A new affliction has recently washed over the world of astro-imaging—namely: sensor tilt. Everywhere I look, sensor tilt is either screwing up an imaging system or it’s being blamed for screwing up an imaging system. And that has led to fixes such as the Octopi camera interface adapter that allow fine adjustment to both sensor spacing and tilt. Faster optical systems and cameras with smaller pixels both work to tighten the requirements on camera tilt and position but to a large extent, the difficulty of aligning cameras has been exacerbated by the really poor mechanical design of many of the components that need to be stacked up in the image train. Poor design is an issue that has caused me to tear my hair out in frustration so many times that I thought I’d post a brief review of common problems along with a very short introduction to optomechanical design principles. My background in optics is primarily in interferometry and optical testing, but I’ve had a serious interest in optomechanics for as long as I can remember. To be clear, the field of optomechanics addresses the specific design challenges associated with optical systems. It involves understanding how to analyze and mount optical components to minimize strain and mechanical distortion, how to achieve and maintain precision alignment, how to select proper materials, how to tolerance mechanical requirements, and how to deal with thermal effects. Lying at the core of all of this, are the basic principles of precision positioning and from what I’ve observed, this is where nearly every (not quite all!) astro-equipment manufacturer falls apart.
Let’s look at just a few examples of poor design that have driven me crazy. Our first example comes from Planewave Instruments. The photos below show the end views of the main dovetail/carriage adapter that they supplied with my 20” telescope back in early 2020. They've shipped a lot of these things and for the life of me, I can’t understand why anyone thinks that it’s acceptable. The mechanics are so faulty that the entire weight of the telescope is held between a couple of pinch points in the carriage and the strength of the connection fall onto the screws themselves. Furthermore, the screws bind so terribly as the dovetail begins to engage that it’s hard to tell when the screws are properly torqued. Grade: D-
As an offer of good will, I suggested the simple design modification shown below to fix the design without having to start completely from scratch. In the end, Planewave declined to fix it so I returned it and replaced the carriage with one from Astro-Physics. The Astro-Physics carriage is beautifully designed and implemented. Designing a well implemented dovetail/carriage is not that hard and in spite of that fact, Planewave isn’t the only company that has had problems pulling it off. I’ve seen similar issues from Celestron with their orange extruded dovetails as well, but I’ve made the point so we’ll skip those photos.
Next on the list: I’ve got a FLI filter wheel for my F/6.7 refractor and for no extra charge, they included a mechanical problem that is so difficult to fix that I had to replace the entire assembly with another brand. In order to try to minimize the thickness of the filter wheel, the FLI design uses only two setscrews located at 120 degrees to mount a dovetail on the camera side of the wheel. The diagram below shows what happens when the screws are tightened. The two sets screws do what they are supposed to do, which is to draw the dovetail down into the mounting hole, but since there is no opposing screw (or lip) on the other side, the camera gets tilted. No amount of shimming will fix this problem! They would have been better off using just a plain cylindrical surface instead of a dovetail. It’s difficult to solve this problem in a clean way without seriously re-machining some of the parts. Grade: F
I replaced the FLI FW with a new one from Pegasus Astro. The Indigo filter wheel mounts 50 mm round filters, it is very inexpensive and at first blush, it looks like a very simple, if not, elegant design. It is very thin, it runs off of USB, and it allows bolting adapters directly to the filter wheel. Unfortunately, it doesn’t take long to uncover multiple problems with this design. First, they use arcing slotted holes to mount adapters and those holes open directly into the filter cavity, which will cover the filters with dust in no short order! What in the world were they thinking??? Fortunately, they designed bolt on covers to cover the holes and as fast as I could complain, they shipped them to me free of charge from Greece. The next problem rears its head when you mount the filters. The diagram below shows the problem.
The screw and washer retainer system is workable but the way that it’s implemented is just sloppy! In my professional life, I’ve shipped optical equipment that I designed all over the world and the first rule is that any screw that isn’t locked, WILL come loose in shipping. Since I’m sending this FW to Chile, I have to address this issue in a serious way.
The final problem with the Indigo FW involves the friction drive wheel on the edge of the filter wheel. It's the blue wheel in the photo above. Unfortunately, they fixed the position of the friction wheel. It’s adjustable for wear, but that’s only possible if you are there—as in when the FW is used in your backyard. That really doesn’t cut it when you are shipping equipment to a very remote location where service can be difficult and where you certainly don’t want to be opening the FW to expose the filters to dust. The drive wheel should have been pre-loaded against the edge of the filter carrier using a simple flexure spring. Additional cost: Very low. As it is, the mechanical design of this product is not very sophisticated. Grade: C-/D+. The good news is that they could clean up this design fairly easily…if they want to.
Last on my list is an issue that probably affects a lot of QHY camera owners. When I switched filter wheels, I had to use the QHY supplied carriage designed to mate to the V-notch “dovetail” on the front of the camera. I was stunned to see just how poorly constrained this whole system is! First, the carriage is seriously over-sized to allow for centering the camera. Let me get something straight here: The only reason you might want to center the camera is if your mechanics are so screwed up in the first place that you can’t mount the carriage where it is supposed to go! More adjustments almost always makes for more misalignment! With this system, it becomes very hard to tell when your camera is precisely centered, but fortunately, centration is not very critical on most systems. Here’s a diagram of how the camera mounts.
The much, much bigger problem is that QHY designed the V-notch spacing so that the centering screws engage it with the camera hanging in space—referenced to absolutely nothing! The centering screws are mounted in loose fit threaded holes and the tips of the screws aren’t coned or spherical. There is simply nothing that defines the angle of the camera! It might be close enough but more likely, the camera will be tilted a bit, no matter what you do. So, in the end you can’t tell when the camera is centered and you sure as heck can’t tell if it’s tilted either. Furthermore, the spacing from the front surface of the V-notch carriage to the sensor depends on how well the screw hole positions are controlled and on the unknown shape of the screw-tips. This is just a mess. The guys at QHY appear to be experts at electronics but in my opinion, their opto-mechanical skills are seriously flawed. No wonder so many folks are buying Ocupi adjusters! Grade for the QHY V-notch mount: F-
Fortunately, there is an easy fix to better control the tilt of the camera shown in the diagram below. First, a conical tip can be ground onto the tip of the centering screws. Second, by carefully measuring the top gap between the V-tail adapter face and the V-tail carriage (using feeler gauges), a circular washer can be cut from plastic shim stock with a bit of extra thickness (maybe 0.010” to 0.015”) to allow the centering screws to contact the bottom edge of the V-notch to pull the V-notch adapter face tight against the shim washer. Astro-Physics uses a V-notch system on their field flatteners but they’ve implemented it correctly so that the retaining screws only engage one edge of the V-notch pulling the adapter and mount tightly together along a reference surface. The diagram below shows the fix. Alternatively, the holes for the screws could be repositioned using a mill and an index table, but that's a lot more work. I'll do it if I need to move the camera inward. The ultimate fix would be to just redesign that whole front V-notch part to be a straight dove-tail...as it should have been in the first place!
Finally, I’ll give an honorable-mention to Celestron for their utterly poor mechanical design of the secondary mount on their Edge series of SCTs. That design is not well constrained in angle or position and that results in having to realign the secondary every time it is removed or after something as simple as a car trip. Years ago, I completely redesigned the secondary mount on my prototype C14 using kinematic principles and demonstrated on an interferometer that it could hold alignment to within about 5 fringes across the pupil even after pulling the secondary out and remounting it. (The factory version controlled tilt 20-40 times worse!). It had super easy to use, orthogonal tilt adjustment and using CNC, it could be made for only a few more dollars than the existing design. I offered the design to Celestron but it wasn’t worth it to them so they passed. Too bad. It worked incredibly well and the added cost would have been minimal.
What’s the solution?
I’ve tossed out a lot of examples of poor designs but I don’t want it to sound like all problem and no solution. So, let’s look at a few basic design principles that hopefully will begin to seep into the Astro-Equipment market.
Kinematic Constraint
A rigid body has six degrees of freedom, which includes 3 translational and 3 rotational axis. A degree of freedom can be controlled by connecting it to a rigid mechanical constraint. The principle of kinematic constraint is to control any unwanted motion (in position or angle) along a single degree of freedom with a single constraint. A kinematic constraint is geometrically defined and repeatable to a high degree of accuracy—often limited only by the mechanical stiffness of the constraint itself. A degree of freedom is over-constrained when more than a single constraint controls motion along that particular axis. Over constraint causes binding, distortion, and can lead to poor positional performance. In the worst case, it can also allow unwanted motion along that axis. More is not always better!
Let’s look at some quick, very simple examples of kinematic constraint.
1) A perfect ball on a perfect flat surface constrains a single degree of freedom
2) A high precision ball bearing rolling in a precision ground V-groove constrains two translational degrees of freedom so that the ball can only roll along a single axis.
3) A single ball in point contact with three balls restrained within a circular ring constrains all three degrees of translational freedom.
While we are at it, here are some common non-kinematic design choices.
1) Attaching one flat plate to another to constrain tilt. Any departure from flat in the way of high points (or from contamination such as dirt on the surface) makes the angle uncertain. There is no way to tell how the plates contact. Yes, you can lap them together and you can go to a lot of trouble to reduce errors but no matter how careful you are the repeatability and accuracy of controlling angle by mating two large area surfaces will be limited. Another form of this method is to butt two edges together. Again, this is not a kinematic solution.
2) Bolting one plate to another to control position. Bolt holes are oversized and bolt sizes vary. Bolting two plates together simply connects them with very little position control.
3) Controlling angle with two widely separated pins that fit into mating precision holes. This is an example of an over-constrained design. The pins and holes have to be co-aligned and the hole spacing has to perfectly match the pin spacing. Furthermore, a pin in a hole does not precisely control the position of the mating parts. This is what diamond pins are made for!
Pseudo or Semi-Kinematic Design
Psuedo Kinematic design uses the same geometry and concepts at a pure kinematic design but allows for very small over-constraint. For example, in order to control the tilt of a flat surface, you would ideally place it in contact with three non-linear rigid balls (for single point contact). However, space or cost constraints may make it difficult to implement a full kinematic design. In that case, you might simply machine the mounting base to include three very small raised rectangular pads for the mating surface to mount to. Since the raised pads do not provide a single point of contact, they over-constrain the solution. However, this kind of pseudo-kinematic approach can improve mechanical performance enormously over a totally non-kinematic solution. If you want to connect the two surfaces, placing a connecting screw through the raised pads minimizes the mechanical distortion of either plate.
Another not so obvious example of a psuedo-kinematic constraint is a ball in a conical hole. The problem is that a ball intercepts a cone along a circle and you don’t have a way to generate a geometrically perfect conical hole. The true kinematic solution constrains the ball position using point contact with three touching balls below.
When carefully applied, psuedo-kinematics can solve a LOT of the issues that I've listed here without adding significant to the cost of a product.
Conclusion
It is unfortunate that there appear to be so few companies in the astro-equipment world that are even aware of kinematic design. Worse, as I've mentioned, I've even tried to help a few of them with their design problems and I've been turned down. So, I get the impression that it's not a very high priority to get the mechanics right. That means that we customers are often left dealing with marginal designs--even from high end companies! In my experience, the one company that works very hard to produce excellent mechanics is Astro-Physics. Their designs are almost always spot on. I want to add that I haven't examined products from every company so I'm sure that there are others out there who are paying attention to the mechanics as well. Telescope and equipment alignment would be far, far easier and more precise if more components were properly designed and made! Starting with something that's a mess and then having to add another adjuster to compensate for all of the errors is fundamentally the wrong way to go about things, but that's what we are left with.
I can write a lot more about this stuff and how it applies to sensor tilt problems (and other stuff) but I don’t want to write a book here. If you are interested in this stuff and want more, a really good (and readable) handbook is “Field Guide to Optomechanical Design and Analysis” by Katie Schwertz and Jim Burge, published by the SPIE press. Katie’s master’s thesis also contains a lot of useful stuff and it’s worth a look. You can find it here: https://www.optimaxsi.com/wp-content/uploads/2014/01/Useful-Estimations-and-Rules-of-Thumb-for-Optomechanics.pdf
- John
Let’s look at just a few examples of poor design that have driven me crazy. Our first example comes from Planewave Instruments. The photos below show the end views of the main dovetail/carriage adapter that they supplied with my 20” telescope back in early 2020. They've shipped a lot of these things and for the life of me, I can’t understand why anyone thinks that it’s acceptable. The mechanics are so faulty that the entire weight of the telescope is held between a couple of pinch points in the carriage and the strength of the connection fall onto the screws themselves. Furthermore, the screws bind so terribly as the dovetail begins to engage that it’s hard to tell when the screws are properly torqued. Grade: D-
As an offer of good will, I suggested the simple design modification shown below to fix the design without having to start completely from scratch. In the end, Planewave declined to fix it so I returned it and replaced the carriage with one from Astro-Physics. The Astro-Physics carriage is beautifully designed and implemented. Designing a well implemented dovetail/carriage is not that hard and in spite of that fact, Planewave isn’t the only company that has had problems pulling it off. I’ve seen similar issues from Celestron with their orange extruded dovetails as well, but I’ve made the point so we’ll skip those photos.
Next on the list: I’ve got a FLI filter wheel for my F/6.7 refractor and for no extra charge, they included a mechanical problem that is so difficult to fix that I had to replace the entire assembly with another brand. In order to try to minimize the thickness of the filter wheel, the FLI design uses only two setscrews located at 120 degrees to mount a dovetail on the camera side of the wheel. The diagram below shows what happens when the screws are tightened. The two sets screws do what they are supposed to do, which is to draw the dovetail down into the mounting hole, but since there is no opposing screw (or lip) on the other side, the camera gets tilted. No amount of shimming will fix this problem! They would have been better off using just a plain cylindrical surface instead of a dovetail. It’s difficult to solve this problem in a clean way without seriously re-machining some of the parts. Grade: F
I replaced the FLI FW with a new one from Pegasus Astro. The Indigo filter wheel mounts 50 mm round filters, it is very inexpensive and at first blush, it looks like a very simple, if not, elegant design. It is very thin, it runs off of USB, and it allows bolting adapters directly to the filter wheel. Unfortunately, it doesn’t take long to uncover multiple problems with this design. First, they use arcing slotted holes to mount adapters and those holes open directly into the filter cavity, which will cover the filters with dust in no short order! What in the world were they thinking??? Fortunately, they designed bolt on covers to cover the holes and as fast as I could complain, they shipped them to me free of charge from Greece. The next problem rears its head when you mount the filters. The diagram below shows the problem.
The screw and washer retainer system is workable but the way that it’s implemented is just sloppy! In my professional life, I’ve shipped optical equipment that I designed all over the world and the first rule is that any screw that isn’t locked, WILL come loose in shipping. Since I’m sending this FW to Chile, I have to address this issue in a serious way.
The final problem with the Indigo FW involves the friction drive wheel on the edge of the filter wheel. It's the blue wheel in the photo above. Unfortunately, they fixed the position of the friction wheel. It’s adjustable for wear, but that’s only possible if you are there—as in when the FW is used in your backyard. That really doesn’t cut it when you are shipping equipment to a very remote location where service can be difficult and where you certainly don’t want to be opening the FW to expose the filters to dust. The drive wheel should have been pre-loaded against the edge of the filter carrier using a simple flexure spring. Additional cost: Very low. As it is, the mechanical design of this product is not very sophisticated. Grade: C-/D+. The good news is that they could clean up this design fairly easily…if they want to.
Last on my list is an issue that probably affects a lot of QHY camera owners. When I switched filter wheels, I had to use the QHY supplied carriage designed to mate to the V-notch “dovetail” on the front of the camera. I was stunned to see just how poorly constrained this whole system is! First, the carriage is seriously over-sized to allow for centering the camera. Let me get something straight here: The only reason you might want to center the camera is if your mechanics are so screwed up in the first place that you can’t mount the carriage where it is supposed to go! More adjustments almost always makes for more misalignment! With this system, it becomes very hard to tell when your camera is precisely centered, but fortunately, centration is not very critical on most systems. Here’s a diagram of how the camera mounts.
The much, much bigger problem is that QHY designed the V-notch spacing so that the centering screws engage it with the camera hanging in space—referenced to absolutely nothing! The centering screws are mounted in loose fit threaded holes and the tips of the screws aren’t coned or spherical. There is simply nothing that defines the angle of the camera! It might be close enough but more likely, the camera will be tilted a bit, no matter what you do. So, in the end you can’t tell when the camera is centered and you sure as heck can’t tell if it’s tilted either. Furthermore, the spacing from the front surface of the V-notch carriage to the sensor depends on how well the screw hole positions are controlled and on the unknown shape of the screw-tips. This is just a mess. The guys at QHY appear to be experts at electronics but in my opinion, their opto-mechanical skills are seriously flawed. No wonder so many folks are buying Ocupi adjusters! Grade for the QHY V-notch mount: F-
Fortunately, there is an easy fix to better control the tilt of the camera shown in the diagram below. First, a conical tip can be ground onto the tip of the centering screws. Second, by carefully measuring the top gap between the V-tail adapter face and the V-tail carriage (using feeler gauges), a circular washer can be cut from plastic shim stock with a bit of extra thickness (maybe 0.010” to 0.015”) to allow the centering screws to contact the bottom edge of the V-notch to pull the V-notch adapter face tight against the shim washer. Astro-Physics uses a V-notch system on their field flatteners but they’ve implemented it correctly so that the retaining screws only engage one edge of the V-notch pulling the adapter and mount tightly together along a reference surface. The diagram below shows the fix. Alternatively, the holes for the screws could be repositioned using a mill and an index table, but that's a lot more work. I'll do it if I need to move the camera inward. The ultimate fix would be to just redesign that whole front V-notch part to be a straight dove-tail...as it should have been in the first place!
Finally, I’ll give an honorable-mention to Celestron for their utterly poor mechanical design of the secondary mount on their Edge series of SCTs. That design is not well constrained in angle or position and that results in having to realign the secondary every time it is removed or after something as simple as a car trip. Years ago, I completely redesigned the secondary mount on my prototype C14 using kinematic principles and demonstrated on an interferometer that it could hold alignment to within about 5 fringes across the pupil even after pulling the secondary out and remounting it. (The factory version controlled tilt 20-40 times worse!). It had super easy to use, orthogonal tilt adjustment and using CNC, it could be made for only a few more dollars than the existing design. I offered the design to Celestron but it wasn’t worth it to them so they passed. Too bad. It worked incredibly well and the added cost would have been minimal.What’s the solution?
I’ve tossed out a lot of examples of poor designs but I don’t want it to sound like all problem and no solution. So, let’s look at a few basic design principles that hopefully will begin to seep into the Astro-Equipment market.
Kinematic Constraint
A rigid body has six degrees of freedom, which includes 3 translational and 3 rotational axis. A degree of freedom can be controlled by connecting it to a rigid mechanical constraint. The principle of kinematic constraint is to control any unwanted motion (in position or angle) along a single degree of freedom with a single constraint. A kinematic constraint is geometrically defined and repeatable to a high degree of accuracy—often limited only by the mechanical stiffness of the constraint itself. A degree of freedom is over-constrained when more than a single constraint controls motion along that particular axis. Over constraint causes binding, distortion, and can lead to poor positional performance. In the worst case, it can also allow unwanted motion along that axis. More is not always better!
Let’s look at some quick, very simple examples of kinematic constraint.
1) A perfect ball on a perfect flat surface constrains a single degree of freedom
2) A high precision ball bearing rolling in a precision ground V-groove constrains two translational degrees of freedom so that the ball can only roll along a single axis.
3) A single ball in point contact with three balls restrained within a circular ring constrains all three degrees of translational freedom.
While we are at it, here are some common non-kinematic design choices.
1) Attaching one flat plate to another to constrain tilt. Any departure from flat in the way of high points (or from contamination such as dirt on the surface) makes the angle uncertain. There is no way to tell how the plates contact. Yes, you can lap them together and you can go to a lot of trouble to reduce errors but no matter how careful you are the repeatability and accuracy of controlling angle by mating two large area surfaces will be limited. Another form of this method is to butt two edges together. Again, this is not a kinematic solution.
2) Bolting one plate to another to control position. Bolt holes are oversized and bolt sizes vary. Bolting two plates together simply connects them with very little position control.
3) Controlling angle with two widely separated pins that fit into mating precision holes. This is an example of an over-constrained design. The pins and holes have to be co-aligned and the hole spacing has to perfectly match the pin spacing. Furthermore, a pin in a hole does not precisely control the position of the mating parts. This is what diamond pins are made for!
Pseudo or Semi-Kinematic Design
Psuedo Kinematic design uses the same geometry and concepts at a pure kinematic design but allows for very small over-constraint. For example, in order to control the tilt of a flat surface, you would ideally place it in contact with three non-linear rigid balls (for single point contact). However, space or cost constraints may make it difficult to implement a full kinematic design. In that case, you might simply machine the mounting base to include three very small raised rectangular pads for the mating surface to mount to. Since the raised pads do not provide a single point of contact, they over-constrain the solution. However, this kind of pseudo-kinematic approach can improve mechanical performance enormously over a totally non-kinematic solution. If you want to connect the two surfaces, placing a connecting screw through the raised pads minimizes the mechanical distortion of either plate.
Another not so obvious example of a psuedo-kinematic constraint is a ball in a conical hole. The problem is that a ball intercepts a cone along a circle and you don’t have a way to generate a geometrically perfect conical hole. The true kinematic solution constrains the ball position using point contact with three touching balls below.
When carefully applied, psuedo-kinematics can solve a LOT of the issues that I've listed here without adding significant to the cost of a product.
Conclusion
It is unfortunate that there appear to be so few companies in the astro-equipment world that are even aware of kinematic design. Worse, as I've mentioned, I've even tried to help a few of them with their design problems and I've been turned down. So, I get the impression that it's not a very high priority to get the mechanics right. That means that we customers are often left dealing with marginal designs--even from high end companies! In my experience, the one company that works very hard to produce excellent mechanics is Astro-Physics. Their designs are almost always spot on. I want to add that I haven't examined products from every company so I'm sure that there are others out there who are paying attention to the mechanics as well. Telescope and equipment alignment would be far, far easier and more precise if more components were properly designed and made! Starting with something that's a mess and then having to add another adjuster to compensate for all of the errors is fundamentally the wrong way to go about things, but that's what we are left with.
I can write a lot more about this stuff and how it applies to sensor tilt problems (and other stuff) but I don’t want to write a book here. If you are interested in this stuff and want more, a really good (and readable) handbook is “Field Guide to Optomechanical Design and Analysis” by Katie Schwertz and Jim Burge, published by the SPIE press. Katie’s master’s thesis also contains a lot of useful stuff and it’s worth a look. You can find it here: https://www.optimaxsi.com/wp-content/uploads/2014/01/Useful-Estimations-and-Rules-of-Thumb-for-Optomechanics.pdf
- John
