(with special attention to the LX200 drives and including instructions for repair and "tweaking" of them)

In the process of repairing two LX200 declination drives, I have gleaned some information which seems to fit together and make good engineering sense.  I have designed similar systems and feel it is time to try to come to closure on some of the issues involved in the design of telescope drives.  These drives must meet exacting specifications if they are to reliably point a telescope tube to one arc second.  That is dividing a circle into 1.296 million parts. To attain this accuracy of pointing and to generate a steady motion for guiding the telescope a sophisticated electrical drive system must be used.

Most of the drives used are closed loop control systems including control electronics, the motors and feedback encoders.  The motors are connected to the main drive axes of the telescope tube through a set of gears and a worm. The reduction gears, worm and main shaft often are not inside the control loop but are extensions, through the gears, of the positions of the drive motor shafts.

One source of slackness or looseness in these types of drive systems is the lash in the gear train between the motor and the main drive shaft. Any defects or lash in the gear train is not fixed simply through control of the motor shafts since this part of the system is outside the feedback loop.  However backlash or dead zone in the gear train can be compensated for to some extent with a backlash setting generally  provided by the electronic part of the system.  In the LX200 design this feature is only provided on the declination drive.  In the case of the RA drive some of the most dominant periodic errors can be improved by a drive correction training technique such as is provided in the LX200 design.

There is often no absolute encoder to give the position of the telescope main shafts because of the difficulty of making an encoder of the required precision.  (more about such encoders later)   To keep track of the absolute position of the telescope, the number of turns of the motor shaft must be known and kept track of by a counting mechanism consisting of a computer and encoder on the motor shaft..  This is accomplished as follows in the case of the LX200.  The motor has an optical transducer on the shaft which sends to the control electronics a series of pulses which the electronics counts and compares to a computer generated number.  The number of counts from the motor shaft tells the computer, indirectly, where the telescope is pointing. When the motor encoder counts match those required by the computer the motor is stopped.  This is an extremely accurate way to make the motor shaft turn the correct amount but it does not by any means guarantee that the telescope tube has moved correctly because the of lash and other errors in the reduction gears, worm and main drive gear.

In the case of the LX200 drive, counts are generated by a disk with 90 slots which are measured by two photoelectric pickups.  The reason there are two pickups is so that there is no ambiguity about the number or direction of the count generated by the rotation of the disk.  With only one pickup it is possible to have false pulses generated and the direction of the encoder disk cannot be determined. With two transducers very slightly offset, the transducers will generate a four state pulse code that gives both the number of counts and  the direction the shaft is rotating.  This is a standard encoder technique known as bi-phase coding.  (I will not go into coding schemes here for the sake of brevity.)  At the normal RA (sidereal) drive rate the motor shaft turns once per 8  seconds this yields a count of  45 pulses per second.   So in order to drive the motor at the sidereal rate, the computer generates 45 pulses per second and the encoder is forced to respond with the same rate.  When the motor gets behind the computer applies more current to the motor and speeds it up.  And vice versa.

In a similar way, once the telescope is synchronized with a known position, it moves the telescope to a new position by simply demanding a calculated number of counts from the encoders on the RA and Dec motor shafts.  This is a very simple, inexpensive and accurate method of positioning a mechanism.  One slot movement of the motor shaft encoder corresponds to 1.3333 arc seconds of motion.  There are 11.25 slot moves per second for the RA drive rate.  (the encoder delivers 4 pulses per slot move)  This means that the motor shaft moves with a position precision of 0.333 arc seconds referred to the telescope tube.  It must be remembered that this accuracy applies only to the position of the motor shaft.  Because of the factors mentioned above, the errors in the gear drive are not corrected to this accuracy by any means.  It is certainly not difficult for the computer to generate the required differential position distances in the form of RA and Dec motions or for the Alt and Azm motions in the form of a pulse count. There is a memory and a computer chip of considerable power in the telescope control system.

Now consider the more difficult issue of correcting the backlash in the drives. For the Dec drive there is a backlash correction number which is entered by the user who selects it manually by observing the motion of the telescope when commands are given. The computer simply remembers to add or subtract this number to the appropriate move command.  Thus the seemingly difficult backlash correction is taken care of relatively easily by the computer through the same motor drive counting mechanism.  This system only works if the mechanical lash is symmetrical and consistent.  We will see that this is not usually the case.

The correction for the irregularities in the worm gear are also taken care of by the user through the worm correction facility provided in the LX200 computer. This correction is entered by the user in the form of E and W pushes of the direction keys on the keypad while manually guiding the telescope through the eyepiece.  The computer knows the rotational position of the worm gear by means of a transducer on the worm gear shaft.  For an 8 minute period, one turn of the gear, a total of 200 corrections are entered. This is one correction for each 2.4 seconds. At the nominal RA rate the computer sends 45 pulses per second or 108 pulses in a period of 2.4 seconds to the accumulator which the motor must match. Pushing the E key stops the motor for the pushed period.  This action also must subtract some pulses from the number entered by the computer for that period.  Pushing the W key doubles the speed of the drive.  So this action must add to the number of pulses entered by the computer during the period in question.

The precise algorithm used to add or subtract pulses in the computer for the 200 periods is not known.  The number of pulses added or subtracted during each 2.4 second period is small (probably only 3 or so) it is most likely that pulses are added or subtracted slowly only for the brief periods that the keys are pressed. The details of the algorithm would be interesting to know, but it is not an issue of principle concern here.  It is apparent that the short pushes of the E and W keys that the operator enters to keep the telescope on track during the training period are used to adjust the computer output many times during the 8 minute period of the worm. A sort of incremental averaging takes place which smoothes the motion of the main shaft.  It is probably sufficient, at this time, to know that Meade has provided a very nice scheme for correcting the worm drive rate in 200 increments per revolution and that the user can train the worm through setting up the smart drive mechanism with considerable accuracy.  Typical rate errors of 50 arc seconds can be reduced by a factor of 10 or more.

Meade states, in their instruction manual, that the Smart Drive can be trained and retrained as often as the operator likes and that a sort of averaging takes place. The computer chip in the telescope could easily be programmed to do a very nice algorithm that responds to the normal rate and the key pushes and enters in a file an appropriate correction which is then used to drive the motor/reduction gear/worm as necessary for smooth RA. motion.  We do not know the details of how the worm rotation is corrected.

The main gears in the LX200 have 180 teeth.  One turn of the worm is 2 degrees of motion of the optical tube.  There is a 60 to one reduction in the gears which means that one turn of the motor shaft causes 120 arc seconds of motion of the optical tube.  The optical tube must be aligned to a known star and the computer told the position of the star.  This action sets the synchronization of the optical tube and the computer.  From this point onward the telescope moves in a differential mode.  For example, the “goto” command tells the motor to make the required number of turns so that the tube moves from where it is pointing to where it should point. The accuracy of each successive pointing operation is dependent upon the accuracy of the previous one.  It is suggested in the operating manual that with critical alignment of the telescope, the “goto” commands will be accurate to 2 arc minutes or better.  This is certainly believable.

This type of position control is differential position control as contrasted to absolute position control..  Absolute position control would require an encoder on the telescope shaft itself.  In this case the control system would know the exact pointing direction of the telescope.   The differential encoder measures the amount of motion from one point to the other.  That is why it is necessary to establish a precise pointing direction after the telescope is turned on.  This is normally done by one of the techniques described in the operation manual.  From that point on, the telescope keeps track of its pointing direction by counting the pulses from the encoders.  Such a differential system can work very well if the loading on the mechanical system is well balanced and symmetrical and the drive does not slip or make an error at anytime during operation..

The encoder and electronics can easily count pulses and keep the motor shaft synchronized to the computer commands.  The principle problem with a system of this sort is that the main pointing shaft is not inside the control loop.  It would be if the
encoder were directly on the pointing shaft.  Unfortunately, an encoder accurate to 1 arc second would have to generate  369 X 60 X 60 =  1,296,000 pulses for one revolution of the declination shaft.  This is quite impossible.  If the worm intersected a main gear with 360 teeth, one turn or the worm would have to be divided into 3600 divisions.  If one required pointing to 1 arc minute for the latter system the encoder would only have to have 360 divisions. Either encoder technology is well within modern design capabilities.

So there are several ways to effect accurate computer controlled pointing. The LX200 system is reasonably good and inexpensive to implement.  But the gear reduction and worm drive must be quite accurate mechanically.  Other schemes can be devised but may be more expensive to effect.

Return to Beginning

Tweaking and Rebuilding the Dec Drive

The following are my experiences while rebuilding two declination drives on the two LX200s that I own. One is a 10" and the  other a 12".   Both telescopes developed large amounts of  lash and showed retrograde motion in the declination drives. This made them unsuitable for auto-guiding for imaging.   The instructions indicate that some lash is to be expected when changing direction; doing North to South reversals. The manual says that values of 2 to 4 seconds are normal.  Since the declination drive speed in normal guiding mode is 15 arc seconds per second of time, one can correct for the delay by entering a number into the computer to correct for the lash.  Nominally, the number entered is 15 times the number of seconds of delay. This number is entered once and need not be changed.  The number entered clearly corresponds to the number of arc seconds of mechanical lash in the declination drive.  Technically the lash should be entirely in the gear reduction train and should be quite symmetrical.  Since the maximum lash that can be corrected is 99 arc seconds, the actual delay time must be less than 6.6 seconds.

Many users have found this correction does not always work. Often, users have found much larger delays and delays that depend upon the position of the declination axis, the direction of the reversal, the loading on the telescope and many other elements.  Hysteresis, dead zones, of up to 15 seconds have been reported. I too have found all of the above effects. The delay can vary from a few seconds to 10 or 15 seconds depending on many factors.  If the delay were in the motor/reduction gear train as expected, it should not vary  much since the "winding up" of the gear train is similar in either direction. Loading effects on the declination axis are not strongly reflected back into the gear train because of the almost unidirectional transfer
of forces through the worm gear. Typically it is not possible, with a low pitch worm gear, which this is, to turn the worm at all with any amount of torque on the main gear.  Breaking of the gear would likely take place first.  However, loading of the main gear, as by unbalance of the optical tube will greatly increase friction between the main gear and the worm.  Thus with an unbalanced optical tube, considerably greater drive force through the reduction gearing is necessary.

A goal of this study of the declination drive is to determine the sources of excess declination drive lash and to eliminate them so that the drive will come up to the specifications required for the declination lash correction to work properly.  The main gear and its clutch mechanism are impressively well build and quite strong.  It is nice to have a 146 mm diameter declination gear since it should provide good pointing accuracy.  It might be noted that the same size drive is used on the 8", 10" and 12" LX200s.  So while the gear is adequate for the two smaller telescopes it is somewhat marginal for the 12".  Many users have discovered that end play in the worm gear mounting contributes to the reversal delay. This is certainly an important effect.  With the given characteristics of the gears, a quick calculation shows that 1 arc second of motion of the telescope tube corresponds to only 0.355E-3 mm of axial motion of the worm.  This is a required tolerance that is incredibly tight.  Thus end play in the worm drive must be eliminated as completely as possible.  The worm must be "snug" in its bearings and the entire drive platform must be snug in its pivot mount.  Adjustments are provided for in the Meade mount via an end screw on the worm shaft bearing and a screw on the platform mount bearing that can be adjusted.  Both should be tightened enough to eliminate all possible end play. The end play results directly in rotary motion of the main gear and thus in the pointing accuracy of the telescope.  I found the mechanism in one of the telescopes well adjusted (nice and tight) but the other had significant end play.

There is however another source of play between the worm and the main gear.  This is the radial motion of the worm with respect to the main gear.  For some reason, the worm in this design is on a "floating" platform which allows for motion of the worm radial to the axis of the declination drive.  It is hard to understand why this "floating" action is as large as it is. No other, of about a dozen worm/gear drives I have inspected, has an action that allows for the large motion that this one does.  If one carefully measures the "float" action one finds that the worm can move as much as 0.5 mm radially. The amount of motion depends upon the direction of reversal and also on the accuracy of balance of the telescope about the declination axis. If the full "float" motion of the bearing platform is allowed, it results in 0.08 mm motion of the main gear edge which is 220 arc seconds of motion of the telescope tube. This is a motion ratio for the worm to main gear surface of  7:1 which seems a bit large for this type of drive.

The forces upon the worm that push it partly out of engagement with the main gear are caused by the friction in the declination bearings plus the forces due to unbalance of the telescope tube.  This seems to be one source of the varying delay in reversal operations.  It is also the source of retrograde motion.   After evaluating numerous operations of the drive with different unbalance loads, it became clear that the amount of "float" is large, irregular and not necessarily repeatable.  Motion of the bearing platform was measured with a precision dial indicator and varied from 0.025 mm with the tube well balanced to the full 0.5 mm with a substantial unbalance. In terms of tube motion this amounts to about 11 arc seconds with the tube balanced to 220 arc seconds with substantial unbalance. The amount of unbalance used was 0.1 Kg-meter.  Again both drives behaved in a similar manner.

While the smaller of these "floats" can be compensated for, the larger cannot since the declination lash correction is 99 arc seconds maximum. There is also a time factor involved with the resettling of the "floating" platform to its new stable position.  In addition, the platform takes on a different position when the tube is being driven compared to what it takes when it is allowed to rest. This settling of the platform position after ending a motion cycle causes the tube to drift of the order of 2 to 20 arc seconds.  The drive mechanism seems to sort of relax after being exercised. Both drives did similar things but each to a different, and unpredictable, degree.

The computer based correction scheme would work with constant mechanical relaxation, it does not work well when the relaxation is variable and erratic. As well as being dependent upon the reversal direction, the "float" and relaxation motion was much smaller for a telescope tube that is perfectly balanced about the declination axis because then only the friction forces and acceleration forces must be overcome.  There is a small spring under the floating bearing platform that presses the worm against the main gear. If this spring is strong enough, it can keep the worm pressed properly against the main gear as long as the unbalance is small and the forces required to move the telescope tube are small.  As the unbalance gets larger, the spring no longer maintains good contact between the worm and the main gear.  The concept that the telescope tube should be kept unbalanced to keep the drive wound up in one direction is not valid in the case of the worm design. Unbalance only increases friction in the drive and requires greater drive force.  Adding unbalance generally will not help nor work consistently even if the end play and pivot play have been "tweaked" out.

Unfortunately making the spring much stronger than the original causes the force and thus friction between the worm and the main gear to become too large and the drive binds. The tiny motor which drives the gear train is not nearly as strong as the motors used in many drives. This is an unfortunate limitation on any attempt to redesign and/or rebuild the drive, as I have found out. Replacing the motor with a stronger one would probably require redesign of the drive electronics as well at which point the entire system would have to be redone. Thus the only thing that can be reasonably done to improve the drive is to limit the maximum "float" action of the worm platform and the motor/gear reduction parts of the drive. This can be done within the rubric of "tweaking" the mechanism.

One might wonder about the design of the mechanism in the first place.  Why is the worm on a "floating" platform at all.  One reason would be to keep the worm, on its floating platform and via a spring, to be held in optimum contact with the main gear.  Another would be to allow for slight run out of the main gear. The main gear run out should easily be kept to under 0.05 mm on a gear with a 73 mm radius. In the case of the gear measured the run out was 0.1 mm.  This is not a particularly refined tolerance but it is not very bad either. Several other  worm/gear drives investigated had better tolerances and did not use the "floating" worm arrangement.  If main gear tolerances are typically 0.1 mm, there seems to be no reason for a "float" of 0.5 mm.  In fact, there is an adjustable stop on the floating platform that limits the disengagement of the worm to the 0.5 mm observed. It seems that this adjustment could be tightened up to limit the "float" to be not more than required for the main gear run out.  Reduction of the allowed worm platform motion was tried and does reduce the looseness of the drive linkage and the maximum slack allowed.  To do this, the motion limiting screw needs to be raised toward the bottom of the platform.  It was possible to tighten this tolerance until only 0.03 "float" remained on one drive and 0.05 on the other.  This caused significant improvement in the total slackness within the drive systems.  The retrograde motion was reduced but not eliminated.

Looseness in the drives was improved greatly. Only about 80 to 120 arc seconds of slack remained compared to 220 arc seconds without the adjustments described.  Now another strange motion of the declination pointing mechanism was observed. When the motion was reversed in either direction a small retrograde motion remained. This was finally traced to the mounting between the bearing platform plate and the gear train housing on which the motor is mounted.  Unbelievably, the entire drive
train/motor housing is attached to the worm bearing housing with four small bolts and a thick rubber ring or gasket (actually a small "O" ring.)  Thus the whole reduction gear train housing can move with respect to the worm bearing and when it does it allows the worm to rotate with it. The amount of motion on one drive was 0.5 degrees rotation of the worm. On the other it was 0.2 degrees. This corresponds to an angular motion of the telescope tube of 10 or 4 arc seconds. Before the motor drive train can move the worm any amount, the rubber gasket must go from clockwise to counter clockwise compression limits. (or vice versa for a change in the opposite direction.)  This working of the rubber gasket is undoubtedly complex and may cause jerky motion of the worm often observed during reversals.  First retrograde and then correct motion is sometimes observed.  It is not at all clear exactly why this strange phenomenon takes place.  It was however, observed to be repeatable over many reversal cycles.  It must be related to the use of a rubber coupling element in the drive chain.  It is a weird hysteresis
phenomenon which would not take place in a linear system.

It is very tricky to get at the rubber "O" ring.  The entire gear drive assembly has to be dismantled.  This operation is full of traps and should not be attempted unless you are ready to replace a broken motor/gear drive assembly in the case that you ruin it.  The drive is assembled from the inside out and at several points items are glued into place and press fitted. It is exceedingly difficult to take apart. The gear drive assembly was taken apart however and then tightly bolted to the worm drive platform and the entire drive reassembled. The second drive was similarly reworked after the first was improved greatly.

Additionally, In both drive trains, it was found that the gear at the end of the worm shaft was not tight. In one case 5 degrees and in the other 3 degrees of looseness was found.  This accounts for most of the remaining looseness and consequent hysteresis in the gear reduction system.  Both gears were removed and found to have play between the plastic gear and the steel worm drive shaft.  The gear, probably nylon or delrin, has a flat "keyway" on one side which simply had become distorted and no longer locked angular position of the gear to the "keyway" on the shaft.  This was fixed by filling the distortion with epoxy and locking the gear to the shaft with an added lock washer under the retaining bolt.  This fix holds the gear very tightly to the shaft.

Operation of the drive mechanisms now took on a considerably different nature.  The following motions were observed with no hysteresis correction entered into the computer. There was now no retrograde motion. Instead, there was no motion at all for about 3 seconds.  This corresponds to 45 arc seconds of drive demand with no telescope motion. When stop action is called for, the tube now stops immediately as it should and subsequently does not move at all.  This is both correct and necessary because it means there is no overshoot or drift.  It does however require another 3 seconds for motion to take place in the
opposite direction. This confirms the symmetry of the 3 seconds of  hysteresis in the drive.  This amount of delay is similar to that expected when the drive is operating to specifications stated in the operators manual.

One must conclude that when requesting reversal of declination motion, there is a total windup in the gears, worm and main gear of 45 arc seconds. This seems like a lot of windup in the gear train but it is only a very modest set of plastic (with some metal) gears.  The system as adjusted is now very tight mechanically, but still very smooth running.  Since this wind up is symmetrical and consistent in amount, it can now be compensated for by the declination lash compensation.  The compensation entered into the computer simply causes the drive motor to windup the required amount in the desired direction so that mechanical lash is absorbed and the forces applied are just enough to start motion of the declination axis.

Both drives, after many hours of remodeling and "tweaking" are now operating fairly well.  They seem to be smooth and reverse with consistency.  Apparently no amount of "tweaking" will make the coupling between the motor shaft upon which the encoder is mounted and the declination axis absolutely tight. This is to be expected with the very simple gear train used.  Only expensive spring loaded gears as used in precision servomechanisms would be free of mechanical lash. Thus it is fortunate that a very clever computer fix for this problem has been provided.  The mechanical hysteresis problem is probably extant in most drives of this type but is usually not amended.

The conclusion of this study and experiment is that the floating worm drive design while a bit unusual is probably necessary in a mass produced driv