. In the event, I achieved reasonable alignment on the bench but with sub-optimal centering of the mirrors (fig.22). As an experiment it was interesting but relied upon the focuser tube being aligned to the secondary’s mechanical center, which in my case was not adjustable. The drawback of the original method on an amateur scope is one of scale; a minor displacement of the wire cross-hair in a small space introduces a large arbitrary optical axis angular error when its purpose is to define the optical axis to the secondary. Two other things came to mind; not everyone has an open-tube RCT and the focuser-mounted laser is a better way to define an optical axis to the center of the secondary. It was a useful exercise, however, and some of its lessons are blended into my collimation plan and echoes those parts of other techniques that align spider reflections.

fig.21 The view from the front, though a camera, showing the concentric mirror reflections and aligned spiders. This is sometimes referred to as the “hall of mirrors” test. The test is repeated for a second spider at 90°. To perform this check, sight down the telescope from a few feet away so that the spider and its reflection coincide. If the mirrors are aligned on this axis, the receding reflections of the secondary baffle will be symmetrical about the spider vane. Any small misalignment causes subsequent reflections to deviate further off center. Having confirmed it is aligned on this spider, repeat for its neighbor. It can be used to align either mirror to the other.

fig.22 This is the “theodolite” view from the back, with the fine crosshair and its reflection coincident in the secondary mirror. Here, you can see a small misalignment; although the spiders are aligned, the mirrors are not concentric. This arises due to a tiny displacement error on the crossed wires, which is more significant on a small RCT, such as this.

Validation

So how does each fare? To confirm the validity of each bench-testing technique, after optical testing I returned the collimated RCT to the bench and evaluated it with the various tests. Gratifyingly, they all confirmed collimation and did not suggest any significant “deviation”, unless the mirror had shifted laterally. This indicates the collimation setting is within the usable tolerance of each method and confirms bench testing for coarse adjustment and star testing as the fine tune. The tolerances of each method are different, however, and after doing some sensitivity analysis, working backwards from a perfectly collimated RCT, I was able to make a simple comparison, summarized in fig.36, at the end of the chapter.

Artificial Stars

The universal law of stargazing applies; three months of cloud follow any significant astronomical purchase and obviously our thoughts then turn to using artificial stars for alignment. With an artificial source, the device is placed tens or hundreds of meters away and the telescope aimed at it (with mount tracking disabled). There are advantages and disadvantages of this approach:

1 The telescope is stationary, so tracking issues are eliminated.

2 Testing can be done in daytime or at night, within reason, depending on air turbulence.

3 The telescope is normally in a horizontal attitude, which challenges the mirror-support system on large aperture instruments, and may lead to unexpected results when the tube is later on.

4 A horizontal aspect is susceptible to near-ground air turbulence.

5 An artificial star needs to subtend a smaller angle than the telescope’s effective resolution.

6 The necessary small hole size (less than 0.5 mm) and long distances may be difficult to achieve in practice.

7 A starfield conveniently has many stars over the entire image, allowing simultaneous evaluation over the entire CCD image – something that is particularly useful for two-mirror systems. A single artificial star may take more time to evaluate for multiple positions.

8 Long focal length telescopes may not have sufficient focus extension to focus on a nearby artificial source and the additional focus extension may also change the focuser-tube alignment.

9 Reflecting telescope aberrations change at close focus distances and will impose a practical limit on the closest target placement. As such, an artificial source requires a little planning; its distance and size parameters are interlinked with the telescope specification and each user has to determine their own compromise.

fig.23 This commercial artificial star consists of a bright white-light LED behind a laser-cut pinhole. The pinhole is 0.1 mm diameter (100 micron) and is suitable for a range of popular RCTs, providing its distance is sufficient to create a pinhole angle that when subtended at the sensor, is less than the Airy disk radius of the telescope under test.

9-Ball Tester

It occurred to construct a multiple-star target, using 8 balls mounted in a circle and with a central ball. In that way, these allow simultaneous assessment of central and peripheral aberrations with ease. At 25x the focal length, and with a KAF8300 APS-C CCD, it requires the target to be a little less than 350-mm (14-inches) square. In one of those scrap-heap-challenge moments, I found a black plastic seed tray, approximately 400 mm square and mounted nine ball bearings in a 300-mm diameter circle with bathroom sealant (fig.24). This size will suit a APS-C chip user at 25x the focal distance. Other assumptions facilitate other setups; one might add more balls inside the outer circle to work with a smaller chip size and so on. At night, I use a white LED torch, with a 50-mm diameter reflector. I diffuse its flat lens with a piece of tissue paper and point it towards the balls from about 6 m (18 feet) away, to subtend 0.5°, as the Sun does. I do get some weird looks from my neighbors, but they are used to me and my mad experiments now. An alternative to this Heath-Robinson method is to make use of any calibrated telescope jog commands in your telescope driver. I use TheSkyX Pro and apply a 10 arc-minute jog setting to move a central star to the cardinal points within the image frame. Since the Paramount has very little backlash, I am able to precisely move a single star around in a repeatable manner around the frame, without resorting to analog slew methods using sustained button-presses on a handset.

fig.24 My9-ball star tester. The nice thing about this is its repeatability; there is no need to find a good patch of equally bright stars. Their placement allows one to quickly assess the balance of aberrations around the optical axis. Here the diameter of the array is about 300 mm, suitable for a APS-C sensor placed at a 25x focal-length distance.

Classical Star Testing

Star testing lays bare the optical performance of any optical system. Used properly, it can distinguish minute aberrations in an out-of-focus image that would otherwise be visually indistinguishable in the diffuse blob of an in-focus image. Used properly it is very revealing; for instance, it is easy to fear the worst and believe one has pinched optics when focused stars are irregular in shape, when in fact a de-focused image identifies it as coma, arising from misaligned mirrors.

In the case of a RCT, one prominent method consists of two cyclic events; removing coma on a central, de-focused star using primary mirror adjustments and subsequently balancing the astigmatism in the surrounding image area using small adjustments to the secondary mirror tilt. The two adjustments interact to some extent and a couple of iterations are normally required to create symmetrical aberrations. Star testing is very sensitive to any number of aberrations and is a substantial subject in its own right. The ultimate guide is the book Star Testing Astronomical Telescopes by Harold Suiter. Although RCTs are only mentioned in passing, there are many examples of other optical configurations with central obstructions, but not necessarily aligning dual, curved mirrors. The book has an excellent section on the use of artificial stars, either illuminated pinholes or reflections off shiny spheres as substitutes on a cloudy night. The prior parameters for star testing are derived from his recommendations. These involve a star test of a single star, which makes perfect sense for many optical configurations with a plane mirror or with few adjustments. In the case of the RCT, the primary mirror is often adjusted to optimize the appearance of a central star but in addition, uses multiple star positions to confirm the optical balance in the image periphery, largely determined by the secondary mirror alignment.

The following star-testing process is a slight adaptation of that suggested by Rich Simons of Deep Sky Instruments in their support documentation and uses the imaging CCD camera to confirm the results, rather than an eyepiece. In doing so, it avoids the complications of further focus extension, or the use of a diagonal, and follows the following iterative process applied after bench alignment:

1 Center a star and alter the primary mirror tilt so an outside-of-focus star is an evenly lit circular annulus (i.e. remove on-axis coma).

2 Balance the image, with small adjustments to the secondary mirror tilt, so that the aberrations in the image periphery (mostly astigmatism) are radially symmetrical.

3 (Optional) If the optimum mirror separation has not yet been established, before fine-tuning with a second iteration of 1 and 2, focus the image, plate-solve and use the image scale to calculate the effective focal length and compare with the telescope specification. At the same time, check for extreme field curvature and tilt with a program such as CCDlnspector. (Field curvature is sometimes confused for aberrations in the image corners. This can be a symptom of an incorrect mirror separation; as the mirrors become closer, the field has more spherical aberration correction). If these two test results correlate, adjust the mirror separation.

4 Repeat 1, 2 and 3 until you have an evenly lit circular annulus in the center of field, with a balanced outer field, such that any astigmatism is evenly distributed and astrometry confirms the correct focal length.

5 Give a gentle thump with the heel of your hand to the RCT back-plate, and flip the telescope, to relieve the stresses and check the collimation is still true. (If you have ever built or trued up a bicycle wheel, you bounce it to relieve the torsional stresses in the spokes.)

Just before we move on, now is the time to tape over one set of primary adjustment screws (and secondary adjusters, if there is no central bolt) to prevent mirror-separation creep.

Star Testing Parameters – De-focus

Successful star testing requires stable atmospheric conditions. The best target is a bundle of bright(ish) stars at high altitude that fill the CCD sensor, say a loose cluster, and on a night of good seeing. You need a good central star and ones around the periphery to conduct the full alignment. The star test is conducted outside of focus, in that, having focused the image, the CCD is moved outboard, or, if you employ a secondary focuser, the secondary is moved away from the primary mirror. With a large secondary obstruction, the focuser should be moved a distance of about 5–8 aberration wavelengths (n). To equate that to a stepper motor offset, requires the aperture ratio and the following equation from Suiter’s book:

where F is the focal ratio, n is the number of aberration wavelengths and X is the wavelength (assumed 550 nm). Fig.25 compares the out of focus star appearance for a range of aberration wavelength positions.

For n = 5, my f/8 RCT requires a 1.4-mm outboard movement. Its Lakeside focus motor has a step size of 3.7 µm, so evaluations take place after moving outwards by about 380 steps. Incidentally – this same equation gives an indication of the depth of focus, using n= ±0.25 wavelengths as the criteria. In the above example, the equation simplifies to:

For the RCT used in the example above, the depth of focus is about 140 µm, or about ±15 steps.

fig.25 This series shows a bright central star that is progressively de-focused. In this case, the focuser was moved outwards by 100 ticks between each image. For this 10-inch f/8 RCT, 100 ticks represents about 1.3 aberration wavelengths. The dark hole is a feature of the central obstruction of the secondary mirror and as the star energy is spread over a wider number of pixels, the overall intensity reduces. As you move further out, it is easier to see the slight on-axis coma. The ring is slightly fatter and dimmer in the lower left hand quadrant. The push adjuster nearest this position was screwed in by about 1/8th turn in this case. Typical evaluations take place for n = 5–8 aberration wavelengths.

Testing Times

It pays dividends to be patient and do star-testing when the conditions are optimum. At night, choose a high-altitude star, with the RCT pointing up, to avoid mirror shift issues and to minimize air turbulence. Use a red filter too, as long wavelengths are refracted less by the atmosphere (and hence by turbulence). For artificial stars, ground turbulence in daylight hours is best in the early morning and over grass rather than concrete. I have had success for a short time around sunset too, before night-time cooling accelerates ground turbulence.

Star-Testing Exposure Time

An optimum exposure renders the star annulus without distortion (from atmospheric seeing or tracking errors) and produces bright mid-tones, so any variations in illumination are easy to evaluate. My Paramount MX tracks well without autoguiding (PE is about 1” peak to peak) and I found exposures between 10 and 30 seconds were optimum. Over that time any effects of seeing conditions are averaged out, just as with autoguiding exposures. For exposures under a few seconds, non-ideal seeing conditions confuse the interpretation (which is easily confirmed by comparing repeated exposures). Some texts suggest to use Polaris (with an apparent magnitude 2), since it is immune to tracking issues. I find this to be too bright and, to avoid saturation, requires a 0.1-second exposure. Apart from seeing issues, not all CCDs can achieve short exposures, especially those with mechanical shutters. I suggest to accurately polar align, to avoid drift, locate a loose cluster with magnitude 5–6 stars and then one can image comfortably with 5- to 20-second exposures, without resorting to a severe screen-stretch to enhance their visibility.

Fine Primary Adjustment (18)

To correct on-axis coma, make tiny adjustments (typically 1/8–1/32 turn) to the primary-mirror adjusters to even the illumination and thickness of the annulus of the de-focused star. This unevenness is caused by residual coma that causes the diffractions rings to lose concentricity (fig.26). On my RCT, the three primary adjusters are in the 12, 4 and 8 o’clock position. I alter the 4 and 8 o’clock adjusters, in pairs. In that way I think of them logically as up/down and left/right controls; up/down by tightening or loosening them equally and left/right by loosening one and tightening the other in equal measure. There are other logical methods, but for me, I find thinking in an orthogonal sense is easier to relate to the CCD image.

fig.26 On the left is a central star with bad coma. Its annulus is narrower and brighter on the left and the primary mirror requires pushing on the right and pulling on the left in equal measure. The middle image is an improvement. The image on the right is cropped from the center of a full-frame image of a loose cluster, de-focused by 10 aberration wavelengths, showing good out-of-focus symmetry.

fig.27 These three figures show a balanced set of aberrations, since all of the de-focused stars are radially symmetrical around the middle of the image. Deep Sky Instruments, in their own collimation instructions for their RCTs, explain the process very well and usefully classify the elongated star shapes: A “pointy” star has its major axis pointing towards the image center and a “flat” star’s major axis is tangential. The figure on the left shows under-correction (pointy stars) which is often an indication of the mirrors being too far apart. The figure in the middle is a perfect scenario and the one on the right shows over-correction (flat stars), normally as a result of the mirror separation being too small. There is always some field curvature with an RCT and the likely outcome is mild under-correction in the corners of the image. For larger sensors, a field flattener is required. The trick with collimation is to identify the dominant ‘pointy’ stars and make secondary mirror adjustments to make them flatter, so that the degree of elongation is equal at all points in a circle, and at the same time, radially symmetrical.

To save time, the first thing one should do is to understand the relationship of the adjusters to the image, which is made easier from one session to the next if the camera angle is consistent. In my setup, with the off-axis guider port uppermost, I push the mirror on the “dim” side, to make the annulus thinner and brighter. I also follow an adjustment regime: The push- and pull-screws have different pitches, so I always use the push (grub) screw, with its finer thread, to set the primary mirror position and the coarser pull-screw to secure the mirror. This either means I back off the push-screw first and then tighten the pull screw, or in the case of pushing the mirror out, say by 1/4 turn, back off the pull-screw by half a turn, tighten the push-screw by 1/4 turn and then tighten the pull-screw to secure the mirror. This helps to achieve a consistent tension and is especially useful if you are trying to undo a prior change. Even so judging precise adjustments is tricky and if it helps, mark the wrench position with some sticky tape next for the two primary adjusters on the back-plate. A change to the primary mirror tilt shifts the star position on the sensor and, for an accurate assessment, it needs to be close to the center. Re-center the star and repeat the process until you have an evenly lit and symmetrical annulus. TheSkyX and SGP have useful center commands to accomplish that in under a minute.

Since the two mirrors interact, it is a first-order simplification to suggest that each mirror only affects either on-axis or off axis aberrations. If the bench alignment was not accurate, the primary mirror had shifted laterally or the secondary was a long way off, it is difficult to achieve circular disks in the center of the image, or they may appear almost circular but with the Poisson spot (Argo spot) off-center. A central spot is another indicator that the primary mirror is in its optimum position. It may take a few iterations of adjusting primary and secondary mirrors to converge on an optimum adjustment. (In a perfectly collimated system, you might also discern the Airy disk of a focused bright star.)

Fine Secondary Adjustment (19–20)

Once this central star looks even, it is time to move onto tuning the secondary mirror. If the primary mirror is centered on the focuser assembly, a laser alignment of the secondary is often close to optimum and needs the tiniest of adjustment to achieve a good balance of aberrations around the image center. Good balance may not necessarily result in perfectly round donuts all over; looking at fig.27, one can see, dependent upon mirror separation, that the outer stars may be rendered as oblongs, indicating astigmatism. In fact, a RCT has field curvature and the star donuts will be slightly elongated towards the corners.

fig.28 These three figures show various unbalanced aberrations, requiring secondary mirror adjustment. The image on the left shows distinct elongation (pointy stars) in the top left corner, requiring adjustment, the middle image is the same, in this case the pointy star is opposite the flat star in the bottom right, requiring a lesser adjustment to balance the aberrations in the same direction. The image on the right is almost there, with less distinct orientations requiring an even smaller adjustment.

The trick is to make this elongation radially symmetrical about the center. (This is quite difficult to judge and I found myself increasingly relying upon the hall of mirrors test to adjust the secondary to the new primary position and then confirming collimation with a star test of a loose cluster.

These final adjustments need a steady hand and preferably a T-handled Allen key or hex wrench to apply small and precise movements. (If you are concerned about dropping a wrench onto the primary mirror, why not attach it to a wrist strap?) The best seeing conditions are near the Zenith and if you do not relish the prospect of balancing on a ladder in the dark, choose the highest altitude setting that still allows you to reach the secondary adjusters with one’s feet on terra firma.

The process begins with an image of some bright stars; a loose cluster or the 9-ball tester is ideal. This image uses the same focus position used for the primary adjustment. The stars donuts will likely be a range of circles and oblongs as you scan around the image. The trick is to identify where they are most “pointy”. This term is coined by Deep Sky Instruments in their on-line RCT collimation guideline. They define pointy stars as those whose major (long) axis point towards the image center, or in cases where there are none, opposite those oblongs stars whose minor (short) axis is pointing towards the center (fig.27), which they term “flat” stars. Again, in my setup, the adjustment convention is to “pull” or loosen the secondary bolt closest to the pointy star (as viewed from the rear of the scope) but please realize your orientation may differ. Having made a tiny adjustment, re-assess a new image and continue to tune out the imbalances. If the central stars show signs of elongation, repeat the primary adjustment process and then check the image balance once more. As you approach collimation, the differences are subtle and may require several image downloads to be sure of the necessary (if any) adjustment. The example image in fig.29 is almost perfectly balanced. There are some slight asymmetries, that suggest a need to loosen the top adjuster a fraction, and tighten the other two.

Image Scale and Mirror Separation (21–23)

Mirror separation is something that is often taken for granted. One authoritative text evaluated mirror separation by comparing Strehl ratios at different mirror separations. That measurement is an overall assessment and not the most critical assessment of stars in the image periphery. The distance, and hence the focal length, has a big effect on aberration on stars around the periphery. Thankfully, plate-solving not only returns the image center coordinates, but also the pixel scale, to three decimal places. If the pixel size is known too, an accurate assessment of the effective focal length is a short equation away:

fig.34 This shows what to look for when balancing aberrations with the GoldFocus system. Each of the red squares represents a GoldFocus readout as in fig.33, for each of four star positions, placed at 120° intervals (conveniently using a mount’s jog commands from a center position).

After initial star testing, I had what seemed to be poor field curvature (fig.35). The initial star test confirmed an image scale of 0.564”/pixel without binning, which with a 5.4 µm pixel size indicated a focal length of 1975 mm. The focal length increases with reduced mirror separation and the estimated separation error was about 2.5 mm too far apart. The stars in the image corners were obviously elongated along a radial axis, indicating under-correction. As mirror separation is reduced, the degree of spherical aberration correction is increased. The configuration is also very sensitive; when the mirror separation reduces by 1 mm the focal length increases by about 9 mm.

In theory one can move either mirror; but it is preferable to adjust the secondary position. (Extending the primary mirror fixing bolts makes it more vulnerable to lateral forces.) I unscrewed the secondary lock ring and baffle by about 3.5 turns and then gently tightened up the lock ring (fig.15). The flange between the lock ring and the secondary baffle facilitated an exact measurement using a Vernier caliper. After two successive adjustments, a plate-solve confirmed an image scale of 0.577”/pixel equating to a focal length of 2,000 mm (as per the optical design). This has a considerably flatter field, indicated on the right-hand image of fig.35. Remember, a RCT design has a curved field and some degree of of axis astigmatism is to be expected. My 10-inch RCT has acceptable field curvature when used with an APS-C sensor but it is advisable to use a field flattener for high quality images when imaging onto larger sensors.

fig.35 These CCDlnspector plots show the field curvature of my RCT before and after an initial alignment. A test of the image scale revealed the focal length was 1,975 mm. The image on the right shows the field curvature after the RCT was adjusted to a focal length of 2,000 mm. The image has better correction but the RC design inherently has some field curvature.

fig.36 This chapter deliberately evaluates many alternative techniques, some of which overlap on purpose. To summarize, this table outlines my experience of their practical capability and how they may potentially be combined. The popularity of affordable RCTs will encourage further developments over the years and application of current SCT collimating products to RCTs. In these pages 1 have avoided any methods that remove the secondary mirror altogether and shine a laser through the central fixing hole; it potentially voids the manufacturer’s warranty and is also too intrusive if one is only wanting to update an otherwise roughly collimated scope. Clearly one does not have to use all procedures to collimate a scope but choose a coarse (or medium) and fine adjustment process from each column. Some limitations are a function of the RCT’s mechanical tolerances. My preferred methods for collimating my RCT are highlighted with a red border. There are many more combinations that are equally valid.