Dome & Shutter Design Telescope Drives
The observatory's location has had a chequered history. Originally located in Forest Row, East Sussex, between 1986 & 1990, it was mothballed for five and half years and observing during the hiatus carried on from Muswell Hill, north London. During the summer of 1996 Brayebrook Observatory was relocated to Little Eversden, alongside the Lord's Bridge radio interferometer, near Cambridge. The skies at this rural locality are very dark, the limiting zenithal magnitude frequently reaches +6.5m. There is no light pollution and sparse local street lighting.
The observatory has been redesigned, raising the prefabricated 'zintech' wall onto a 40-inch course of brickwork, and providing a working underfloor area suitable for storage and ancillary equipment.
Assessment of Wind Loading in accordance with BS6399:Pt.2 1995 & BRE GUIDE to Wind Loading of Structures: Pt.2 1995
@20m/s (45mph) - subcritical flow, Reynolds No <1200
Horizontal Load: 953lbsf
Vertical Load: -5719lbsf
Resultant Load: -5789lbsf@80º
@30m/s (67mph) - transcritical flow, Reynolds No ~1200
Horizontal Load: 2383lbsf
Vertical Load: -2859lbsf
Resultant Load: -3722lbsf@40º
@40m/s (90mph) - supercritcal flow, Reynolds No >1200
Horizontal Load: 4289lbsf
Vertical Load: -953lbsf
Resultant Load: -4394lbsf@121/2º
The dome wind loadings have been based on two different approaches to the problem. Code of Practice, BS3 Pt.2 1972 treats the problem as a dynamic loading, assuming no drag. In this instance I have divided the dome into a series of latitudinal strips, assuming the direction of airflow across the horizontal diameter of the polar axis. Using Hutton's formula for the equivalent loading of an inclined plane, I have integrated the loads between the ground and apex for each strip and determined the resultant vector.
BS6399:Pt.2 considers the effects of drag, and considers a gust factor. The guide to the wind loading of buildings considers the airflow over the dome. In this I have assumed the dome to be a rough hemispherical surface. When the airflow over the dome is lamina, the Bernoulli effect produces a very high vertical component, but the moderate horizontal component will cause the dome to deform along the inside perimeter of the leeward edge. The Bernoulli effect diminishes at the transcritical flow region because of drag around the skirt. In the supercritical region there is no lift, only the horizontal component, heavily concentrated on the windward pole of the dome. On the leeward edge there is a horseshoe shaped turbulent wake which produces a modest windward horizontal component. This resultant translocational force produces a high overturning moment.
There are choices when designing an observatory
dome, between incorporating the rail on the dome itself or on
the wall, and whether to adopt a bi-parting or an up-and-over
shutter. Each have their pros and cons. I decided not only to
place the rail on the wall, but to make it a part of the fabricated
structure. The reasons for this were twofold. The dome could be
redesigned as a flexible structure, and the rail could be rigidly
supported, which is easier when attempting to maintain circularity.
The disadvantage of placing the rollers on the dome is in routing
electrical power to the drive system. In order to keep wind resistance
to a minimum, I also adopted the up-and-over shutter. It requires
a more powerful drive than a bi-parting shutter, but one can make
full use of the shutter width without having to open the shutter
completely. This is useful when observing in daylight and in windy
The problem of transferring electrical power from the incoming single phase supply under the observing floor to the dome was solved by using gas recombination batteries and an AC Powerverter mounted on the dome either side of the drive motors. This is a proprietary unit that not only converts 12VDC to 240VAC, but also acts as a soft start device and monitors the charge condition of the batteries and periodically institutes a recharge cycle, ensuring maximum life and optimum withdrawable capacity.
The shutter is controlled by a Parvalux 95W 7 rev/min capacitor start induction gear motor. This turns a windlass, which opens and closes the shutter via an endless cord. Shutter travel overrun is prevented by limit switches.
The dome is driven by a Parvalux 38W 12.5 rev/min capacitor start induction gear motor and a unit ratio timing pulley coupled to one of the dome rollers.
Both drives may be operated by a hand paddle, but there is a proximity sensor in the form of an induction lead along the shutter opening and a 50kHz oscillator around the end of the telescope tube, which trips the dome drive relays and maintains the telescope-shutter alignment once set.
There is a torque limiter fitted to the dome drive motor to prevent the gearbox being damaged in the event of an overload.
All the dome and shutter drive controls are routed to a service panel on the shutter transom, and can be accessed remotely via a 12-pole jack plug.
The dome is 121/2ft diameter with a 3 foot wide shutter.
When designing my dome I was mindful it should turn easily, so it mustn't weigh a lot, but it must be windproof. The shutter mustn't pop out of its track and it should be as wide as at least three times the telescope's aperture, The dome must be sufficiently rigid not to flex when it has a relatively wide slot cut out of it. Lightness and stiffness are conflicting requirements because in trying to add sufficient strength to the dome to prevent it distorting when it has a wide shutter opening, it could easily end up so heavy it would be a real struggle to turn round.
Timber or glass reinforced plastic ('fibreglass') seem easy materials to work with, but weight for weight they lack rigidity compared to steel, and they need a support structure, which adds a lot of weight compared to the weight of the cladding. They also need regular maintenance. Steel is relatively inexpensive and has a high strength to weight ratio. Corrosion resistant steels are too expensive. Aluminium is light, corrosion resistant and readily formed, but it lacks rigidity. On the other hand, there are copper-aluminium alloys ('duralamin'), that are both readily formed and have a higher strength to weight ratio than steel. Moreover they are, to all intents and purposes, maintenance free and also highly heat reflective.
All things considered, the ideal material is 'half-hard' aluminium which is a solution treated semi-hardened dural alloy. Weight for weight it is about the same price as mild steel sheet.
A truly hemispherical dome entails manufacturing 'gores', sections of the surface of a sphere. They are curved to the sphere's radius in both horizontal and vertical directions, and they are extremely awkward to make. They also require fastening to a rigid framework that must be made from either steel or dural angle, that must be rolled to the correct radius, or curved timber frames that are both heavy and labour intensive.
What if instead of opting for a true hemisphere, I opted instead for a faceted shape that closely approximated a hemisphere, and in order to save time and expense, I made the facets with straight sides, and abandoned the idea of cladding a skeleton framework and instead fastened the facets to one another to form a "monocoque" structure? Since the dome must have an opening which allows the tube to look vertically upwards, the facets would not need to meet at the dome's apex, but a set distance down. That radial distance would correspond to half the width of the shutter.
The next consideration was the thickness of the dural sheeting, and how much of it I would need to buy! 16SWG (0".028) is an easily worked sheet thickness suitable for domes from 9ft. to 18ft. diameter. It comes in standard sheet sizes (8ft. x 4ft.), and my task was to make the biggest dome from the smallest number of sheets that made use of the most sheet area, so I had the minimum wastage.
I designed my dome along these lines in December 1983 with the assistance of the late Alan Young , who ran a small fabrication workshop called Broadview Engineering in Burwash, East Sussex. Alan sadly died in January 1990 aged only 52 years, and he is sorely missed by all who knew him, if only because he was a very practical man full of clever ideas.
I settled on a "pseudo-gore" formed from 3 separate trapezoidal facets. Each facet has a joggled flange, top and bottom, and a return fold on both sides. Because the facets are flat they can be cut out of stock sheet using a workshop sheet metal guillotine, and joggles formed on a bench press and a brake fold. The three separate facet shapes were drawn out using a template, and templates were also used to form the joggles and for drilling the rivet holes. They were then sealed and pop-riveted together along their joggles. They thus formed a single 'pseudo-gore" side panel, and when fastened side to side, a "quasi-dome" with a hole at the top. This hole was the same diameter as the shutter width.
To assemble all the side panels Alan used a jig. This took the form of a vertical pole (a scaffold pole), with a tubular ring with spokes welded at right angles to the top. (The finished fixture resembled a May pole with a Catherine wheel on top). Around the bottom of the pole was placed the base ring, which would become the bottom lip of the dome once all the side panels were riveted in situ. I used 2"x2"x1/4" dural angle, cut and formed into a 42 sided polygon. This was done by making hacksaw cuts in the upper side of the angle, and heating the outer side with a welding torch, just sufficient to soften it so it could be bent to the correct angle (171º:26 '), and then welding the saw cuts together once the ring was completed. Care was taken not to distort the ring when heating it. It must lie flat and not become twisted or buckled, and most importantly, it must be the same distance across flats all the way round. (A 1/16" leeway is tolerable-but no more). The last side panel was made to suit so as to accommodate accumulated build error.
Next Alan needed to fabricate an up stand for the shutter track. I could have chosen bi-parting shutters, but I dislike them, not just for the reasons I have already given, but because although they are simpler to make and drive, they catch the wind, and are easily popped out. An up-and-over shutter is best, but because it must open beyond the dome apex, it must either be made in two separate sections (as in the William Herschel & Keck telescope domes) or have a separate drop down flap at the bottom. (This option is cheaper because only a single shutter track is needed). The bottom flap folds in half outwards when I need to observe near the horizon. The shutter track is formed from "U" section, rolled to the desired radius, with the opening facing outwards. The shutter wraps around the outside of the track and runs on pairs of rollers set on spiders. There is a brace and turnbuckle set in top hat cleats at the bottom of the shutter transom. This prevents the shutter opening being forced inwards when subject to high wind loads, which in turn makes it impossible for the shutter to be lifted out of its track.
It is a simple matter to automate both
dome rotation and shutter opening/closing. If considered at the
outset, it need not be the insuperable problem most advanced amateur
astronomers make it out to be. If you want to be able to drive
a dome with a small, economical motor, it must turn freely and
easily. The monocoque design of my dome results in a light but
extremely rigid structure, much more so than a true hemisphere
formed from a framework clad with gores. But in pushing a dome
round from the side it will not of its own volition move in a
circle. Every action has an equal and opposite reaction. Push
tangentially, and the dome reacts by trying to move sideways.
The only thing that forces it to go round is the roller system.
It is possible (in fact it is the norm!) to fasten the rail to the dome lip and place the rollers on the wall, but the rail needs to be heavy and rigid otherwise it will flex under the weight of the dome, and the dome will jam. This requirement leads to a rail so heavy that the dome weight rises to the point where a considerable effort is needed to push it round, and motorisation becomes impractical. Better to place the heavy rail on the wall so it is no longer a moving part, and place the rollers on the dome.
In deciding on a rail section several factors have to be considered. If it is rolled from strip it flexes easily and twists and buckles, so it needs lots of side stiffeners and support cleats that interfere with the rollers. If it is angle to tee section it is expensive and difficult to roll into a true arc. What is needed is a section that is as stiff sideways as it is up and down, but can be easily rolled into an arc. The obvious shape that fulfils these criteria is the tube. After calculating the deflection between supports under the static dome load plus wind and snow loads, I selected Schedule 10 (47mm) steel tube (about 4mm wall thickness). The rail outside diameter is about 8-inches less than the across flats dimension of the dome, so as to provide clearance for the roller support brackets and the roller standoff from the dome skin. The rail tube was rolled at a mill in five sections to a template, to a tolerance of ±1/8 inch, and plug and butt welded on site into a 12-foot centre line diameter. The rail was located on wall stanchions using 4"x4"x1/4" welded steel seating plates. When bolted in situ the rail was measured and found to be round to within `1/4 inch and adjusted dead level. It was important to brace the sections with G clamps and scaffold poles and arc weld in short radial segments, one at a time, checking on the roundness as we progressed. If one joint were welded at a time, the intense local heating around the joint would cause the tube to expand and force it out-of-round, and it would have been impossible to get it back round afterwards. The butt welds were dressed flush, and keeper plates welded underneath.
The number of rollers is a major contributory factor to rolling resistance. The dome tries to move sideways when it is pushed, so if the rollers have a flat cross-section, both horizontal and vertical rollers would be necessary to ensure the dome went round. The trouble with this two dimensional approach is that it ignores the very real problem with any rail/roller system. Because the reaction of the roller on the rail due to the weight of the dome is not inelastic, the rollers produce friction which causes rolling resistance. The accepted wisdom is that the rolling resistance is directly proportional to the load per roller, and therefore the more rollers used, the lower the load per roller (assuming all the rollers make equal contact), and consequently the combined rolling resistance remains the same, regardless of their number. And adopting the ball race principle of the roller bearing, the more rollers the better. Nothing could be further from the truth. Rolling resistance is caused by elastic contact between rail and roller. The logic used to advocate lots of rollers assumes the contact is inelastic; that is it treats the surfaces as infinitely hard. The contact area is not directly proportional to load, doubling the load does not double the contact area, it only marginally increases it. In fact the fewer rollers used, the lower the rolling resistance. The minimum number of rollers possible is three. But, this assumes the dome is perfectly rigid, which obviously it can't be. So for a dome diameter of 10ft. to 12ft. I reckon 6 rollers is a sensible choice, and for a dome between 14ft. and 18ft. 12 rollers. However, only half of them need be in permanent contact. The ones in between can stand off by about 1/8 inch. They merely provide sideways or lateral restraint, i.e. they prevent the dome "crabbing".
Roller profile and size is also a contributory factor to rolling resistance. Small rollers suffer high axle loads, which in turn produces high frictional losses and increased rolling resistance. As a rule of thumb, the rolling contact line diameter should be at least twice that of the rail diameter. Advantage can also be taken of the dome panel design to stand each roller off a square section tubular framework, bolted to a lower panel. Since the panel is inclined 12 º.7 to the vertical, the axle of the roller is declined 13º which provides a semi-kinematic advantage in the bid to encourage the dome to want to move in a circle rather than sideways. Because the rail has a tubular cross-section, I chose a roller section that runs on both the outside and the inside of the rail simultaneously, so as to combine the actions of both horizontal and vertical rollers in a single roller. To minimize the contact area I adopted flat contact faces, not curved to match the rail section. Such a roller is "Diablo" shaped.
The rollers were machined from Nylon 66. I selected this material because it will never need lubricating, and in fact it both grips the railtube and scrubs it clean in a pressure polishing action. The roller axles were fitted with roller bearings, and cover plates which can be removed so the bearings can be occasionally regreased.
To prevent the dome becoming derailed in a high wind rail restraints are necessary. A dome, properly designed and fabricated, is a windproof structure. The effect of a wind is a complex matter. Enormous loads are generated that distort the dome's shape. If the shutter is trapped by the track so it cannot pop out, then the dome remains windproof. But it can be displaced off the rail, so it must be restrained by a series of brackets that run around the inside edge. These do not have to lock beneath the rail, contrary to what is almost universally claimed by so called experts. The dome cannot be lifted straight upwards, but it can be overturned. As the dome is pushed on any side, the rail restraints grab on the inside edge opposite and lock the dome to the rail. This design was incorporated into my dome in January 1987. It withstood a force 9 gale later that month, a force 11 storm in March of that year, and the Great Storm of October 15/16th.'87, when the wind reached storm force 14. It was not otherwise held down. However a conventional dome of true hemispheric shape, also built by Alan Young and placed atop the east tower of Alexandra Palace in January 1987, blew off during the Great Storm, despite being tied down. My cleats had not been fitted! The trouble with tie downs is they work loose with continuous buffeting, until the dome is derailed. Once that happens it is no longer windproof and suction produced by the Bernoulli effect causes sufficient lift to overturn it.
To motorize the dome all that was needed was to fasten a drive belt gear and a torque limiter/scroll clutch to one of the contact rollers and drive it with a right angle gear motor. I chose an appropriate dome rotation rate of one revolution every 4 minutes, and worked out the acceleration/time diagram from there, determined the torque and speeds needed and selected a suitable gearmotor from a manufacturer's catalogue. The 12VDC gas recombination power pack and the powerverter were placed on a framework, and their weight helps force the drive roller down onto the railtube and cause it to maintain hard contact. The torque limiter on the drive belt gear is essential. It prevents the motor stripping the bronze worm wheel in its gearbox. When the motor's windings are first energized at start up a current surge is produced which can overload the drive gear. Also if something prevents the dome from rotating, the drive belt gear simply over-ratchets.
The simplest way to drive the shutter is the direct way, by an endless nylon cord system via a windlass. I intentionally avoided selecting steel cable because it hasn't sufficient give, and it is noisy. Instead I chose thin nylon climbing rope. It has the decided advantage that it stretches under tension. This is very important because it allows the motor time to accelerate and decelerate. If there were no give in the cord the drive gear loadings would rise at start-up and shut-down which would then require a much more powerful and expensive gearmotor. The shutter travel is both limited and the direction toggled by waterproof limit switches. It is not possible to accidentally drive the shutter against the end of its travel.
For discussions on dome designs go to FORUM
The hour and declination axes are fitted
with worm and wheel drives. The hour angle wormwheel is fixed
and the worm, which is mounted on an eccentric by which it may
be disengaged, is carried round wit the declination axis. The
declination wormwheel is located on the counterweight end of the
axle and is an integral part of the declination setting circle.
The worm is also mounted on an eccentric, by which it may thrown
out of engagement.
The axes may also be slewed by stepper motors at up to 5ºper sec, in RA and 7ºper sec. in DEC, and also driven at apparent sidereal rate. The motion in each axis may be controlled to either compensate for differential refraction, or to follow a solar system object by offsetting at the known differential rate in hour angle and declination. Because the equatorial is aligned to the true pole, declination drift due to differential refraction is greater than for an equatorial aligned to the refracted pole, as recommended by King. [King, Edward Skinner, Sc.D., A Manual of Celestial Photography (1931). King recommended elevating the instrumental pole to the refracted pole to minimize errors in following at apparent sidereal rate. (See also Sky & Telescope Vol.78,5,p.538-544, Nov.1989.) The equations King employed for hourly rates of change due to refraction were derived from a paper published in Monthly Notices of RAS Vol.LVII No.2.p.50. On the inequality in the apparent diurnal movement of stars due to refraction, and a method of allowing for it in astronomical photography by Prof. Arthur A. Rambaut, M.A., D.Sc. King modified them to account for the instrumental pole being aligned to the refracted rather than the true pole.]
The principal difficulty of driving an equatorial telescope with stepper motors is that the difference in drive frequency from sidereal to slew rate is too great for the motors to accommodate. If a slew frequency is chosen within the motor's range then the step angle is resolved at the focal plane when following at sidereal rate. Alternatively, if a step angle is chosen well below that resolvable at the focal plane, the motor will be unable to slew the telescope at an acceptable rate without stalling because of the extremely high frequency needed and the concomitant output torque reduction.
The conventional way of avoiding these limitations is to abandon stepper motors altogether and use either a closed loop DC servo system or Selsyn drives, which for the uninformed pose a logistical minefield. The difficulty is avoided accumulated errors affecting field acquisition, and in venting or ducting away heat from the servo system.
A comparatively simple way of avoiding all these difficulties is, in the case of the hour angle drive, to retain the single phase synchronous gearmotor as the sidereal drive and use it in line with a dual shaft stepper motor that is used for slewing and coarse adjustments.
The stepper motor is coupled to the worm direct by a bellows coupling and the synchronous gearmotor coupled to the shaft extension via an electromagnetic clutch. When slewing, the synchronous motor is declutched, and once the object field has been acquired the stepper motor windings are de-energized, and the clutch re-engaged, enabling the synchronous motor to continue driving at apparent sidereal rate. Any fine rate corrections are then made using a programmable variable frequency oscillator (VFO).
The declination drive uses a similar arrangement, but fine motion is provided by a high resolution stepper motor coupled to a planetary gearhead.
The slewing/coarse adjustment stepper motors are 200 steps per rev frame size 32, two phase hybrid, driven in half step bipolar mode. Because of the high polar moment of inertia of the Calver (4.8 kg.m.s2) torque of about 2.5Nm at 1kHz is needed for accelerating the telescope to slewing speed from rest. The PSU's are switch mode constant current, 40V 5A for the drivers and 5V 5A for the logic. This makes the controller relatively immune from mains sag and switching transients. The frequency is ramped; the ramp time may be adjusted using a 1Mohm trim pot to suit the rotor and drive inertia. It is set to match the motor when operating at its highest usable running frequency.
The oscillator frequency is fixed by a capacitor across an LM324N op amp, the variation being 2.5 Hz per pF, and adjusted using a CMOS4040 12 stage binary divider. The value 1220pF fixes the input frequency to the driver at 2048 Hz. Increasing the capacitance across the op amp reduces the input frequency.
The slew frequency is selected by switching between pin outs on the divider and the control pin on the driver. The frequency may be reduced whilst running, but any increase at the high end can stall the motor.
Extremely low frequencies may be selected for clocking the DEC fine motion stepper motor for drift compensation by switching in a tandem divider using a 4011 NAND gate.
Dual axis control is facilitated by joystick operated microswitches alongside which are BCD decade rocker switches to alter the frequency. The CMOS 4040 dividers, 4067 multiplexers, 4011 NAND gate and UCN5904B two phase bipolar drivers are housed in a handpaddle with the switch gear. The PSU's, oscillators and the driving transistors are mounted separately on a large heatsink away from the telescope, beneath the observing floor.
The dividers may be controlled from a PC board and an open loop control system employed. The winding field will collapse sufficiently rapidly to enable the stepper motor to be used as an encoder to update the hour angle whilst being rotated by the synchronous motor, and the matched frequency ramping will ensure that a significant number of pulses will not be dropped during slewing, enabling a fairly precise slew to pre-specified co-ordinates.
At present I am designing a control system upgrade, around 5 phase steppers with a 27 arcmin step angle, and a planetarium package called SkyChart2000, and a Tangent Instruments compatible BBOX controller.
Differential atmospheric refraction causes the hour angle of a star to be diminished as it approaches upper culmination, to a continually decreasing extent, and for the apparent position to be equally effected as the star increases in hour angle west of the meridian.
[In his attempts to reduce the amount of guiding needed in following for longish periods King was anticipated by Davidison & Hinks (Monthly Notices of RAS Vol.LVIII No.1.p.4 On the Apparent Diurnal Motion of Stars in relation to the Adjustment of the Polar Axis of a Telescope by C. Davidison & Vol. LVIII No.8.p.428 On some attempts to counteract by instrumental adjustments certain effects of refraction in stellar photography by Arthur R. Hinks BA) All the equations employed in drafting rating tables were derived using trigonometric approximations in order to facilitate reduction.]
The rational of elevating the instrumental pole above the true pole to counteract to some degree the effects of refraction is valid only where the declination axis is not fitted with a drive capable of affecting accurate drift compensation. Where the drive rates of both Axes may be preselected, and the instrumental pole is brought into coincidence with the true pole, it becomes possible to not only drive the telescope at apparent sidereal rate, but also to follow solar system objects whose differential motion is known.]
With the equatorial aligned to the true pole apparent diurnal motion occasions a significant part of the difficulty in making a telescope follow accurately for long periods of time where the sidereal drive is regulated to a constant rate.
The sidereal drive is a Crouzet synchronous gearmotor operating at 240VAC and designed to run from a 50Hz supply frequency. The rate is controlled by a 1MHz crystal based VFO using a programmable frequency selector.
For zenith distances less than 85º, the change in drive frequency needed to take account of apparent sidereal motion in hour angle changes sufficiently slowly to permit a frequency to be selected appropriate for quite a wide span of hour angle, in most cases thirty minutes or more.
To determine the appropriate frequency for a given declination and hour angle the following equations are employed, which I have derived from Tatum et al [Tatum, J.B., Hankanen, H.N., & Ackroyd, E.E. 'A slow-motion crosshair drive for long-exposure photography of fast moving objects' J.Br.Astron.Assoc.,97,2.p.95(1987)]
I have taken as a working value of A dH.dt, 1.015 mean seconds per hour. Although the equations are tedious to reduce by hand they may be readily solved on a PC/MAC.
For the general interest of those wishing
to utilize this method of controlling a synchronous gearmotor
to match the apparent diurnal motion, I have prepared an Excel
For discussions on drive control systems go to FORUM Return to Brayebrook Observatory Top of Page Return to Telescope Drives Return to Home Page