Calypso Telescope Periodic Servo Error Correction Procedure

Bruce Truax
btruax@dld-llc.com
engineer@calypso.org
860-276-0450

Rev A.

August 30, 2002

Introduction

During commissioning of the Calypso Telescope we found that there were sub arc second periodic errors in both the Elevation and Azimuth axis Farrand encoders. The Farrand electronics balancing procedure was supposed to remove these errors through a careful balancing of the since and cosine channels of the detector. While performing the balancing procedure we discovered that, while the balancing procedure did minimize the periodic errors at 1024/revolution, it did not eliminate the error. In fact we observed errors not only at 1024/rev, but also at 1056/revolution and 2048/rev. Farrand did not have a good explanation for these errors, nor could they recommend an electronic or mechanical alignment method that would allow them to be eliminated. Repeated testing of the errors showed that they were relatively stable in both amplitude and phase when referenced to the Farrand ripple clock (which is received once sector for each of the 512 sectors of the encoder).

Having determined that the errors were stable we reasoned that it would be possible to correct them by adding and equal and opposite error into the servo control loop. The Ziatech servo software was modified to allow the addition of up to three sinusoidal correction coefficients. The frequency, phase and amplitude can be specified independently for each of the three coefficients in the servo axis configuration file. All that is needed is a method to measure the error and determine the values for the correction coefficients.

This procedure provides instructions for measuring the periodic error for each axis, computing the correction coefficients and entering the coefficients into the servo error configuration file. The elevation axis is the most critical because errors in the elevation axis map directly to errors in image position while errors in the azimuth axis are reduced by the sine of the Zenith angle. Fortunately the elevation axis can be measured with the building closed so it can be done even in poor weather conditions. The Azimuth axis test needs to be run with the building open on a nice day with low wind. Less than 10 mph would be ideal but the test will probably produce acceptable results in winds up to 15 mph.

While the periodic Farrand error is reasonably stable, the error does drift with time and/or temperature. It is recommended that this calibration procedure be repeated every six months for optimum telescope performance.

Equipment Required

The following equipment is required to perform this procedure.

• Two people, one to perform the test and a second one to act as a safety during the elevation axis test. While the elevation axis test can be performed with the building closed, the elevation axis cannot run through the entire range of travel without hitting the ridge of the enclosure. It is therefore important that a person be present on the observing level during the test. This person should be ready to press an emergency stop button if the telescope is in danger of striking the ridge of the enclosure.

• Phillips PM3382 Oscilloscope and two probes

• 9-Pin to 9-Pin null modem serial cable

• Instrument PC

• Servo Test Terminal Strip

• Computer connected to Calypso Network running MathCad and copies of the Aximuth and Elevation Encoder Tuning worksheets.

Procedure

Note: The bulk of this procedure will work for both the Elevation and Azimuth Axis. Where there are differences, the Azimuth axis instructions are in red.

A minimum of three measurements will be taken for each axis. The first measurement will record the axis performance with the current correction parameters. This baseline will be used to determine if the new correction coefficients provide improved performance. The second set of data will be recorded with the periodic error correction disabled. This data will be used to compute new coefficients. The new coefficients will then be entered into the servo axis configuration file (AZIMUTH.CFG or ELEV.CFG), and enabled. The third set of data, taken with the new coefficients will be analyzed to determine the corrected performance. Each axis should be less than 0.07 arc seconds peak for all three frequency peaks.

1. Using a rag and some WD40 wipe down the elevation axis drive sector. This will require releasing the brake and moving the telescope. When completed, park the telescope. When completed, park the telescope and re-engage the brake.

Remove the diamond plate and drive ring covers. Using a rag and WD-40, wipe down the drive ring, idler wheels, motor capstan and particularly the Heidenhain capstan. This will require releasing the Azimuth brake and rotating the axis to various positions. When completed, park the telescope and re-engage the brake.

2. Connect 10 conductor ribbon cable from the Servo Test Terminal Strip to JX on the VL1297 board in slot XX of the Ziatech chassis. (see Figure 1).

Figure 1 – Servo Terminal Test Strip

3. Connect scope probes as follows:

Probe	Connection
Channel 1 Signal	2M10FC pin2 (Figure 2)
Channel 1 Ground	2M10FC pin 1 (Figure 2)
Channel 4 Signal	Elevation Axis Ripple Clock Azimuth Axis Ripple Clock (Figure 3)
Channel 4 Ground	Elevation Axis Ripple Clock Azimuth Axis Ripple Clock Ground (Figure 3)

Figure 2 – Servo Test Connection Points

Figure 3 – Channel 4 Trigger Probe Connections

4. Press the escape key on the Servo Controller keyboard. If all of the axes are properly parked the MASTER control program should exit and you should get a DOS prompt.

5. Type EDIT ELEV.CFG (AZIMUTH.CFG) and use the cursor to scroll down to parameter number 9. Set the value of this parameter to 7. Press ALT-F, X and answer YES to save the file.

6. Open the other axis CFG file in the editor and set parameter 9 is set to 0.

7. Reboot the servo controller by cycling servo power.

8. Setup the Oscilloscope as follows:

a. Press DSO button until the display reports that the scope is in Digital Mode.

b. Connect the null modem serial cable from the RS-232 port on the rear of the scope to COM1 (Incorrectly labeled COM2) on the instrument PC. (See Figure 4)

c. Channel 1 amplitude – 2V/Div

d. Channel 4 amplitude – 5V/div

e. Channel 4 trigger – ON

f. Channel 4 trigger level – +2V

g. Timebase Mode->Acquisition Length->1 ch @ 32Kpts -> return

h. Text Off

i. Timebase Magnification – 1/32

j. Timebase – 32 seconds/div

k. Channel 4 – OFF

l. Timebase delay = 0

m. Set the vertical position of the Channel 1 trace to the center of the screen.

Figure 4 – Serial Port connection on Instrument PC

9. On the instrument PC open the TELNET application on the desktop.

10. Telnet to charon and log in as TCC with the password scope2point.

11. When you get the $ prompt in the telnet window type TELRUN and wait for the TCC to complete the startup procedure.

12. Enter the command TALK TCC_AZ. (Note that this command is the same for Elevation and Azimuth axes.)

13. At this point the second person should go to the observing level with a 2-Way radio. We are going to start moving the elevation axis and the person on the observing level should be prepared to press an emergency stop button if the telescope gets too close to hitting the structure.

Open the building and make sure that the telescope is free to swing a full 360 degrees in Azimuth with the elevation axis parked.

14. Press F8 on the servo controller keyboard to view the servo status screen. Moving the telescope to clean the drive ring very likely caused an axis overspeed. Press the F4 key until all errors have been cleared. This may take one or two presses.

15. Return to the axis status screen by pressing F8. Press F7 as necessary to view the status of the axis under test.

16. Startup the axis by entering the following command in the TELNET window: ALT SEEK.REF (AZ SEEK.REF)

17. Watch the axis status screen and note when the axis has completed initialization and has stopped moving.

18. Enter the following command in the TELNET window: ALT MOVE 10 (AZ MOVE 0)

19. When the axis has reached its position enter the following command in the TELNET window: ALT MOVE 7 0.17 (AZ MOVE -2 0.5625)

20. Put your finger on the SINGLE button the scope and watch the Farrand Sector value on the Ziatech monitor screen. It will count up and then count down. After it has reversed direction and begun counting down, press the SINGLE button as soon as it reaches sector 523.

Put your finger on the SINGLE button the scope and watch the Farrand Sector value. It will count down and then count up. After it has reversed direction and begun counting up, press the SINGLE button as soon as it reaches sector 359.

21. As soon as the Sector counter decrements (increments) the scope should trigger and begin recording data. The data acquisition will take 320 seconds during which time the axis will move from 7 degrees to 61.5 degrees (0 degrees to 180 degrees).

22. Note that the scope screen has the word WAIT at the center bottom. When the data acquisition is complete, the word WAIT will disappear. As soon as the word WAIT disappears, press the F6 (Shutdown) key on the Servo Control keyboard. This will reverse and park the axis.

23. Examine the data. If the scope trace goes off scale or has any large discontinuities then the drive disk or Heidenhain capstan is not clean. Clean the axis drive components once more and try again from Step 14.

24. Run the SCOPEXFR program on the Instrument PC. When asked which serial port is being used, enter 1. Type in the file name (usually ELERRn (AZERRn) where n is the data set number) with no extension. Wait for the data transfer to complete.

Figure 5 – Instrument PC with all three windows

25. Open the log file in the C:\DATA directory and make an entry for this data set recording all of the acquisition parameters and the purpose of the test (use one of the previous entries as a template). This is the Baseline data set.

26. Press ESCAPE on the Servo controller console to exit MASTER.

27. Edit the <axis>.CFG file and change parameter 65 to 0. The first data set recorded the axis performance with the current set of correction coefficients. This next data set will record the axis performance with no correction. This is the data set, which will be used for computing the new coefficients.

28. Restart the servo process by cycling SERVO power.

29. Repeat steps 16-26 to take and store a second set of data.

30. If it is not already running, start up the Mac in the library.

31. Once the computer is running, open the Farrand Tuning folder on the desktop. Inside this folder there should be two MathCad files El Farrand Tuning.mcd and Az Farrand Tuning.mcd. There may also be some old Data files. If any old data files exist, determine their creation dates and make a new Subfolder named Servo Data dd-mm-yy and drop the data files in the new folder.

32. Under the Apple Menu, open the Chooser and select Dave File Sharing.

33. Select the Instrument PC and mount the C disk. The password is calypso.

34. Navigate to C:\DATA and copy the data files (the ones with the .txt extension) to the Farrand Tuning folder.

35. Open the El Farrand Tuning.mcd (Az Farrand Tuning.mcd) file.

a. On the first page of the file you should see the following equation:

Data :=READPRN(AZERRafter _txt)

36. Edit the name of the file to match the name of the first data file.

37. Print the first three pages of the worksheet on the laser printer. This is the baseline, which we will use to compare the final results. The frequency plot on the third page will be used for the comparison.

38. Change the name of the data file to the second data file taken. This is the file, which will be used to compute the new coefficients.

39. Scroll to the bottom of the sheet and wait for the computation to complete. This could take a few minutes.

40. Scroll up about three pages from the bottom to just above the time series graph. You will see three phase offset values on the right of the page in the equation:

Change the three values of phase to either 0 or 180 to minimize the residual noise on the data. This can usually be done by observing the time series plot immediately below this equation but it may be necessary to look at the frequency plot at the bottom of the screen. This step is required because MathCad does not have an arc tangent function valid for all 360 degrees of phase.

41. Once the correct phase offsets have been chosen, scroll to the bottom of the screen and record the Frequency, amplitude and phase values reported in the following equations:

42. Open the ElCorrection Coef History (AZCorrection Coef History) file and record the existing coefficients (available from the <axis>.CFG file) if necessary and the new coefficients. Use previous entries as a template. If the previous person did their job, the existing coefficients should already be in the file.

43. Print the entire MathCad worksheet.

44. Print the file and take it to the electronics room.

45. Edit the <axis>.CFG file. Scroll to the bottom of the file and enter the new coefficients. Also edit parameter 65 and set it back to 1 to turn on the new correction coefficients.

46. Restart the servo controller by cycling servo power.

47. Repeat steps 16-25 and take what should be the final set of data.

48. After saving the scope data, return to the Mac and copy over the latest data set.

49. Analyze the latest data set and print the first three pages. Compare the Frequecncy analysis at the end of the second data set to the analysis on page 3 of the final data set. These two should be similar and the final data set should be better than the first data set. If this is not the case, edit the <axis>.CFG file and forward the data to a telescope engineer.

50. Create an archive folder for the data on the Mac and drop the data files into the folder. Repeat this process on the Instrument PC.