This is a personal interpretation of the design of a computer control system with comments that are based on my own experience. I have taken the liberty of mentioning, at several points, my own example of how a revised design might address perceived problems.
This is not an attempt to decipher the details of the design of the main computer board in the LX200. It is written to help owners understand better just what is in the computer system and how it interacts with the keypad, the motor drives and other external items. The main board (MB) is very complex. It has a powerful computer chip, memory, logic chips to interface the computer to the outside digital world, including the keypad and encoders and analog outputs to drive the motors on the two axes of the telescope. I do this because it is sometimes more comfortable to own and work with a piece of equipment that has been described in some detail. All in all, the computer works very well. Some of the interface electronics has been reported to give problems from time to time. I will offer an analysis of the operation of the MB with particular attention to the interface electronics. Schemetic drawings are available by clicking here.
Go to: Main Board Components: Basic
Computer Operation: Clock and Timer Circuits:
Problems Reported: Motor Drive
The computer can calculate co-ordinates for both Polar and Alt/Azm mounting, correct for atmospheric distortion, find thousands of objects, respond to external commands, store PEC data and do much more. The microcomputer, at the core of the system, is a 68301 Chip. This is a Toshiba chip which is a clone of the well known Motorola 68000 series. (but not one in the Motorola line) There are two Sharp 64K memory chips and a large AT&T gate array. The latter is a collection of logic gates custom programmed for the particular application. There are a few minor chips for input and output of serial signals, two large EPROMs which carry the main programs and data, Ver 3.34 in this case.
There is also a crystal oscillator and two Dallas DS1202 date chips.
These output Day, Month and Year information. The system is probably booted
from an Atmel EEPROM, 24CO4A but this is a detail that while not too important
at the moment, is discussed in more detail below. There are also
several operational amplifiers and four transistors which are associated with
the analog outputs to the motors. One very interesting chip is a
motor driver chip, UDN2993B, made by Allegro. More about this chip below.
There are two small power operational amplifiers, L2724, which drive the motors.
Many of these devices and their functions will be discussed below. The
board is a very high quality, multi layer, glass board. It is quite complex
and appears to be very well laid out. I noticed only one added (soldered
in) wire in the analog circuitry. A concern about the design is that it
is dated and several of the devices are now hard to get or discontinued.
The greatest complexity in any digital design is in the firmware that boots
and runs the system. This is unfortunately not accessible from a casual
inspection of the board.
The basic purpose of the computer system is to take various commands from the keypad and deliver signals to the motors that move the telescope optical tube to the desired position. Additionally, the computer translates time and in some cases guiding signals into signals which correct the motion of the tube. All this is done with a setup that requires only a few inputs to the computer to tell it the starting position of the telescope optical tube. Position calculations are done on a differential basis. That is, the next position is dependent upon the last position and the distance, length and direction, to the new position. The more accurate the last position and the smaller the move, the more accurate the final position will be. The computer has tables of positions of 65K stars and can also calculate the positions of the important movable objects like the planets, sun and moon. Additionally the computer takes care of housekeeping duties; like the display of the coordinates, local time and GMT and a few more minor but important calculations such as the worm correction data (PEC), the declination lash setting and remembering other settings within the mode group of commands such as location.
There are actually hundreds of things for the computer to do to make the system
friendly to the user. These, I feel, have been carried out with great
success. This GOTO telescope is a real joy to use in most scenarios.
On the other hand some problems have been reported and the system itself is
not well described or explained in the instruction manuals. Most
of the chips used are standard, reliable and well understood. As complex
as the system is, it does not compare in complexity to a PC and operating system.
In my opinion, the computer system, its interface with the keypad and with the
several accessories such as the RS232 port and the CCD port work very well.
It has a very good array of commands for external computer communications which
also work well.
There are two Dallas Semiconductor DS 1202 clock/calendar chips on the main board. They are controlled by a 32.768 KHz crystal which controls the accuracy of the timers. One of the chips is setup to deliver local time information and the other set up to give sidereal time for and is reset when star alignments are made The tow clocks are not connected in any way. The crystal oscillator circuit provides clock information for the clocks. Since the accuracy of all timing is determined by a crystal, it should be both accurate and stable. If there is a problem with the timing, it will have a serious affect on the RA drive rate as well as other timed functions. The crystal frequency can be adjusted with an external capacitor, but accurate time/frequency measuring equipment is required to make these adjustments. A more tedious means is to simply observe the time generated by the telescope and see if it is running fast or slow and then tweaking the crystal with a small parallel capacitor. (if it needs to be slowed).
The most practical action that the user can take is to simply synchronize the
clock with and accurate clock, like WWV, at the start of a session.
The timer will not drift much over a period of a few hours.
Of particular interest to the user will not be those parts and systems that
operate well, but those that cause problems. The problems that have been
reported have been primarily with the electronic and mechanical interface to
the telescope structure itself. That is, the motor drivers, the motors
and the encoder system. First a few words about the basic decisions required
to move a telescope in a suitably precision way. The telescope is moved
by an electromechanical servo system. The theory and design of such systems
is well understood both theoretically and practically. In the case of
most modern telescopes motion is accomplished by stepper motors or by DC servo
motors. A basic design decision that has to be made is which sort of drive system
to use. Stepper motors have the advantage that they are easily incorporated
in computer driven systems. They have several disadvantages as well.
It is hard to get very large ratios of the slowest to the fastest drive speed.
In the case of a telescope, one would like as much as 1000 to 1 ratio.
(which is what the LX200 does) Perhaps it would be wise to settle for
several hundred to one to simplify the design. But even several hundred
to one is tricky to get with stepper motors. Another disadvantage of stepper
motors is simply the fact that they step. This basically means that they
constantly shake the mechanical structure. They have to be made to step
fast enough under the slowest drive conditions so as to not shake the telescope.
This means about 3 to 5 steps per arc second of optical tube motion. The
LX200 does 3 steps per arc second. (It is not a stepper motor but a sort
of quasi stepper system., see discussion following.)
A DC motor drive has some advantages. It has very high torque for its size, is inexpensive, has maximum torque at stall, has smooth motion and can be operated over a very large speed range. The disadvantage is that there is no easy way to tell how fast it is rotating and how far it has gone. It requires an external encoder to provide this information to the computer driving it. One way is to provide a tachometer feedback. This basically analog system has been used but has some of the usual disadvantages of analog systems such as accuracy and drift. There is a very cleaver way to get the advantages of the DC motor and those of the stepper motor. The system is well known and well understood. The LX200 uses it. The DC motor shaft is provided with an encoder that generates a large number of pulses as well as direction information as it turns. Then the direction and distance the motor has moved the telescope are kept track of by the computer. The LX200 has such a bi-phase encoder built into both the declination and the RA drives. It becomes a question of how this system, which is basically excellent, is actually brought off in practice. Because of the speeds required, there will be a gear reduction box between the motor shafts and the worms that drive the main gears. Since the gear box in the LX200 design is not within the computer/encoder feedback loop, it must be of very high quality. Unfortunately it is not of high quality and the motor is too small for the application. (in my opinion) But remember, the entire motor drive for the LX200 sells for $150.00. A very strong gear/motor drive system, such as I am building for my new mount will cost $200 for the motor and encoder and $150 for the gear deduction unit. Both drives require the main gear and worm mechanism in addition. It is hard to make a breakdown for the LX200 drive but the motor itself is very modest and thus might cost only a dollar or two. These considerations are not so much a criticism of one design over the other as they are of the economics of making and selling lots of something over a no holds barred design and making of just one model. I believe the LX200 design would fare better with a motor that was several times stronger and with a higher quality gear train.
When one part of the design is changed, particularly made stronger and larger. There are consequences for other parts of the system. In this design it is the power operational amplifiers that drive the motors that would be under stress. These amplifiers, L2724 made by SGS Thomson, are dual operational amplifiers made to run on a single sided power supply. There is one chip for each axis. The L2724 is in a 9 pin SIP package with a heat sink tab. It is rated at 28 volt max power supply and an output current of 1 amp max. The power dissipation is rated at 10 watts. (50C case temp.) The chip is designed to drive small DC motors. The circuit shows an internal snubber diode. (positive side only) With DC motors, I prefer to see external snubbers used. The chip shows thermal protection on the chip. On the face of it, this looks like an adequate chip to drive the LX200 motor. (But it is not over designed by any means I would say)
Some failures have been reported. Only an extensive evaluation of the circuit under the most extreme conditions expected would tell the full story. My feeling is that the chip was designed for use with VCR and CD capstan and take up motors. These do not operate under the stressful conditions experienced by the telescope drives, especially stalled or near stalled operation. I have looked over the detailed specifications on the motor driver chip to determine the suitability of using this chip in combination with the power operational amplifiers that are used. While the design is a bit unusual, probably do to the limited selection of chips available at the time of the design, there is no reason to believe that the design is flawed. Never-the-less, I am skeptical about the design and would not do it the same way. The electrical signal at the motor in the LX200 is incredibly noisy and has switching artifacts of large size on it. It looks like some additional filtering should have been used. One of the most mysterious parts of the MB design involves the conversion of the digital information to motor driver commands. I have to assume for the moment that the computer calls for a set number of pulses for the axes to move to get to a new location. The motors report back to the computer how many tics the encoder has made and when the correct number have been made the computer stops the motor. The mechanical feel of the motors is that they are controlled to one tic accuracy. (This does not of course move the telescope 0.3 arc seconds because of the lash in the gearing but it has the potential to do so.) The motors are in any case very tightly controlled.
Exactly how the computer sends the logical command to the motor is still not
clear to me. The chip that is the center of this operation is an
Allegro USD2993. This is a dual H-bridge motor driver chip. It is,
by the way, discontinued though Allegro reports that they have a substitute
for it. (whew!) The data sheet for this chip indicates that it is logic
controlled and puts out some sort of PWM signal. The signal is amplified
by the power operational amplifiers and sent to the motor. The circuit
is complex enough that I have not been able to decipher it completely,
The signals in various places are pulse like and sort of PWM looking.
The final output however looks voltage limited. This element
of the design will have to have more analysis done on it.
Summarizing, it looks like the design of the output stages driving the motors is adequate under normal drive conditions but may be marginal under extreme conditions such as stall or near stall operation. The larger, stronger motor drives that I have designed for my own new mount will be driven by a significantly higher power DC amplifier using discrete components with substantial heat sinking. This more powerful amplifier design will not be costly, perhaps only several tens of dollars. It is important to remember that a more powerful design for the electronics, motor and gear train will yield a design that can deliver large forces to the worm and main gears in the drive. Thus, mechanical slip clutches need to be installed to prevent damage to the remainder of the mechanical system under conditions of jamming. You donÉt want to loose a camera, diagonal or finger, after all. (particularly not the observers)
Now we can see that the basic operation of the system that the computer takes into account all of the command inputs, looks in the tables, determines the current position and sends a signal that moves the telescope to the new position. This signal is the number of degrees in Dec and RA or in Alt and Azm that the telescope must move. The motor responds by rotating and sending back to the computer pulses from the encoder. When the right number of pulses has been received by the computer, it stops the motor. For the LX200 the number of pulses is about 3 per arc second of motion. The computer has simply rotated the motor shaft the correct amount. It does not directly measure where the tube is pointing because there is no encoder on either shaft. When a CCD guider is added, the situation is quite different. In this case, the CCD guider looks at a star and tells the computer how it needs to move the tube to re-center the star on the guider chip. In this case, the tube is in the feedback loop so that control can be made as accurate as the CCD chip can manage.
One of the problems in the system is that of lash in the mechanical parts of the system. Mechanical lash is a dead zone in the feedback loop. Control about the dead zone can be unstable. This causes jumps and jiggles in the operation. A tight system, one with very little mechanical lash, will behave more smoothly. The idea of my new design is to tighten the system mechanically as much as possible. That includes reducing the mechanical lash to a very small amount. Because the friction and other forces increase when this is done, stronger motors to drive the system and higher powered amplifiers are required. The result will be a smoother running mechanical structure that will move firmly and precisely. (I hope)
Overall summary: I believe the computer and basic design concepts of the LX200 are very good. The shortcomings of the LX200 are, I think, entirely in the size and strength of the drives and their gearing. This is manifest by the apparent result that the larger LX200s, which used the same drives as the smaller ones, have more problems. A coupling of the computer system of the LX 200 to a drive and mount of greater mechanical integrity would be close to ideal, I believe. I would hope that the 16Ë LX200 with its larger bearings, gears and drives would behave very well. I hope and trust, that my new design will do well. : - )
Return to Beginning
Return to LX200 Electrical Index
Go to Home Index for Doc G's Info Site