Backfocus and tilt

Understanding the issue of backfocus and tilt is critical to being able to take good astromonics images. In this article I would like to discuss the technical issues that occur when working with large sensors and Celestron RASA 8 telescopes. Of course, much of the information contained here is true in general and applies to all optics.

First, let me briefly remind what backfocus and tilt are.

In the most general terms, backfocus is the correct distance of the sensor from the housing. The value of the backfocus is given in its specifications by both the manufacturer of the telescope and the camera.

• In the case of a telescope, it will be the distance at which the camera sensor should be placed from the plane of the telescope to which subsequent optical elements are mounted (connectors, camera, etc.). When I mean RASA telescope, it will be the distance from the metal plate of the front part of the telescope; the plane with the M87 side thread on which the accessories are screwed. A significant number of telescopes have a backfocus of 55 mm, but the RASA 8 has a backfocus of 28.73 mm. However, the larger version of this telescope, the RASA 11 has a backfocus of 72.80 mm. In comparison, the EdgeHD 1100 telescope’s backfocus is as long as 146.05 mm.

• For a camera, backfocus is the distance of the sensor from the front plane of its housing. For example, cameras with smaller sensors of 4/3″ or 1″ format (e.g. ZWO ASI 1600, ASI 294 or ASI 183) are usually characterized by a 6.5mm bacfocus. In the case of APS-C or full-frame astrophotographic cameras, the backfocus is typically 17.5mm. Some designs (such as the ZWO ASI 2600, ASI 2400 or ASI 6400) have a detachable tilt plate and after taking it off and revealing the housing the backfocus is reduced to 12.5 mm. Then the sensor distance is no longer counted from the outer plane of the tilt plate but from the outer plane of the housing itself. It is also worth adding that in the terminology used for classic cameras (DSLR or mirrorless), backfocus is called flange focal distance. A list of values for different models can be found here.

Placing the sensor exactly at the backfocus point is extremely important: there, and only there, does the manufacturer guarantee the correct image geometry: the best star shape in the entire frame, the least distortion, optimum sharpening, etc. Depending on how the telescope’s optical system is designed, we can approach backfocus with different tolerances. In general, the greater the backfocus length, the greater the tolerance. In many refractors, SCTs, or MAKs, the tolerances are quite large, but in RASA 8, with backfocus 28.73 mm. This value, specified by the manufacturer, shows that, theoretically, we are interested in accuracy to one hundredth of a millimeter. In practice, however, in most cases (depending, among other things, on seeing), a difference in effect within a tolerance of 0.5 mm is generally acceptable.

Of course, the closer we are to the value specified by the telescope manufacturer, the better the effect we will achieve.

When we go beyond the tolerance range, the shape of the stars will start to change. Optical distortion will appear, the stronger the further to one side from the optical axis. In addition, this distortion increases with distance from the center in a non-linear fashion. This is because the optical focus surface of a telescope is not flat, but curved; it is closest to the plane precisely at the backfocus point. If we move away from there, then although we can focus at the center by turning the adjustment knob, the closer we look to the edge of the frame the bigger the problem becomes. Alberto Ibañez has posted an excellent article on his blog, explaining in it the relationship of tilt to the curvature of the plane of focus.

Difficulties with backfocus and tilt in RASA 8

In case of the RASA 8 with f/2, the manufacturer’s documentation lists the backfocus value as 28.73 mm. Many cameras with APS-C and full-frame sensors have a backfocus of 17.5 mm. This means that between the front of the telescope and the front housing of the camera we can use a connector that is exactly 11.23 mm thick. At least a couple of significant problems arise here:

• 11.23 mm is not much, especially if we want to put a filter drawer in there. The filter itself is about 7-8 mm thick, so we don’t have much space for any housing, let alone a decent camera attachment for the telescope.
• Since RASA 8 backfocus is very strict, it may be even enough to tighten the thread too much or too little for optical defects to appear. The problem gets bigger the more threads we have to tighten: camera to camera adapter, telescope adapter, etc. If we add to this the number of connections in the intermediate elements and the number of components forming subsequent connection planes, we have a recipe for a difficult problem to solve, especially with larger sensors. For example, a significant proportion of filter drawers are manufactured from three independent elements: front, rear, and center.These elements are bolted together and additionally screwed to the rest using central threads. This number of connections results in deviations from the ideal geometry that are difficult to predict. In most optical sets, this is not very important, but in RASA, unfortunately, it usually is.
• Different filters have different optical densities. The value of 28.73 mm is the backfocus for RASA, assuming a 2 mm thick narrowband filter is used. For broadband filters, the backfocus will be different, because the light travels in a completely different optical path, encountering glass at the end that changes the final shape of the plane of focus in different ways. For this reason, we should use different filters and change the distance between the camera and the telescope accordingly, at the same time taking care to keep the sensor parallel and not to introduce tilt.
• Cameras with large sensors do have special plates for tilting, but they are very problematic in use and minimal movements of the adjusting screw cause gigantic changes. Moreover, after taking the camera off and putting it on again, it turns out that this time the tilt looks a bit different. To make the adjustment again, you have to take the camera off, but after putting it on again, you often do not get back to the previous state and the problem loops.
• Such cameras are also heavy, often weighing even a kilogram. The moment of force resulting from gravity pulls the camera downwards, which can cause it to micro-fall. Then tilt appears, its characteristic feature is that it causes distortions in different parts of the image, depending on which part of the sky we point the telescope at. This translates into the problem of which edge of the camera sensor is actually pointing down.
• The sensor should be placed perfectly centrally. Unfortunately, some camera adapters for RASA telescopes have a metal threded connector which causes the camera not to be mounted exactly centrally but to be moved to the side. The result is an effect similar to that caused by tilt or bad collimation. In the case of such play there is also a danger of the camera sliding down during a long shooting session.
• Finally, there is also the effect of thermal expansion, which is difficult to estimate. For example, the coefficient of expansion of steel is 12 microns per meter per degree Kelvin. With a 2 cm thick metal component and a temperature difference of 30 degrees, we have about 7 microns from metal expansion alone. For aluminum (for 6063 alloy), it’s 2 times more, so about 25 microns per meter per Kelvin. In the case of plastics, it’s 5 times as much, about 60 microns per meter per Kelvin. Now let’s take a camera whose casing is made of aluminum, the interior of which is made of copper and plastics, it’s hot from work and the part near the sensor is frozen e.g. to – 15 degrees. And maybe someone still has a dew-heater turned on at the front? It’s hard to estimate, but it’s quite likely, that such a kind of temperature differences can cause changes in the dimensions of the parts holding the sensor in the right position.

Problems arising in practice

All of the above issues are reflected in constantly recurring posts on discourse forums and social networks. People are fighting with the problem of tilt and backfocus in different ways, looking for the reasons in their own mistakes, problems with adapters, cameras, telescope, incorrect collimation, distorsions of the mirror, corrector and many other issues. Even a cursory reading of these discussions shows that the problem does exist and is serious, with some people having stronger and others weaker symptoms. Of course, there are also people who don’t notice the problem – they may be very lucky, or simply the pursuit of optical perfection is not the most important thing for them.

Possible solutions

RASA 8 and the large sensor cameras form a system with such a small tolerance, that in practice it is difficult to devise a technical solution that would permanently remove all difficulties. It seems that the only such way would be to weld the camera permanently in the right position with a predetermined, single permanently mounted filter. This is because every element that is subject to adjustability here is a potential source of a present or future optical problem. Every assembly or disassembly of the system may introduce new disturbances and cause problems again. Rather, one must become accustomed to the fact that although one will strive to obtain the best possible image, various difficulties will always arise to some degree. It’s important to understand that the goal of correction should be a better picture, but not necessarily a perfect picture. Such realistic expectations will allow us to work more calmly with astrophotography.

Different people have different ways of dealing with these types of problems. An individual approach is one of the characteristics of the beautiful art of astrophotography. When creating Astrodevice technical solutions, I was guided by my own approach, which I would like to share here.

I believe that, first and foremost, you need to have efficient equipment. You need to make sure that the camera and telescope do not have any optical defects. If they are damaged, they should be repaired. You also need to be sure that the mount is working well, the mechanism is accurate, the whole thing is well balanced, and you have good guiding. If the equipment is in good condition, every effort should be made to ensure that the connection between the telescope and the camera is as perfect as possible: it should have the ideal optical length, as few connections as possible, and mounting planes that are as parallel as possible. This should ensure a very good result.

Finally, it is worth considering software correction of the final image using a tool such as BlurXTerminator, which does an excellent job of repairing any minor optical distortions that may remain in the image.

Astrodevice filter drawers

The filter drawers were designed in such a way that their possible adjustment could at least partially remedy the above problems. I have used several important design solutions in them:

• The drawers are bolted to the camera housing with several screws rather than a single thread. To make it possible, you have to first unscrew the original tilt plate from the camera. By doing so, first of all, we gain another 5 mm of space (this is the thickness of the tilt plate). As a result we have 16.23 mm of manageable distance, which is already enough to create the whole mechanism of the filter drawer with the mounting. Secondly, we get a rigid connection in which there is no room for differences in the rotation of the threaded connection.
• Previous commercial versions of Astrodevice drawers featured a calibration screws system that ensured their front and rear planes were perfectly parallel. In the current model designed for self-printing, the geometry and printing method are selected to maintain the maximum accuracy that can be achieved in the Z axis in 3D printing. The entire main module of the drawer, its core, containing the front and rear mounting planes, is printed as a single solid. So if you have a well-calibrated printer and follow the printing instructions, your drawer should have perfectly parallel planes and a thickness of exactly 16.2 mm. After adding a camera, which has a backfocus of 12.5 mm without a tilt plate, you will get exactly 28.70 mm, which successfully falls within the tolerance limits of the correct value for RASA 8.
• The drawers of the M87 series have their own built-in nut which makes them exactly centred. The nut also makes it the only moving element. The only free degree of freedom is the clamping force on the telescope.

What's interesting, Celestron's calculated backfocus of 28.73 mm works very well with narrow band filters (S,H,O and more complex filters such as NBZ or L-Extreme). For some people, at this distance, the backfocus (and so the shape of stars) can become problematic for R, G, B filters whose optical length is slightly shorter. Similarly you should keep in mind that filters of different thickness have different effective optical lengths.

Some time ago, there was a discussion about this on Cloudy Nights. Based on the telescope manufacturer’s answers, in the summary one user gave such data:

adding, that

As you can see, the difference between no filter and the narrowband filter is 1.46 mm.
This is a very large spread for the ability to micro-adjust, with the requirement that the drawer planes remain perfectly parallel. The drawers I’ve designed use every available millimeter to accommodate their mechanism while maintaining proper rigidity. In order to shorten the optical path, we would have to make the drawer thinner and ask the user to fine-tune it every time they change filters. Not only would this be absurdly problematic, but even if someone insisted, we would still have the problem of making a precise enough adjustment: in most cases these experiments would lead users to frustration, would uncalibrate the drawer, and would introduce false tilt. It would be enough for the user not to know exactly what drawer thickness they wanted to set for a given filter model (to the nearest hundredth of a millimeter) and they would no longer know whether to attribute the bad effect to incorrect distance, tilt, their own mistake, drawer design, or something else. Excessive adjustability would also introduce a lot more false light, which you would have to fight with a side cover. All of this leads us to conclude, for the time being, that it would be better to create a drawer adjusted to the single distance recommended by the telescope manufacturer, rather than a breakneck design with too many degrees of freedom.

Concluding remarks

Much of the problems that users attribute to backfocus and tilt disappear if you take care of the following:

The mounting planes should be as paralell as possible. This is very, very important. My tests of drawers of some manufacturers show that the dimensional inaccuracy measured at various points on the perimeter of the plane where the drawer adheres to the camera is even 0.4 mm (!). If someone screws such a drawer to their RASA, they will probably have a significant problem with backfocus and tilt. For this reason, the drawers I have designed have adjustment screws to be set that provide precise distance correction and compensate for natural manufacturing deviations.

A number of backfocus problems arise at the threads. A little lighter or a little stronger tightening and we already have a difference of a few hundredths of a millimeter. With several threads the problem may add up to tenths of a millimeter. That is why I developed a drawer with a built-in M87 screw cap: it should have the near-perfect dimension and there is no connecting element along the way which would introduce length distortion.

Another problem is the force with which we tighten the system to the telescope. This is actually part of the problem described above. That’s why I suggest that the drawer should be tightened sufficiently but not forcefully.

The next problem is gravitational pull. As mentioned earlier, large cameras, such as the ASI 2600, weigh almost 1 kg. When we mount the whole thing on short M42 threads, we may have to deal with micro-distortion. Its characteristic feature is that the image distortion changes depending on the point on the sky where we point the telescope. This happens because as the position of the telescope changes, a different part of the sensor is directed towards the physical bottom. Having to deal with this problem, users often tighten threads more firmly, but this in turn starts to cause problems described above. For this reason I think that, if possible, a good solution for ensuring the best possible image quality is to use a drawer whose only degree of freedom is to be tightened to the telescope, i.e. a drawer with a large M87 thread, which potentially can better cope with possible micro-drops. Of course, this does not mean that accessories mounted with smaller threads are bad – just that a larger thread is always stronger. Additionally, in the M87 drawer the section which in other cases is screwed on M42 thread, here is structurally completely rigid.

All of the above problems can occur even with a fully functional telescope and camera. In fact, we should state this assumption at the outset: we can only attempt to solve problems when the telescope and camera are in perfect condition. Unfortunately, sometimes a telescope exhibits astigmatism, the effect of which can easily be confused with tilt. The main cause of astigmatism in telescopes such as RASA is the curvature of the primary mirror – as a result of shock, the mechanical part that keeps the mirror in the correct plane may shift, or the focus adjustment mechanism may malfunction, resulting in optical distortion when pushing and pulling the mirror. In the case of heavier mirrors in other telescopes, this phenomenon is often caused by the gravitational fall of the primary mirror (known as mirror flop). A characteristic feature of astigmatism is that the image of the stars is distorted linearly rather than concentrically, as in the case of tilt. When you look at the photo, in the case of tilt, the center is good, the stars are point-like there. The closer to the edge, the more stretched the stars are (towards the center), most obviously in one of the corners. In the case of astigmatism, the stars in the center are also stretched and therefore the direction does not lead to the center of the frame at all. If you think about it for a moment, you will probably come to the conclusion that the same effect can be caused by poor telescope guidance - the stars may simply be stretched. That is why diagnosing distortion is usually a long and tedious problem that requires many tests.

So before we can talk about the drawer, backfocus, and possible tilt, we must be absolutely sure that the telescope, camera, and balance of the mount and guiding are absolutely flawless. If they are, and imperfections still exist, many backfocus or tilt problems can be easily solved by keeping a few things in mind:

• If you are using a narrowband filter, you should remove the glass window from the telescope. It changes the effective length of the optical path.
• If you are using the original Celestron M42 adapter and you have a felt mounted under it, you should consider detaching it. Felt introduces extra backfocus and is also soft and does not align evenly to hundredths of a millimeter. Thus, it may introduce tilt.
• Some original M42 metal nuts supplied with the telescope do not have a centering ring. Make sure that when you mount the camera using the original accessories supplied by Celestron, the camera is perfectly in the optical axis of the telescope. If you can move it from side to side – you have a problem. To avoid camera centering issues, I developed a drawer with a built-in nut in the M87 version. Thanks to it the camera is mounted perfectly centrally. Misalignment introduces a number of optical distortions that users may mistakenly attribute to bad backfocus, tilt or bad collimation.
• In the original M87 mount nuts that came with the telescope, there is a plastic clamping ring. It is elastic and its use can potentially lead to problems with camera micro-sagging. It can be removed from the original nut in effect tightening metal to metal without any soft intermediaries.

In summary, there are many mounting issues that can contribute to optical distortions. Users often try to deal with them by inserting shims, adjusting tilt or, as a last resort, re-collimating the telescope. Very often it doesn’t give good results, it leads to frustration and the real problem lies somewhere else. Often, it is neither the telescope nor the camera, but the element connecting the two and the way in which this connection is implemented.