Brayebrook Observatory Log Brayebrook Observatory Log Brayebrook Observatory Log Brayebrook Observatory Log Brayebrook Observatory Log Brayebrook Observatory Log Brayebrook Observatory Log

Changing the Calver's drives from dc / ac synchronous
to a computer controlled stepper system


AWR STEPPER DRIVE IMAGES

BACKGROUND:

When I acquired my 10-inch Calver in December 1981 there was no extant
drive. There were slow motions, and the RA & DEC wormwheels, but no
drive of any description.

I had Ron Irving recut the wormwheel teeth because they were so badly
damaged, and also had him machine new stainless steel worms, and build
new worm support brackets because the surviving parts were so badly
rusted.

Reference to Calver's catalogues in Ron Irving's possession indicated
the original drive would have been a ball governed, regulated clockwork
affair, powered by a massive falling weight. There was no point in
attempting to build another. It would only drive the RA shaft, and
provide no means of making DEC adjustments.

I decided to use Crouzet motors; an ac synchronous motor with a 30kg.cm
rated gearbox for the RA worm, speed 1/3rpm, and a 12Vdc reversible
motor with a 1rpm gearhead for the DEC worm. The RA motor was coupled to
the worm via a bellows coupling with a friction clutch set at 10kg.cm.
The DEC motor was coupled to its worm via a small Reliance Gears
friction clutch coupling, adjustable between 0.8 - 3.0kg.cm. The idea
being that the worm could then be turned by hand with an extension
handle connected to the opposite end via a Hooke's joint.

With the guidance and assistance of Terry Platt, in 1984 I constructed a
VFO and a switched mode 12Vdc psu to power each of the motors. This
worked satisfactorily, to a point.

I lapped in the RA worm and wheel to minimize periodic error and
backlash, but I saw no need to do so for the DEC worm & wheel. The VFO
frequency could be set to anything between 40.000Hz and 59.999Hz by
means of thumbwheel switches. The thumbwheel settings could be seen in
the dark via a slaved LED display immediately above, built into the
handset. Nominal sidereal rate was 48.465Hz and mean solar rate was
48.333Hz.

The telescope tracked well, but correcting for DEC drift due to
differential refraction proved tiresome and awkward. The DEC worm was
very tight in some parts of the wheel, and slack in others. The Reliance
friction clutched coupling also proved to have slightly too low a torque
rating. Backlash was also much greater on the DEC worm and made field
centering somewhat hit and miss.

Pointing the telescope was easy, especially on objects of known
declination where Hargreaves' method could be employed. Frank Hargreaves
recommended this method for large telescopes within domes. You point the
telescope roughly where you know the object ought to be, set it
accurately in declination, lock the telescope in declination and then
station yourself at the finder and sweep in hour angle until the object
is seen in the finder field.

Starting tracking from this point also proved to be a hit and miss
business. The reason for this was because the wormwheel on a No.1 Calver
is fixed, and the worm, connected to the declination head, travels with
the telescope and dec axle, around the hour axle and fixed wormwheel.

To start tracking, the worm had to be engaged into the wormwheel. Both
RA & DEC worms are carried on swinging arms controlled by cams. These
arrangements used to be referred to as "dog clutches" from their earlier
use on treadle lathes. When engaging the RA worm, unless by happenstance
the worm and wheel teeth were in precise alignment, as the worm bedded
into the wormwheel teeth, there would be a slight shift in the
telescope's hour angle causing the image to move away from the centre of
the field, and at times, particularly when using a medium power, out of
the field altogether. It then became a tedious chore to return to the
finder and use the VFO's RA overide to recentre the object, and then
refine the adjustment at the eyepiece.

Because the VFO overide had a range of only ±10%, this procedure could
sometimes take several minutes. Obvioulsy these limitations made
repointing the telescope tedious.

What was evidently required was a means of slewing the telescope and
then being able to either centre or track, using electric motors. The
question at the time though was the optimum way of doing so, given the
cost of stepper motors or DC Servo motors, and their respective control
systems, in the 1980's.

A LITTLE THEORY:

I began looking at ways of building a viable system in 1988/89.

DC Servo motors were immediately ruled out. They were far too costly;
far too bulky for the torque rating needed (120Ncm), and the
tachogenerator feedback loop control needed to slew the telescope to
specific coordinates utterly fiendish in its complexity.

So I began considering stepper motors. The initial problem concerned
their inherently lower dynamic range. The most affordable stepper motors
were 200 step per rev 2 phase bipolar hybrids. However they would not
rotate at much faster than 6 revs per second without stalling, whether
driven in full or half steps. This corresponded in my case, about the
polar axis, to a max slew rate of 4.6 degrees per second, and about the
declination axis to a max slew rate of 6.3 degrees per second. I tested
the motors, frame 90 Vexta procured from Oriental Motor, using a
breadboarded chopper cct built for me by Terry Platt in spring 1989.

The other problem was heat dissipation. I roughly calculated that 60
Watts per phase would be needed, and that the phase voltage would need
to be roughly 40Vdc to get the slew speeds I was after. Most of this
energy is unused during RA track and needs to be dissipated through a
resistor dropper cct. Each of the resistors would have to be wirewound
with at least a 50 Watt rating, and have to be of the type which may be
bolted to a heat sink. At the time such resistors were quite expensive,
as indeed were the Power MOSFETs.

The next consideration was track step angle. In full step mode the RA
track step angle would have been a massive 13.966arcsecs per step, and
the DEC step angle 18.947arcecs per step. Neither are acceptable, either
for smooth RA tracking, or DEC adjustment when guiding.

If the RA track step is to be to all intents and purposes undetectable,
the step angle must be no more than half Dawes' limit, or for a 10-inch
aperture, 0".228arc. But how to reduce it? In RA the reduction would
need to be roughly a factor of 60, and in DEC a factor of 80. I did go
as far as purchasing a planetary gearhead for the DEC motor, and laying
out a design for a two motor DEC adjust and slew system using a bevel
gear differential. The problems though grew as the design matured. The
difficulties of backlash and the weight and the complexity of the
support bracketry simply made this solution impractical. Also there
simply wasn't adequate room to fit a similar setup to the RA worm.
Enough, back to the drawing board.

I did for a short while take a close look at a microstepping controller,
but at the time 6bit translation table IC's were prohibitively
expensive. A 6 bit translation table enables a two phase stepper to be
characterised for 64 microsteps per step driving. Each microstep would
then be; in RA: 0".218arc & in DEC: 0".296arc, which would have been the
ideal solution. Except I could neither afford it, nor find anyone able
to build the controller, and my knowledge of programming did not extend
beyond Basic!

And so the project was shelved, and as things transpired, I was not to
return to it until 2000!

DESIGN: A new approach to stepper motor driving

When I returned to the problem 11 years later several companies were
offering commercial custom solutions. After spending roughly six months
comparing various systems I decided to use the services of Alan Buckman
at AWR Technology, based in Deal, Kent.

One of the big difficulties in driving a heavy telescope with stepper
motors is their inherently restricted dynamic range. I wanted to track
with a step angle roughly half Dawes' limit, yet slew at up to 5
degs/sec, so the frequency range needed to be 1200:1.

In order to make an appropriate motor selection I firstly determined the
weight of the moving parts and their polar moments of inertia. Trueblood
& Genet provide a few useful formulae extracted from an engineering
handbook (ref p31 Microcomputer Control of Telescopes, Willmann Bell,
1st ed). I prefer to use both the Machineries Handbook published by the
Industrial Press, and Low's Pocket Mechanical Engineering Handbook
published by Longmans (now out of print). I also prefer to work in
imperial units because I have a feel for rightness or wrongness of the
result that I do not get when working in ISO units. If I need to compare
my answers with metric values I make a conversion afterwards.

The moving parts weigh 745lbs:10ozs, of which the :

tube assembly weighs 256lbs:13ozs;

the declination axis, saddle and setting circle and wormwheel and drive
weighs 125lbs:4ozs;

the counterweights weigh 243lbs:3ozs;

the polar axle including the RA wormwheel, setting circle and the slow
motion wormwheel weighs 120lbs:6ozs.

Calculating the polar moments of inertia of mass about the three axes;
the mechanical axis of the tube, the polar axis and the declination
axis, is tedious, but it is not arithmetically difficult. What has to be
done is simple; each individual item has to be weighed or have its
weight calculated, and the offset of its cg from the relevant axis
measured or calculated, and the appropriate inertia formula used to make
the calculation for that part. The parts you need to consider typically
include the tube, the primary cell, the secondary cell, the finder, the
guide 'scope, the rackmount, the dec axis, dec c'wts., dec circle, polar
axis, RA circle. You can neglect the ra and dec worm assy's. The
calculations need to be carried out for each item about each of the
three axes, and the results summed for each axis in turn.

To give you a feel for what sort of result you might expect, the
smallest polar moment of inertia of mass will be about the mechanical
axis of the tube, the greatest will be about the declination axis. In
the case of my Calver I obtained:

Polar Moments of Inertia of Mass:
about the telescope tube axis: 4.61ftlbsec^2
about the polar axis: 52.57 ftlbsec^2
about the declination axis: 80.56 ftlbsec^2
[notice the units - foot x pounds x seconds squared ]

These values can then be converted to a load inertia (the inertia that
resists acceleration forces) by multiplying by the acceleration due to
gravity (32.2ft/sec^2).
[the resulting units then become foot squared x pounds]

This method avoids the pitfalls of either slug or poundal units.

The other thing that needs measuring is the RA & DEC wormwheel ratio. My
RA worm reduction is 464:1 and the DEC worm reduction is 342:1.

Once these values have been determined they can be used to calculate
the torque requirements for slewing. The <torque calculation> I made is
based on the mechanical properties of the drive system. It takes account
of the mechanical efficiency of the worm drive; the rubbing friction
losses, and the inertial loading caused by a given acceleration.

My Calver's worms are 1/2"-12BSW, and the wormwheel tooth width is 3/8".
The worms are made from Immaculate 5 stainless and the wheels from PB2
phosphor bronze. At track the rubbing speed will be 0.05 ft/min and the
stiction coefficient 0.444; the coefficient of friction 0.065. In the
worse case scenario, when accelerating the 'scope from track in RA to
slew the torque required to overcome inertia is 1.3inlbs. Combined with
friction this increases to 3.7 inlbs. About the DEC axis the torque to
overcome inertia is 2.7inlbs and combined with friction this increases
to 8.8inlbs. (A useful rule of thumb when performing these calculations
is that the friction torque is typically twice the inertia torque for
plain shaft bearings).

From Newton's third law of motion, every action has an equal and
opposite reaction. The torque needed to overcome the inertial load
resisting the acceleration force is therefore reflected as an inertial
load on the stepper motor. It is fortunately reduced by the square of
the drive ratio, in my Calver's case by 215296:1 about the RA motor, and
116964 :1 about the DEC motor.

I needed motors capable of supplying a pull out torque of twice the
combined friction and inertial loads, at a pulse frequency of 1.2kHz. My
initial selection was a Vexta 5 phase frame 60 model UPK564BJY, which is
a rare earth magnet dual shaft. Alan Buckman however was not keen on
this choice because of the difficulties of writing a 6bit translation
table for a 5-phase motor, and having to design a 5-phase driver.
So I compromised on a 2-phase rare earth magnet hybrid frame 60
Sanyo-Denki model 103H7823-0710 dual shaft (i.e. the motor has a
shaft extension for use with an external encoder). These motors have
a current rating of 3amps per phase and a rotor inertia of
0.84E-04kg.m^2 (6.256E-03lbft^2). I procured them from EAO Highland in
West Sussex on a short order delivery of two weeks at a cost of £105.

There should be an approximate match between the load and reflected load
inertias. The load inertia during acceleration of the RA worm is
1692.75lbft^2, and of the DEC worm 2594.03lbft^2. Divide these figures
by the square of the wormwheel ratios and you obtain:
about the RA stepper motor: 7.862E-03lbft^2
about the DEC stepper motor: 22.178E-03lbft^2
This indicated that the RA drive would be able to accelerate faster then
the DEC drive, but both were close enough to the rotor inertia
(roughly 1:1 & 4:1) for the motor selection to be viable. Adding the inertia
of the worm would tend to improve the balance between load inertias,
but if there were still a big discrepency (of the order of a magnitude),
adding a damper disc to the shaft extension can solve the problem.

Now to the step sizes and pulse frequencies. The motor selected is 200
steps per rev. Alan Buckman chose to drive the RA motor with a split
phase, and the DEC motor with a two phase microstepping controller. The
phase voltage at first was 24Vdc, and the current 2.8 amps per phase.
The RA step size at 64 microsteps per step is 0.217arcsec (68.9
microsteps per second sidereal) and the DEC step size 0.295arcsec. Max
slew is value clipped at 16bits, less 1, or 65535 microsteps per second.
In RA this equates to 3.99 degs per sec. and in DEC 5 degs per sec.
During slew the resolution decreases progressively from 64, to 32, to 16,
to 8, to 4 microsteps per step, and finally to half steps, and ends up at
about 2000 half steps per second at max slew.

The prototype drive controller was coupled to a Farnell 200Watt 24Vdc
switch mode voltage/current regulated psu. However the drivers were not
sufficiently powerful to slew the motors without mid range resonance
causing them to stall at about 3 revs per second.

So at my insistance the psu was swapped for one with a 250 Watt rating,
but this did not improve matters, even when the phase voltage was
nominally increased from 24 to 29 volts.

So a third Farnell psu was procured, one with a 350Watt rating,
adjustable between 24-48Vdc. The phase voltages were set to 39 volts,
and the current maintained at 2.8 amps per phase.

However this meant that a second psu was needed for the cooling fans
which operated off 24Vdc and the logic cct's which are 5 volt TTL. I was
fortunate in being able to procure an ex MoD ordered Weir-Lambda 100Watt
switchmode voltage/current regulated psu with a variety of different
output voltages and currents for a modest £10, from a company called WCN
based in Southampton. This psu supplied 24 volts to the fans in the
resistor dropper box cooling system, and the 5 volts needed by the
Intelligent Handset, the joystick, and the drive controller.

There was now sufficient power to slew at the maximum rated frequencies
without the rotor's stalling. The acceleration or ramp rates in both
axes were set at 60kHz/sec, which translates to an acceleration time of
3 seconds. Deceleration is a mirror image of acceleration, except during
an emergency stop, more on which later.

During RA track the RA motor alone is only consuming 30 Watts. However
the power output from the drive controller to the RA motor is 150 Watts
and to the DEC motor 200 Watts, therefore during RA track 320 Watts has
to be continually dissipated. This is achieved through a bank of large
wire wound resistors bolted to a 10"x4"x3/8" dural plate, in turn bolted
to a pair of huge 3.4 deg C/Watt heat sinks. The heat is ducted away by
four 40mm 8cuft/min PAPST fans.

A LITTLE MORE THEORY:

In operation stepper motors always exhibit a drop in torque with
increasing pulse frequency. To compensate for this the current per phase
needs to be factored according to the microstepping rate. This is termed
"characterisation".

Another unwanted property of the stepper motor is something termed
"mid-range resonnance". As the pulse frequency is ramped up, the rotor
begins at some point to oscillate between the field coil poles. There
are several tried and tested ways to reduce mid-range resonance.

Firstly you can deliberately introduce a phase lag in the bi-polar field
coil supply currents. This however is technically difficult to achieve
when the motor does not run at a constant speed which is the case during
a ramp up to slew.

Another method is to use a split phase drive. Instead of driving each of
the two phases separately, and in synchronisation, a single phase supply
current is split into two, each half used to drive one of the pair of
phases of the windings. It is easier to modulate this type of drive, but
there is a trade off because the supply current per phase is lowered by
a factor of 1 over root 2, or to 70% of the two phase equivalent. The
split phase though enables the microstepping controller to drive the
rotor more smoothly, an important consideration for the RA drive.

Another, and until now untried method, is to inject the gap between the
rotor and the stator poles with a magnetic liquid called a FerroFluid. A
FerroFluid is a suspension of microscopic (10 micron) Ferrite particles
in a surfactant. The surfactant is a solvent and lubricant used to
prevent the Ferrite particles from clinging together. The principle of
using such a liquid in the rotor gap is to introduce viscous drag and
hence damping which helps reduce mid-range resonnance.

The data sheets that come with the FerroFluid provide a calculation page
for determining the precise volume of liquid needed to fill the rotor
teeth and the gaps between the rotor stacks. This method will only work
with rare earth magnet rotors. If you withdraw the rotor from an AlNiCo
magnet stepper it will loose its magnetism almost totally and be
rendered useless. The rotor should also not have the gaps between its
teeth filled with epoxy resin, as many cheaper steppers have.

The Sanyo-Denki motors have a laminated stator, so the outside must be
sealed. I used ICI Permabond two pack general purpose epoxy to seal the
sides and to bond 9.0 deg C/Watt heat sinks. To seal the end flanges I
used a proprietory gasket sealant obtained from a local tool merchant.

FerrorFluid behaves in a non-intuitive manner. It does not obey the laws
of gravity when in contact with a magnet, but the laws of magnetism. As
the fluid is injected onto the rotor it immediately migrates all around
the rotor teeth, filling in all the gaps. Once the fluid has been
injected the rotor is fed back into the stator and the end flanges
refitted and the entire unit bolted back together again.

If you have done a good job sealing the stator laminations and gasketing
the end flanges the FerroFluid will stay inside the stator. If not it
will migrate between the stator laminations, or around the end flanges,
by what superficially seems to be capilliary action, but isn't. It is
attracted to magnetized surfaces, and because the Ferrite particles are
only 10 microns or thereabouts in size, the liquid can migrate onto the
external surface of the stator by getting between the laminations.
Sealing the stator's external surface is crucial to the success of this
technique.

I performed some interesting experiments in conjunction with Alan
Buckman at his workshop. George Thompson, the Sales Director of
<FerroTec UK> kindly supplied two 15ml bottles of SMGL14-100/500
FerrorFluid which I injected onto the the withdrawn rotor using a
syringe from an inkjet refil kit. The fluid, although it flows has a
very high viscosity in the presence of a magnetic field, and low vapour
pressure. You must use a syringe with a wide needle otherwise it will
not get drawn into the cylinder from the bottle because the vacuum
produced within the cylinder as the syringe piston is withdrawn will
produce insufficient pressure.

The FerroFluid also reduces the detent torque, in my case by roughly two
thirds, which results in a much smoother rotor action, particularly at
low pulse frequencies. The "detent torque" is the torque needed to
advance the de-energized rotor across an adjacent pair of stator poles.

The torque needed to overcome the rotor and force it across a pair of
energized stator poles is termed the "holding torque".

The torque produced in the rotor by the switching of stator pole
currents is called the "pull-out torque". If the torque produced by the
combined friction and interial loads in accelerating the telescope about
either the RA or DEC axis exceeds the pull-out torque, the rotor will
stall. The pull-out torque falls as the pulse frequency increases. At
some point the rotor will begin to oscillate and stall.

You can usually tell when this is about to happen because the rotor
begins to resonnate and produces quite a bit of vibration. The motor
support structure can begin to ring, and the motor begins to whine,
sometimes alarmingly loudly, soon to be followed by a crunching or even
graunching noise. Not exactly something you want to hear, but in
building a stepper drive system for a heavy telescope, almost inevitably
one you will hear before you get things right. It may sound awful, but
the motors will come to no harm. What will be damaged if you do nothing
about mid-range resonnance are the wormwheel teeth. The vibration
induced in the rotor causes gear train vibration called 'cogging'.
Cogging is noisey, and results in severe wear of gear teeth, known as
'fretting', or sometimes, 'fretting corrosion'.

IMPLEMENTATION:

Having built a drive system that worked when setup on the bench, the
time had come to fit it to the telescope. I had Ron Irving machine motor
adaption plates
out of 1/4-inch brass. This was for two reasons. Firstly
the stepper motors could be bolted to the adaption plates which in turn
could be bolted to the worm support brackets. Secondly, in the event of
an electronic failure, the original drive system could be replaced, so
at least I wouldn't be left completely out of action.

The next thing I needed to do was lap in the declination wormwheel. I
cadged three grades of silicon carbide and two grades of cerox from Jim
Hysom; mixed the carborundum powders with heavy grease, and the cerox
with engine oil. I purhased a DeWalt mains reversible hand drill with a
1/2-inch chuck. With the motors removed, but the telescope still in
situ, I then coupled the DeWalt hand drill to the slow motion end of the
declination worm and began the lengthy procedure of grinding the
stainless worm teeth into the bronze wormwheel teeth. I soon got the
knack of controlling the speed of the drill and was able to slew the
telescope to and from from east to west smoothly. Periodically I would
stop and brush on more lapping paste.

I did this for a period totalling about 48 hours altold, going through
carborundum grades 120, 280 & 400 and 225 & 125 micron cerox, before
finishing off with a long run using Brasso. It was easy to tell when the
Brasso had run dry. The worm began to smoke!

Afterwards the telescope could be slewed smoothly and accelerated more
rapidly, from either east to west, or from due south to near the zenith,
without any sign of the declination motor stalling.

It transpired that the main cause of the declination wormwheel being
stiff in one place and slack in another was because it was not truly
perpendicular to the declination axle. When I clocked the edge of the
wormwheel with a 1/10000-inch reading dti, the deviation from due east
to due west was 12 thou.

Before I began the lapping process only three worm teeth were fully
engaged in the wormwheel. By the time I'd completed lapping in the
declination worm 12 teeth were fully engaged, and the worm had become
noticeably diablo shaped, which is what a hobbed worm should look like.

The torque needed to turn the worm to overcome friction dropped
considerably, and I was able to reduce the backlash because I could seat
the worm much further into the wormwheel before it became tight.

The only remaining task was to bolt on the motors, and install the
various hardwares. Now for the moment of truth. Would I be able to slew
the telescope, compensate for backlash, and initialize the Intelligent
Handset?

After all the preparatory work, which had taken the better part of six
months, it was not perhaps too suprising that it worked from the off. I
also made an interesting independent discovery. I already new that the
telescope was accurately polar aligned on the true pole. It occured to
me that prior to initializing it on a couple of clock stars I should be
able to get the IHS coordinate readout to coincide with the setting
circles simply by pointing the telescope at the celestial pole and in
the meridian, and then performing a single star calibration, inputting
an RA the same as the sidereal time, and a dec of 90 degrees.
And it worked !

The next problem to address was the possibility of accidently slewing
the telescope into an obstruction or the baseplate. There are
observatory control softwares that incorporate excursion limit mapping
routines, but I don't trust software solutions. The first time you
realize something is wrong, is when it is too late. Hardware solutions
can be made failsafe, and can be built so that they rarely fail, if
ever. Hardware motion control limits are better than software. They are
also a lot cheaper.

The primary cell of my Calver, at max slew in both axes, moves at
7-inches per second. The momentum of a 745lb moving mass at such a
velocity is considerable. To prevent collision in the event of a bad
goto command, the primary cell flange has an array of a half dozen
equispaced Moeller-Klockner finger style limit switches RS244-2563 &
RS244-2513. If any one of them touches the pier or baseplate the drive
cct. is broken and the slew ramped down in a second.

To investigate the loads transferred to the worms and the wormwheel teeth
caused by a collision I firstly calculated the angular velocity @
combined slew about both axes, taking a worse case of 5 degs/sec &
4.0 degs per/sec about the dec and ra axes respectively. These combine to
6.4 degs/sec or 0.1117 rad/sec.

The combined polar moments of inertia of mass about both the polar and
declination axes are 130.5 kg.m^2, giving a total kinetic energy [1/2
mass x velocity squared] of 0.815625 Joules.

From my CAD model I measured the radius of action of the primary cell to
be 52".958 and then calculated the velocity @ the primary cell from
v=2(PI)*R/60 = 2 x (PI) x 53 x 1.067/60 = 5.918 ft/min.

I then assumed a deceleration time of 1/100th sec and calculated the
resulting angular impulse. Using imperial inertia values of 80.56
ft.lb.sec^2 about dec & 52.57 ft.lb.sec^2 about ra, which combine to
96.20 ft.lb.sec^2, and the angular velocity value of 0.1117rad/sec, gave
an angular impulse of 10.75 ft.lb.sec.

What would be the result of an iron on iron impact? The factor of
restitution for iron on iron is 0.66. When t = 0.01sec the momentum will
be 1075 ft.lbsf, so with a factor of 0.66, the resultant momentum will
be 710 ft.lbsf.

This torque will be applied at a distance of 37".75 from the ra axis &
40" from the dec axis. The resulting load on the ra worm will be
710x12/37.75 = 225.5lbsf and on the dec worm 710x12/40 = 213lbsf.
Combining the load shared eqaully, the impact load resolves to 155lbsf
per worm.

This load will act on each worm and there are three wormwheel teeth in
full mesh at any one moment. The surface area of tooth engagement
equates to 0.03 in^2, so the shear stress per worm will be 155/3x0.03 =
1722lbsf/in^2. The shear strength of bronze PB2 is 8500lbsf.in^2. A
factor of safety of roughly 5, which is good for this circumstance.

The AWR drive box and the Farnell psu are CE rated and so should be
immune to mains transients, but to be on the safe side I fitted a
supression filter and an RCD cut out. I use the latter to power down the
Farnell psu only. The Weir-Lambda psu is left on continually. This means
the IHS, even if it is shut down, remains powered up.

There is an advantage in leaving it powered up. The Real Time Clock cct
[RTC] is controlled by an oven regulated 32kHz crystal. If the IHS is
shut down and powered down the only power supply to the crystal is
provided by a button cell, and it does not power the crystal oven, which
then drifts off frequency reducing the accuracy of the RTC.

At one stage I seriously considered using one of the serial port outputs
from my Techspan Systems UTC master clock and bipassing the IHS's RTC.
However the RTC when left powered up is accurate to about 1 second per
week. By comparison my Wharton UT1 master clock is accurate to 1 second
a month.

I also considered installing the resistor dropper box under the
observing floor because of concern for the heat it gives off. The
additional large heatsinks and the fans I fitted made that unecessary.

The IHS is needed for initializing the telescope or when performing
goto's, even if the command is sent from a computer. However when
pointing the telescope at bright naked eye objects it is simpler to
steer the telescope using a joystick. I selected a three axis IP66 rated
unit manufactured by CD PRODUCTS, VISTA Ca, c/o QUILLER
in Bournemouth, model HFX44S10, FARNELL PART No.206192.
The x,y motion controls direction and the z-twist axis,
MOVE, SLEW & CENTRE speed selections.
I chose this particular model because it is big enough
to operate in the cold with a gloved hand.

It is possible to interface the drive box through the IHS via a
planetarium software with an LX200 protocol, using a PC RS232 serial
connection. I run Xephem on LinuxPPC2000 on an old 7200 PowerMac.
(I hate PC’s ). However I have made one concession, and acquired a Rock
QuadraXAT PC laptop to use with my SBIG ST237A ccd camera. It was one of
the few laptops made with both USB2 and FireWire400 ports and a DVD
Combo drive, plus the older serial and extended parallel ports plus a
3.5-inch floppy drive (long since obsolete on Macs).

Having said that though the AWR goto protocol is different from the
ASCOM standard so-called LX200 protocol, borrowed from the mid 1980’s
Tangent Instruments ‘B-Box’ controller. Pointing accuracy is better than
±5arcsecs, and tracking accuracy 0”.005/sidereal second.

Additional features include backlash compensation; periodic error
correction; single or multiple star initialization; readout of the
altitude and azimuth of the instrumental pole; readout of either
equatorial or instrumental coordinates for input into a pointing analysis
software such as T Point; read out in RA & DEC or ALTAZ; mean, apparent
or King sidereal track rates; observer’s horizon map; optional automatic
meridian reversal; dome slit azimuth input and dome rotation
synchronization.

ADVANTAGES OF AWR TECHNOLOGY SOLUTION:

If you live in the UK and are thinking of replacing your synchronous RA
drive with a stepper system I can wholeheartedly recommend AWR
Technology
. Alan Buckman has a considerable amount of experience in this
field and had produced several stepper drive systems tailored to
telescopes even heavier than mine.

If you are to get the best out of AWR you must be prepared to do your
homework and not simply rely on Alan to come up with the goods in the
expectation the system will work first time as you anticipate.

The 'standard' AWR stepper drive is either 12Vdc or 24Vdc, and either
1amp or 2amps per phase. If you expect these systems to slew a heavy
telescope, perhaps with stiff plain bearing or nylon slipping axle
clutches, at 5 degs/sec or more you will be in for a big dissapointment.

The system I had Alan build for me is, up to now, the most powerful he
has built. Whether a more powerful one could be built is a moot point
because there is a limitation in the power MOSFETs, and the current that
can be taken by the stepper windings.

One of the decided advantages of making use of AWR Technology's services
is that you can talk to the man who is building your system. I liaised
with Alan continually throughout the design and fabrication phases. I
travelled down to his office in Deal and discussed the engineering
difficulties of what I was attempting to achieve, with his assistance,
in detail. What I did not do was whinge when things went wrong, which
occasionally they did, and just sit on my arse and wait for him to put
everything right, like some fairy godmother.

Yes there are plenty of other commercial solutions out there, but they
do not offer tailor made solutions, they are all overseas, and if
anything does go wrong, or it doesn't come up to spec in the first
place, you are essentially on your own. In fact most of the companies I
looked at do not even offer telephone support, let alone permit you
access to their chief designer!

Alan is continually upgrading the IHS firmware. It is currently at v1.5.
Most competitors do not do this. Most let the piece of kit you purchase
from them become obsolete. When it goes wrong, which if its either Meade
or Celestron it most probably will, sooner or later, what dealer support you
may anticipate under the warranty will have evaporated with the proverbial
morning dew. Your only recommended course will be to 'upgrade'.

How often have we heard that one? Upgrading software is one thing,
upgrading your PC is another, but upgrading either your drive system or
the complete telescope is something else. If you consider Celestron
or Meade to be good manufacturers then please disregard these comments.
If however you have a brain heed my advice and use it. Do your homework,
read up on stepper drive systems, and teach yourself the fundamentals of
drive design. It might look it but it ain't rocket science.

DISADVANTAGES OF AWR TECHNOLOGY SOLUTION:

Compared to some of the more advanced commercial solutions the AWR
controller is unsophisicated. It is not intended to provide robotic
control, or enable the user to execute an observing programme
unattended. The interface between the drive box and a PC also restricts
the use of a more sophisticated computer programme, or suite of
programmmes, that would otherwise make automation possible.
Having said that though, Alan Buckman is working on a robotic
control system
.

Unless you have a purpose built mounting with hollow cable ducts within
the axles, you will inevitably end up with a festoon of cables trailing
from the mount, and also the telescope if you use an autoguider. It is
left to you, the ATMer to sort the problem out. Notice that I have used
coiled leads to the motors. I had to source and order these myself. I
also had to devise a means of supporting the cable at the motor
connector. AWR may provide this service, but only if you ask, insist and
pay up front for it.

Your system, custom built and tailored to your telescope though it may
be, may not work first time. Although Alan will offer advice, and
provide technical support and engineer a solution, it may not be at no
additional cost. Remember you do not simply play a passive role in the
procurement process. You are an active participant, whether you like it
or not, and you are expected to show a certain amount of savvy. The good
news is that at least you will end up with a working system, and that
you can talk to the man who built it for you.

CONCLUSION:

Having had a system which works as well as mine does successfully built
I would not wish to go back to my original synchronous drive,
sophisticated though it was in its day. My Calver is now far easier to
point and control, either from the observing floor or at the eyepiece.
The AWR system has made a big awkward to use telescope easy to use,
almost as easy to use as a small telescope.

This does come at a price. I am not referring here simply to the cost
(approx. £2500). The price I have had to pay is system complexity.
There is much more to think about now, and a steep learning curve.

Future developments include a PEC index plate on the RA worm to make PEC
mapping absolute. I also intend adding HP 8192 pulse per rev
incremental encoders to the shaft extensions to turn it into a closed
loop system. The advantages of a closed loop system over an open loop
system are that, should the motors stall for any reason, say imbalance due
to equipment changes, you do not need to reinitialize the system. I also intend
fitting a system of proximity sensors around the dome rail to provide
dome azimuth information to the drive box. These have already been
procured.

If I had to do it all over again what would I do differently? To quote
the proverbial Irishman, "Well I wouldn't start form here!" What I mean
is that a heavy Victorian telescope with a very long tube made of 16swg
steel, and a German Equatorial with solid cast steel axles in plain
bearings is hardly the ideal setup to turn into a goto system. A better
mounting would be the Torque Tube, which has unrestricted diurnal arcs
at all declinations above the horizon.







Top of Page
Back to Home Page