From :ATM Türk: Amatör Teleskop Yapımı

---

Current revision

10” f/6 computerized telescope using OnStep This is a brief account of how we designed and built our home-made 10” f/6 automated Dobsonian telescope, almost exclusively with DIY techniques.

The project can be seen in two phases: Mechanical design, production and assembly Automation with OnStep, an open-source telescope control package

OnStep is an open-source telescope controller software package, developed by Howard J. Dutton. It was designed and developed for a multitude of Arduino platforms including Teensy, AtMega, BluePill and more. The code is freely available at https://github.com/hjd1964/OnStep and is actively enriched and supported by Howard J. Dutton and the OnStep community.

Apart from the card and step motors, the remaining optical/mechanical parts have either bought or fabricated for the rest of the telescope. Compared to a commercial telescope, ours might have a marginal advantage in the cost, which may not be significant. On the other hand, increasing the size of the primary mirror may justify the cost. This is especially true for the mirror sizes 16” and up. Implementations of OnStep based controllers are more common for equatorial mounts compared to Alt-azimuth but again for bigger telescopes where heavier loads should be moved with precision, it also makes more sense to use it. We also saw that complexity during design & the manufacturing stages of such telescopes might require a group effort / cooperative work. Also, you may need to know what others’ had done in the past, learn about their elegant solutions to the potential problems on the road and even ‘borrow’ some of their innovative design features which we also did without hesitation.

For the OnStep project, we appreciate the remarkable effort of Howard Dutton and Khalid Baheyeldin as well. They patiently answer never-ending questions of the enthusiastic crowd.

At the beginning of our project, we have searched the web to find different motorized alt-azimuth telescopes to guide and ease our design decisions but except the ones based on commercial products or quite simple ones, they did not look attractive enough. Eventually, we came across some elegant designs for mounting the motors, pulleys, and timing belts with minimum complexity. Even though one can find a very detailed plan for a motorized telescope, we were also expecting to encounter some problems which the makers of these telescopes were not providing enough detail about as if everything were working perfectly as planned, without any performance issues. We will also discuss here such problems of our design & fabrication.

Purchasing, material supply & machining the parts In the beginning, since we had started with a readily made 10” f/6 primary mirror, and a 2.14” secondary, the telescope would definitely be a Dobsonian. An Alt-Azimuth type of mount, suitable for even bigger mirrors. It could also be possible to make an equatorial mount for which was more suitable for astrophotography, an alt-azimuth mount controlled with OnStep seemed to have more potential for various reasons, including mechanical superiority and feasibility of using larger mirrors in the future, which we are seriously considering. Actually, one of the prototypes of our OnStep controller cards has been successfully used to command a 20” f/4.5 Dobsonian (so far, the largest amateur telescope in Turkey), providing positive feedback for our future work. One of the two 10” f/6 telescopes which we’ve attempted to manufacture had already a commercially bought primary mirror, it is now operational. The second one has all the mechanical parts have been fabricated and assembled, waiting for the polishing/figuring & coating of the primary mirror.

One member of our study group used parametric CAD application to design all mechanical parts of the telescope. He also performed virtual assembly of these parts prior to actual building. This gave us enough confidence about the form, fit and function of most if not all parts. Using CAD for design should be an integral part of a project and even if you can improvise freely to a degree, you can never be sure without CAD applications’ set of tools. We’ve verified our design in early stages, computed center of mass (which was somewhat critical for balance of the scope) and generated all the 2D drawings needed for CNC router, so a minimum level of guesswork and frustration encountered in manufacturing and assembly.

Of course in a project of such complexity, it is not possible to verify everything beforehand and even though CAD reduces your potential mistakes and saves you time, there remains still some unknown points, which you yourself have to find solutions or trial and error based works.

The bill of material of our telescope includes mostly manufactured parts compared to purchased ones. In the mount, we have used 15 mm thick marine grade birch plywood. For the two telescopes we’ve attempted to fabricate simultaneously, we used two sheets of 122 x 250 cm plywood. Enough to make a pair of rocker box, mirror box, and two secondary cages. Since 15 mm thick plywood would be too heavy for the secondary cage rings, we’ve decided to reduce its thickness using a planning machine, after they are cut with a CNC router. Not a big deal for a total of 4 rings but a small sheet of 6 or 9 mm would definitely be easier. To glue and join these parts we had asked the assistance of a professional carpenter since the workshop had an infinite number of clamps and workshop tools we lack. Without them, assembling the parts of rocker & mirror boxes would be very difficult with enough precision.

To fabricate the mirror cell, we visited a metal workshop, where all welding & truing has been done by a professional. Then two mirror cells have been electrostatically powder coated to black at another shop nearby. To ease the removal of the primary mirror, two bolts are unfastened and the cell pivots around two others to access the primary and take it out from the cell for storage or cleaning. The mirror rests on nylon nuts and supported with two pairs of PTFE bearings, separated with 90° each angle to minimize astigmatism. When pointed towards Zenith ~5 kg mass of primary mirror rests over these 4 bearings. A solution probably more expensive/time-consuming to adapt but better than using a sling or simple side supports.

For the ø 25 mm aluminium (1 mm wall thickness) truss-poles, we have adapted a hexapod solution which enables the collimation of the optics. By turning the truss poles clockwise / counter-clockwise, we can align the optical components and bring them to the same path. This feature not only simplified the complexity of the mirror cell but at the same time enables one to perform the collimation of the telescope without any assistance and back & forth traveling between the mirror cell and focuser. The comfort of hexapod collimation alone well worth the extra cost and burden of connectors to be purchased such as male threaded rod bearings or manufactured items like acetal taps to be inserted into the aluminum tubes. All of the six poles have been connected by the ends with bolts to form a collapsible structure, reducing the clutter and ease the assembly. In order to attach truss poles to both the secondary cage and the mirror box, we have initially used M8 bakelite bolts but they did not work as expectedly and couldn’t provide the compressive strength of the truss poles & secondary cage side and then we changed to M8 screws, tighten by Allen keys. In order to assemble the collapsible truss and connect secondary cage to mirror box, a total of six bolts should be tightened. Even though this was not what we had initially planned but until we can find and implement a better tool-free solution, tolerable for us.

For the secondary cage, we’ve decided to implement a wire spider to mount the 2.14” secondary mirror. This design looks complicated at first, however, the simplicity of actual manufacturing and final rigidity justified our decision. Our secondary mirror is approximately aligned in the center of the secondary cage, brought to the correct position relative to the focuser by a simple jig and then minor adjustments required during collimation are performed using truss poles _another advantage of the hexapod structure. So far, we observed the wire spider holds the collimation stiff enough to resist gravitational loads. The tensioning of 4 pairs of ø 0.1 mm steel wires is done by turning the brass sleeve nuts inserted to a same-sized socket cap screw, where a 0.15 mm, hole has been drilled to its head along the 45° axis using electrical discharge machining. Steel wires are simply inserted from these holes and connected to the secondary holder plate, a 45° aluminum plate which also has screws for fixing the wires. Turning the screws make the minor adjustments of the wires so that they can align the secondary holder perfectly and bring it to the center of the optical path. The above-mentioned sleeve nuts are then inserted into ~180 mm long ø 19 mm aluminum tubes, which connect the rings of the secondary cage to each other.

Other items of the secondary cage-like focuser & Rigel unit finder have been simply attached to the cage and a matte black polyamide (Kydex like) material wrapped along the circumference to block stray light and protect the optical surfaces. The secondary cage is not at optimum weight but with different eyepieces and unit finder, the balance of the telescope is quite well and it’s neither head light nor heavy. The focuser we’ve chosen for this telescope is a 2” Crayford focuser and it has a relatively simple design. It does not have a microfocus knob (with 1:10 reducer) for fine focus because we eventually will control the focuser with OnStep in the future and have fine-focusing capability. Also, we did not use a finderscope (RACI) with similar reasoning, star hopping is not required (or practical) for this kind of computerized telescope.

The mirror box, which holds the 40 mm thick Ostohowski made Pyrex mirror on 9 point flotation cell and a NEMA 17 step motor with a 50:1 ratio high precision planetary gearbox. This motor is connected to a 12 mm diameter steel shaft using a 20 mm wide, 255 tooth circular timing belt (AT5) & 25 tooth dual pulleys, which is also connected to dual pulleys turn the mirror box along altitude axis. Inside walls of the mirror box have been painted matte black (BLACK 2.0 by Stuart Semple) to minimize reflections. The mirror box also holds (by a total of eight M8 bolts) the altitude bearings. These bearings are cut from 2x15 mm thick plywood and one of the pairs is covered with stainless steel strips to minimize wear & increase dimensional stability. The dual pulleys (25 teeth) at the sides of the mirror box are hidden under these altitude bearings to protect them from external elements and to obtain a minimalistic view. Moving parts like timing belts, motors or pulleys of the telescope are not visible externally.

The lowest level component, rocker box has a very simple design. It includes two pairs of adjustable idler bearings to balance the weight of the rest of the telescope and the azimuth motor (again a NEMA 17 step motor with 50:1 ratio high precision planetary gearbox) and corresponding timing belt, a circular T5 with 288 teeth, again hidden from plain sight. Since we do not rely on fine-tuned friction for balance of the scope as this is the usual case for most Dobsonians, we have used a 2 mm thick stainless steel ring between the circular plywood part which a timing belt has been glued across its perimeter and used three transfer bearings on the recessed sockets carved under the base plate (plywood) of the rocker box, so that it may rotate without stiction and do not dig into plywood surface in time, under the considerable weight of the whole telescope assembly. This freely rotating structure has a ‘break’, which is the 16 teeth pulley, that is rotating the mirror box with engaging its teeth to this timing belt, in a tangential position. And this pulley/drive mechanism is again invisible unless you disassemble everything and reveal or turn the rocker box upside down. Most of the computerized telescopes have either eclectic drive mechanisms (which is necessarily not a bad thing but not a visually pleasant thing to the eye) or slip/clutch mechanisms, to enable free movement in both axes, which is again, from our point of view, is unnecessarily complicated. The two 12V batteries and the OnStep controller unit also rests freely on the base of the rocker box. The height of these items are low enough not to hit the mirror box during swivel motion. Cables coming from the step motors are reaching to the controller unit and altitude motor connector must be disconnected during the transportation of the telescope as separate pieces which is mostly the case. While seperating the mirror box from

PCB & electronics We have designed and manufactured three separate PCBs (V1, V2 & V3) for our implementation of OnStep. Each card has the same dimensions and PCB work & soldering has been completed without any external help. V2 & V3 of the card uses a DRV8825 step motor controller, which provides silent micro-stepping (32) of dual NEMA 17 motors where V3 also includes support for a smart hand controller (SHC), a mechanical keypad, connected with a spiral cord cable to the card, a nice and more robust alternative to transmitting commands to OnStep, using a cellular phone or a tablet/laptop etc. We run the card with dual 7500 mAh 12 V batteries (connected in series) to obtain 24 V. The batteries provide approximately ~8 hours of continuous operation time before motors stall. Two connectors coming out from the plexiglass case with perforated to dissipate heat generated by the components are connected to the motors. The on/off switch and WiFi antenna is also located on the case. The card can also be powered by a 12V DC adapter when the telescope is near an AC outlet which is sometimes possible.

We chose to build our telescope controller hardware based on the Arduino BluePill architecture. One major decision that we made was to design and build our own monolithic controller hardware platform rather than a mix-and-match of generic Arduino modules commonly sold on the market. This approach offers several advantages:

It results in an end-product that is well suited to our specific requirements in the context of telescope operation and observational field work, including power, thermal management, MTBF, mechanical, usability, serviceability. Cost-effectiveness, since we buy chips, rather than Arduino modules Suitability for industrialization

The Arduino BluePill architecture is based on the STM32F103 microcontroller by ST Microelectronics. This a ubiquitous and very cost-effective chip, optimizing the bang-for-buck for our needs. Similarly, DRV8825 Stepper motor controllers were chosen to drive the NEMA17 motors mechanically moving the telescope assembly along Altitude and Azimuth axes.

We selected some of the key design parameters as follows:

Input voltage range: 12-30V Nominal battery voltage (typical operation): 24V Motor current range: up to 2A per winding Motor torque range: up to 5 Nm Motor reduction ratio: 50:1 Telescope GoTo speed: 5 degrees/second max

The electronic PCB design process also took into consideration the removal of excess heat to be generated by the motor driver chips under heavy load.

Problems & re-work Almost like being prototyping phase for our design & manufacturing of this telescope, we have observed several problems (in other words, opportunities for improvement) some quite easy to address, some difficult or expensive. Once the telescope was operative and we tried to make precision GoTo’s, we’ve noticed backlash problems and tried to understand how critical they were. Backlash in a computerized telescope may not be noticeable unless you try to aim the scope that requires direction change in azimuth/altitude or both. During tracking, however, the backlash had no effect. For the altitude axis, we used bubble levels in order to rotate the mirror box between horizontal and vertical planes. Using an Arduino code (stepper.ino) we have rotated 90 degrees and counted how many full steps required to rotate 90 degrees. Then returning back would require the same number of steps but due to backlash, we’ve observed that we had around 480 full steps (or in 1.3 degrees) missing which was equal to backlash. Several other tests verified these results and we could be sure of this value. In order to find out and reduce the backlash in altitude axis, we have increased the width of the timing belt from 10 mm to 20 mm and also, increased the number of teeth which this belt drives from 16 to 25, which had no effect on reduction since they remain 1:1 in ratios. We also changed the belt profile from T5 to AT5 in order to decrease flexure in the belt. All these changes helped us to get rid of the backlash to a degree and now we measured it ~0.27 degrees.

Operation of the telescope Compared to a grab and go 8” telescope, our 10” computerized telescope requires some planning for transportation to the observing site. Once you disassemble the truss poles and take the secondary cage away from the telescope, you can also separate the rocker box and the mirror box and and have to deal with four main sub-assemblies to transport. Heaviest of these four is the mirror box, weighing approximately 12 kgs. The 8 kg of rocker box comes second. Secondary cage is quite lightweight and only requires some delicacy since it contains the secondary mirror and thin wires of spider. Collapsible truss poles are also lightweight but it helps to carry them in a tube (plastic / cardboard ø 70 mm in diameter, 80 cm length) to protect the surface finish and minimize the required space. Other small items like batteries, the box holding the OnStep card, cables, eyepieces, unit finder, etc. are practically less problematic to transport. Of course, you may also carry the telescope on a pushcart if you have adequate space in your car / truck or you simply take out the scope from a fixed location (building) to a nearby observation site, like a school, etc. This also simplifies the setup procedure of the telescope and possibly frees you from collimating the optics everytime. Also, you may have the additional comfort of leaving OnStep card & connector cables on the telescope, this also simplifies setup and saves you a couple of minutes. Once the mirror box has been put over the rocker box and collapsible truss poles have attached, you may start collimation and connect to wifi network of OnStep with your cellular phone, On startup OnStep needs to geo-reference itself using 1, 2 or 3 star alignment procedure. Once you perform the alignment, you can track with sideral (%99 of the time) or Lunar / Solar tracking rates or perform GoTo operations directly within Android app or a third party planetarium app like Carte Du Ciel or Stellarium, etc. With the latter path, you may move the telescope by visually selecting targets within planetarium's screen as well as providing coordinates of celestial bodies if you know. Minor GoTo errors can be adjusted by guide operations, where you can use the arrow keys (North / West / East / South) and move at different speeds ranging from very slow to very fast. The slew rate of the telescope during GoTo operations (a changeable parameter in the configuration file) is around 3°/second so it takes roughly 30 seconds to travel from Zenith to horizon and 1 minute from East to West or North to South. Higher slew rates bring the risk of motor stalls and/or hitting the telescope to a nearby obstacle and necessity to making a new star alignment. During manual guiding, the slew rate changes according to your requirements and generally half of these speed is possible.

The focal length of the mirror is exactly 1524 mm so with 2” eyepieces ranging from 40 mm to 7.5 mm (with 2X Barlow lens) magnifications from 38x to 203x can be obtained. The quality of the mirror in this telescope can sustain 35x to 50x per inch of aperture, so theoretically it can work up to 350x to 500x maximum before magnification roll-off and of course atmospheric conditions permit. In order to see how well we can track a target tests we pushed the magnification limit to 662x using a Celestron X-Cel 2.3 mm eyepiece but this was only for the sake of experimentation and the result was not visually appealing at all.

Compared to an ultra-light 10” telescope, our classical Dobsonian design ends up being considerably heavy. Which is not good especially ease of transportation and storage is a priority. But the weight works in favor of us during telescope operation. It is less prone to wind loads (less vibration at the eyepiece) and quicker settle down times if you mistakenly hit the secondary cage or other parts. The total weight of 35 kg’s is too heavy as as single piece to lift even a car to a nearby observing spot 20 - 30 meters away but of course you can make more than one trip from your car to the observing spot to carry the parts of the telescope one by one.

Classical mirror box also helps in resisting the dew and probably protects your mirror against other pollutants. Even without a shroud, mirror is hardly exposed to the elements. This is partly true for the secondary mirror as well. The depth of the cage protects the secondary mirror to a degree. A single ring type minimalistics cage can beat our design choice in weight issues but again classical secondary cage is superior in preventing stray light even without a dedicated baffle extending beyond its end.

Future enhancements Secondary heater / boundary layer fans / shroud / derotator integration / robo-focus / cosmetics

Photos, videos & files https://www.youtube.com/watch?v=eSOFmSh7xH