The reason for this research was the collimation of an Intes MK-72 Maksutov telescope that suffered from a mirror-flop. After the mirror was secured and the telescope rebuilt, collimating the telescope was the next stage. As with a correctly constructed RC, the focuser of the Intes MK-72 is not attached to the main mirror, but to the housing. This means that the focuser, main mirror and secondary mirror each have to be adjusted in turn. To collimate the main mirror, the circular housing of the camera is used to see if it appears concentric with the line through the centre of the focuser and the mounting hole for the secondary mirror in the spider (for more details see the white paper).

The RC-collimation-procedure failed to get the Maksutov-telescope right, which seemed to be caused by the fact that the imaging-chip in the ASI290MC camera used was not in the centre of the housing. An available ASI120MM also appeared to have a deviation, but smaller than that of the ASI290MC and the difference between the two cameras seemed to be reflected in the collimation, with the smaller deviation appearing to lead to a better result. When collimating the main mirror with the above method it is crucial that the image chip is in the centre of the housing, after all the reflection of the circular housing is used when adjusting the main mirror. Therefore we decided to use various ZWO cameras to see what the actual deviation of the image chip is from the housing's centre.

We placed all five cameras in the same way in a test setup, consisting of a collimator with an artificial star and a telescope that holds the camera (see figure 1). A 25cm f/5 newton telescope was used as a collimator (dark blue telescope in the image) with an artificial star made of a 9µm fiber-optic cable (yellow cable in the image) in the focuser. The collimator is set to infinity, creating an artificial star that appears to be at infinity. The telescope holding the camera is a Skywatcher Evostar 72ED (black and white telescope in the image), which is mounted on an alt-azi adjustable table in front of the collimator.

The procedure for testing the cameras is quite simple: the camera is placed and the alt/azi table on which the Evostar sits is adjusted until the artificial star is in the centre of the imaging chip, something that can be easily verified by means of the digital cross-hairs of the camera software ASICAP (see figure 2).

The camera is then rotated 180 degrees along the optical axis, after which it is observed where the artificial star appears (see figure 3). With the camera in this example it is immediately clear that the chip is not centred, but the procedure has been expanded considerably to measure this properly.

Since it is likely that removing and replacing the camera will already lead to deviations, more measurements were needed. For this purpose, the camera is removed and replaced again ten times in the first position (i.e. in which the artificial star is centred), each time taking a screen-shot so that the deviations can be measured (this is measurement A). These screenshots were taken at 100% zoom, so that every pixel of the camera corresponds to every pixel on the screen (we used screenshots, because that's the only way to store the digital cross-hairs as well). The same is done in the second position, i.e. where it becomes visible that the chip is not centred (measurement B), and finally the camera was mounted again ten times in the first position to see if the entire setup was not disturbed in between (measurement C).

The cameras were placed in a 2” compression ring adapter and provided with a mark to ensure that the camera is always in the same rotational position in the telescope when removing and re-inserting it (the ZWO cameras all have a 2” rim, so that the cameras can also be used without a 1.25″ nose-cone). When replacing, attention was also paid to the same order and force of tightening the clamping screws of the 2″ adapter.

Per camera, 30 measurements were made in this way, resulting in the following parameters:

• the repositionability of the camera (mean standard deviation of the individual measurements A, B and C);
• the stability of the arrangement, referred to below as Closing Error A-C (difference between the means of measurements A and C );
• the centring error of the image chip (half the difference between the means of the combined measurements A+C and measurement B);
• the accuracy with which the centring error is determined (determined from the propagation law of the variances from measurements A, B and C).

Since the centre of the artificial star in the screenshots had to be determined manually, it cannot be ruled out that another human error could creep in here. To minimize the chance of this, the measurements have been processed by both authors and the average of these has been used in the calculations. The measurements resulted in the following numerical values ​​(the measurements were made in pixels and converted to micrometers using the known pixel sizes of the cameras):

The above table clearly shows that the measured centring errors are significantly larger than the closing errors and standard deviations (at 1σ, 68% confidence level) and that they are therefore real. The average centring error is 222μm (0.222 millimetres) and can therefore be called considerable. Figure 4 shows the centring errors graphically, with the red dots scaled proportionally to the standard deviations of the measurements. The image was shared with the manufacturer and their response revealed that the anomalies shown are not uncommon.

As we have seen above, the average centring error is about a quarter of a millimetre. However, it was expected that this could not have any influence. This is because collimating the main mirror aligns it with the mechanical viewing axis by aligning a 2 millimetres diameter light source in the visual-back with a reflective ring with a 7 millimetres gap in the spider. This last hole is visible on the screen during adjustment as a dark circle with a diameter of approximately 18 pixels. Each pixel therefore represents approximately 7/18 = 0.4 millimetres and since the spider is approximately one third of the distance between the main mirror and the camera, the centring error of the image chip at the height of the spider would be well under one tenth of a millimetre (on average 0.222/3 millimetres for the examined cameras). Since this is only a quarter of the 0.4 millimetre per pixel image resolution, this can hardly be observed.

However, when collimating the Intes' main mirror there is no spider, so an artificial spider was made using 0.5 millimetre nylon thread threaded through the screw holes for attaching the corrector plate (the corrector plate was removed during this process, see figure 5). Visually, the three wires seemed to intersect neatly in one point, so this should represent the centre of the aperture. Now the three wires are 0.5 millimetres thick and the spider is again about a third of the distance between the mirror and the camera, here too the centring error should be barely noticeable.

Nevertheless, both authors thought they could see that there was a difference between the two cameras used. When adjusting the main mirror, it is not only checked whether the camera housing can be seen centred in relation to the spider, but above all whether the spider coincides well with its reflection. If the latter, it seemed that the camera housing was shown slightly off-centre (see figure 6). And so we decided to see if we could better centre the image chip in the camera housing.

We still had to come up with a procedure for centring the image chip. In itself, the construction of the camera allows perfect centring. The image chip is in fact on a separate print that is attached to the motherboard with four screws (see figure 7). By loosening the four screws just enough to be able to move the printed circuit board, while still providing enough resistance to prevent the printed circuit from springing back, we can try to centre the image chip.

But what is the centre? We first tested on the lathe whether the housing is round and also whether the lid where the 2″ connection is located is centred in relation to the camera housing. With the first camera we tested, this turned out to be within 0.01 millimetres. Later tests with one of the other cameras showed that this is not always the case and that the 2″ connection can show 0.1 millimetre eccentricity in relation to the housing.

The first attempt to centre the imaging chip was visual (see figure 8). The camera was mounted open in the lathe and a second camera, fitted with a lens, was set up opposite it. The second camera was positioned so that its cross-hairs coincided neatly with the central gray area of ​​the imaging chip. By turning the lathe's chuck 180 degrees, the eccentricity was made visible. Subsequently, the recording camera was moved halfway through the displayed deviation and then the print with the image chip was moved into the centre of the reticle. By now turning the claw 180 degrees again, it is clear whether the image chip is now centred (see figure 9). However, the method turned out to be not accurate enough, perhaps because not the entire visual plane is actually used for image production.

The method that does work is by means of a laser. The camera is again mounted in the lathe and the centring in the chuck is checked. The camera is connected and then a stable setup is made with a laser, which illuminates exactly the centre of the digital cross-hairs. By turning the chuck of the lathe again, it becomes clear what the centring error is. The laser can then be moved half the error again and the image chip is moved the remaining distance. After turning the claw again, it becomes clear what the remaining centring error is and the procedure can be repeated until the centring error is acceptably small.

Initially, the laser was aimed directly at the imaging chip, but the resulting spot was too large to work with properly (see figure 10). The spot can be made smaller by bringing the laser to a focus point with the aid of a lens. The solution was found by combining a laser collimator for Newtonian telescopes with a Skywatcher finder-scope and mounting it on a cross stage (see figure 11). The two could easily be combined because the Skywatcher finder-scope is provided with a screw thread with a pitch of 0.75 millimetres and a diameter of 50.3 millimetres, so slightly smaller than the 2″ (50.8 millimetres) outer diameter of the laser collimator. The latter was therefore turned off by half a millimetre and provided with 0.75 millimetre thread.

With this laser setup it is now possible to centre the image chip (see figure 12). However, the check, as discussed at the beginning of this article, showed that after centring there was still a residual error of just under 0.1 millimetre. The cause turned out to be the aforementioned eccentricity of the 2″ connection. After all, the centring was done on the housing, while testing was done on the 2″ connection. The eccentricity of the 2″ connection could be easily determined with a dial indicator on the lathe and corresponds well with the control measurement for the collimator. We can therefore assume that the centring has been successful. When using the camera to collimate primary mirrors, this deviation is of no concern, since it is the outside of the housing that is used for collimation and we have aligned the image chip with it.

The last flaw turned out to be in the C-mount lens. The lens is attached to the camera with a T2 to C-mount adapter, but these adapters are not properly centred. Figure 13 shows the effect of turning this adapter around (i.e. the adapter is removed and replaced in reversed direction). In these images, the camera was mounted in the chuck of the lathe, so it could not move. The ZWO zoom lens was mounted and a picture was taken. Then the lens was disassembled and the T2 to C-mount adapter reversed. After mounting the lens, the second image was taken. The same test was repeated with 25mm and 35mm C-mount lenses with similar results.

However, the shift of the image has virtually no effect on the collimation of the primary mirror of an RC to be collimated. The camera must be rotated slightly in alt/azi in order to get the optical axis of the telescope or lens in the centre of the image chip. This rotation will distort the image of the in the primary mirror reflected camera body slightly into an ellipse, but so little that it will most likely not be visible.

For applications where the concentricity of the image chip to the camera housing is critical, it pays to check the eccentricity of the image chip. However, it should also be checked whether the 2″ connection is concentric with the camera housing, as this is not always the case. If the image chip is indeed not mounted in the centre, then this can best be adjusted by mounting the camera in a lathe and determining the centre with a focused laser. If any mounted lens is mounted non-concentrically, this will ensure that the camera only needs to be tilted slightly during use. As long as the deviation is small, it will not affect the concentric image the camera provides.