Roller Drives

Drive systems for astronomical telescopes present unique challenges in mechanical design. The sidereal motion around the right ascension axis is very slow, about 1 revolution per day yet the precision of that motion over minutes to hours must be very precise. For purposes of discussion let us assume that we have a telescope with a CCD camera and we would like to expose this imager for 10 minutes. Further let us assume that a “good” image would be one that has nice round star images and we are located where the air is steady and we can expect that stars should be about 2 arc seconds in diameter. OK … what is round? As a criterion for roundness, we might say that the difference between the width and the height of the “round” stars be less that 10%.
Putting all of this together we are going to ask our telescope to move through 150 arc minutes as it follows the stars with a precision of less than 0.2 arc seconds, or better than .0022%. Let us now translate this precision into both a worm gear system and a roller system.
Worm gear system
As you recall the way the worm gear set operates is that the gear has a large number of teeth (359 or 360 typically).  Meshed to these teeth is a worm gear that has a single tooth spirally wrapped around the worm face. For one full rotation of the worm, the gear will rotate 1/359 (or 1/360) of a revolution and thus drive the right ascension axle this same amount. With this in mind, let us pick a worm gear that is 12” in diameter and a worm 1.5” in diameter, fairly common sizes for larger telescopes. A drawing of this is shown here.
A practical aspect of worm gear sets is that the worm cannot be pushed against the gear with so much force that the worm can't be turned. To overcome this problem some designs call for the worm to be spring loaded into the gear while others separate the worm and gear by some clearance and use pre loading of the RA axle to keep the gears meshed in one direction to eliminate play.
Back to the drawing, it is clear that the distance “D” between the worm axle and the center of the gear must remain constant. If this changes then as the worm turns there will be a slight “faster” then “slower” rotation of the RA axle as the worm gets closer and farther away from the gear. Now applying our “good image” criteria to the distance “D” we find that it cannot vary more than .0022% or for a gear of 12” we have 6 x .000022 = .000133”. What this means is that the worm must be very concentric to it's shaft (no “wobble”) and that shaft cannot move with respect to the RA axle. Thus the gear and worm must be manufactured to a very high degree of precision. Think about cutting 360 teeth in just the right place and cutting that one tooth on the worm precisely. Even the best worm gear set manufacturers will lap the gears together after they are made to make sure that they are as close to perfection as possible.
Modern commercial drive systems have recently taken a middle road on their worm precision. They build the best that they can within certain monetary constraints and then add some electronics to monitor the worm rotation “wobble” and apply corrections to the RA motor rotation rate to compensate for the “wobble”. This is generically called “Periodic Error Correction” or PEC. PEC requires training, either factory or user, to achieve high precision drive rates.

Roller Drives
The concept of roller drives is pretty easy. A smaller shaft is pressed against a larger disk and as the shaft is turned the disk rotates. The reduction is the ratio of the diameters of the shaft and disk. If the shaft/disk pair is pressed together very hard and the correct materials are chosen for them they won't slip as they rotate. If one cascades two or three of these shaft/disk sets a reasonable gear reduction can be had:
Now everything we said above about worm gear precision above applies to these shaft/disk sets. They must me made concentric, smooth and round. If any of these three things is not precise then the rotational rate of the RA axle will vary.
If the manufacturing precision of these shaft/disk sets must be the same as worm gear sets what's the big deal? The answer is that it is very much easier to fabricate a round disk and shaft to high precision than it is to make a precision worm gear set. Almost any modern machine shop can fabricate a shaft/disk set to the required precision where as there are very few shops that can fabricate a precise worm gear set.

The rest of the story
One difficulty with a roller drive is that the precise reduction from the input shaft to the output shaft is a function of the diameter ratios. I said that a precise shaft/disk set could be manufactured easily and I also said that the three criteria for a good set are concentricity, smoothness and roundness. Yet the reduction ratio depends on the diameter ratios. In order to make a precise roller drive ratio, then the diameters of the shaft and disk must be made precisely … a very hard fourth requirement. But, do we care about the precise reduction? The answer is no. Almost all telescope drive systems use crystal-controlled oscillators to govern the drive rate of the motor. To achieve a precise sidereal rotation rate with a roller drive system simply requires that this oscillator frequency be adjusted to achieve the required rate, something very easy to do.
There are practical aspects to roller drives. The first is that the shafts and disks must be mounted with good bearings so that the required high contact pressure between them can be maintained. Each shaft/disk must be well aligned to each other so that the contact between them is as good as possible. Large individual reductions can lead to rotation errors. Accelerations used during slewing motions of the telescope need to be kept low to prevent slippage.  

Do they work?
Yes they do. As an example of a telescope using roller drives look at the 2.5 meter Sloan Digital Sky Survey telescope at Apache Point, NM. This has achieved very good pointing and tracking using such a drive. Admittedly they use encoders to measure the telescope pointing but the tracking is dependant on the roller system. There have been a few smaller telescopes that use this system including some amateur designs.

Another Twist

Blending the two methods together can reduce the inherent problems with both worm gears and rollers drives. Specifically if one could provide an additional reduction after the worm/gear drive then the amount of "wobble" could be reduced by the amount of the reduction. Next, if a method could be found to increase the contact area of the rollers, then the slippage between the rollers as well as the high contact forces could be reduced. The solution is to use a thin metal band to transmit the motion of the worm gear rotation to the RA axle. The approach is as follows: attached to the worm gear and concentric with it is a round cylinder. Attached to the RA axle and concentric with it is another round cylinder. Now by placing the thin metal band around both cylinders as the worm gear turns it will rotate the RA axle. If the diameter of the worm gear cylinder is smaller than the diameter of the RA axle cylinder then there will be a reduction between the worm gear and the RA axle. With the band contacting a large surface area of each cylinder contact forces between the band and a cylinder are small while the band reduces the alignment constraints between the worm gear cylinder and the RA cylinder.

Such an approach is below:

Back in 1971-1972 Andy Tomer built a drive of this type in his "Phoenix III" telescope. His experience with this drive are not recorded, however, the "Phoenix III" telescope has been recently re-built and outfitted with a DriveScope drive system. Early reports are that the system works very well providing excellent star images over long exposures. At the time of this writing others are using this system with very favorable results.