|
| PrimaryMirrorActuatorBoards |
UserPreferences |
| Magellan Telescopes | FrontPage | RecentChanges | TitleIndex | WordIndex | SiteNavigation | HelpContents | moin.sf.net |
The primary mirror actuator failures on Magellan now account for 50% of telescope downtime. The primary failure point is a DC-to-DC converter that fails or becomes intermittent. The underlying cause of the failures is not known.
Alan Bagish, with with assistance from Ken Duffek and Dusty Clark at MMT, has made a prototype board and, with Patricio Jones, has it mostly debugged. This board puts many parts in sockets and allows for a variety of DC-to-DC converter options. It is not a direct copy of the original but maintains many of the concepts. (We are looking for a more reliable, drop-in replacement board, not a new implementation.)
January 6, 2005, 10 AM. This phone conference is to get acquainted and to find out what's been done at MMT.
Date: Thursday, January 6, 2005 Time: 10:00 am PST Duration: 1 hour Number to call: 1-800-473-8494 Chairperson: Alan Uomoto Pasadena participants - Please meet in the conference room. Thanks, Silvia
Questions and topics:
P. Jones:
This is the list of topics I would like to discuss with people in Arizona:
I may have more topics to talk about, the ones above are what I remember right now. I think that points 4 and 5 should be addressed only there at Arizona.
Pato Jones.
Here is more:
6. Replacing actuator boards. Is it safe to replace an actuator board with other not calibrated with that particular actuator? We assume that is a board which has been successfully calibrated with other actautor.
7. We need to know about the feature I discovered at the teststand, where the 45 degrees cylinder reverses when the force asked passes +/-470 for the new board or +/-490 for the original board.
8. We would like to know wich software was used for creating the original schematics of the cell's electronics. I mean, having them in pdf format is OK, but being able to open them with the software they were made with would be much better.
9. We would also like to understand the porpuse of the analog switch on the board. We know that allows us to enable or disable the integrator of the circuit (by having the INTLO signal high or low respectively), but don't know its application on the supporting system of the primary mirror. I have seen (on the telescopes, that INTLO is always gotten high when an actuator is being used or bump tested). May it be related to panic events?
Regards, Pato Jones.
Travel for Alan Bagish, Patricio Jones, Alan Uomoto:
Notes from our visit to U of Arizona (MMT people). By Patricio Jones. Edited by ....
This is a summary of what we were told or found out during our visit to University of Arizona. Visitors were Alan Uomoto, Alan Bagish and Patricio Jones. Our hosts were Dusty Clark, Tom Trebisky and Ken Duffek (now on LBT).
1. The telescope can be moved (elevation) with the primary mirror down. This is specially suggested when EL is under 45 degrees, because raising it under this angle makes the static supports to be ‘dragged’ by the bolts. Do we have Teflon sleeves around the bolts of our static supports?
2. Mz reaching the limits:
A. HPair pressure may be increased up to 34 PSI (for a start). It may be a reason for this issue. B. The rotation values, the de-compensation in the values of the lvdts of the hardpoint pairs, the Mz value, and other symptoms, are a clear evidence of the hardpoints ‘breaking away’. Remember that the hardpoints pairs are 0-1, 2-5 and 3-4, according to their location in the cell of the telescope. C. The hardpoints have Krytox inside, which is for high pressure and high temperature. Krytox tends to get stiff with low temperatures, so the hardpoints do get harder to move under these conditions. So the motor would not be able to cope with this and the hardpoints would break away. D. The elevation slew speed of the telescope may be decreased for avoiding this Mz issue. E. The hardpoints should not leak. There should be no air flow towards them. F. As far as Ken remembers, 34 PSI was the limit for HP air pressure at the Magellan telescopes. It was limited both electronically (using a zener diode) and with pop off valves. Ken says that Matt Johns wanted to increase this limit, but he doesn’t remember if Matt actually did. G. Ken said that the hardpoints do leak at -22 degrees Celcius (lab tests). H. As I said that Oscar and I had found some air leak (or flow) coming out of the blue structure of the hardpoints at Magellan, they pointed out that a piece was removed from the hardpoints and that these holes may be related to them. They suggested us to just tap and cover them with set screws. I. Tom informed that according to (I didn’t get his name) calculations, 34 PSI is the hardpoints produces 100 PSI stress in the glass, which can be ‘accepted’ (don’t remember the word) for 50 years. 50 PSI would produce 150 PSI stress in the glass, no information regarding time on this one. J. Dusty suggested a test using the spare hardpoint on a ‘test bench’. The idea is to make a graph of the lvdt position while increasing the load cell force (by moving the motor). If the graph turns out to be different from expected (look at the notes), the break away mechanism should be disassembled and checked (some springs may be defective and may need to be replaced).
As a summary of the Mz and hardpoints issue, they suggested us to eliminate all the air leaks in the hardpoints first (measuring the air pressure in place for each one), to increase HPair up to 34 PSI, and to decrease the elevation slew speed of the telescope. The air leaks may be related to internal o-rings or seals failing in the hardpoints.
3. Actuators:
A. In general, it is not recommended to replace an actuator board without calibrating the actuator + board pair. It is OK to replace one for saving a night or the rest of a night of observation, knowing that doing it will not damage the mirror (the board must have passed a calibration with other actuator), and that it will produce an offset in the forces applied, with respect to the commanded ones. The new actuator + board pair must be calibrated as soon as possible. B. The air transducers on the actuators need to be recalibrated more often than what we do (just when an actuator is removed for calibration). It is possible to do it on site, but with the mirror down, and commanding no more than 75 or 100 PSI (using the TEST button on the actuator window, xcell4) for adjusting span (zero can obviously be calibrated with zero command). This adjustment is not as good as the proper one (0~10 Vdc input for 3 ~ 110 or 115 PSI output at the test bench) but it may help keeping the actuator forces closer to the commanded ones. C. Dusty and Ken commented that they had a 48Vdc (unregulated) power supply failure at MMT once. It damaged the DC/DC converters of three boards, all of the ‘orange line’. They eventually found out that the power resistor used for discharging the power supply’s capacitor had opened. They suggested us to check our power supplies (4 per telescope), looking for something really weird (same power supplies are used at MMT without our problems with the converters). There is an idea of replacing two power supplies on each telescope with linear regulated supplies (24VDC or 48VDC), expecting to have failures only in the actuators fed by original ones. We will want to monitor the output of the UPS that feeds the power supplies for power quality and true sinusoidal output. Harmonics may damage the DC-DC converters. An oscilloscope needs to be hooked to the output of the 48VDC supplies and triggered at turn-on to look for any transient events. If LCO has a power quality analyzer (such as a Fluke 43B or 433) we ought to hook it up ASAP.
4. Test stand and new board’s tests.
A. Regarding the reversal of the force on the 45 degrees cylinder I saw at the test stand at LCO, we agreed that it must be the incoming signal (coming from the test stand’s computer), so we have to measure the input signal on that half of the board. They think that either there is a rollover (when passing some undisclosed value in some signal) or there is a failure on the feeding voltage of the ADC in the test stand (should be +/- 15Vdc, in the backplane). B. Regarding the failures on the INTLO signal we have had in our test stand, they pointed out that it is the only output being used in the 48 I/O module. So they suggested us trying to use one of the other 47, which are available. This made me realize that we don’t seem to have the schematics for the test stand, so I asked them if can give us a copy (I asked Alan Bagish to insist on this topic).
5. Cell monitoring windows (mainly from Tom Trebisky). Tom said that we may ask for having more friendly labels on these windows, and that some cells (or even columns) may be taken out of them. Tom also told us that the software doesn’t look for the ‘magic’ position all the time. In fact, once the system achieves that position, it forgets about it. This is the reason why we need to press the ‘magic’ or the ‘autoraise’ buttons every time the mirror’s support system loses the ‘magic’ position.
A. Control window (xcell1): The columns ‘cur’, ‘err’, ‘cmd’ and ‘total’ are related. The first one (cur) shows the current values on the forces and momentums in the cell, coming from ‘the matrix’, which calculate them from the values in the load cells of the hardpoints. The second column (err) shows the difference between the current and the target values (cur – target). It is normal that these values are near zero, because the target forces are zero most of the time (??). This is also the reason why the values in ‘cur’ are always near zero (the idea is to make that the error values are zero). The third column (cmd) shows the outer loop output values needed to take the error values to zero, so this is the reason why they are not near zero. The fourth column (total) shows the summation of the forces of all the actuators in the X, Y and Z direction (being Fy = mirror weight x sin(el), Fz = mirror weight x cos(el), and the summation in X related only to the four cross lateral actuators). The values of the actuator forces come from the force monitors. B. Hardpoint window (xcell0): We understand most of this window, but some things needed to be clarified. The first column at the left (‘lvdtent’) has cells for telling the system to move the hardpoints until the lvdts reach the values written there. The values in the third column (‘lvdt’) are in millimeters. The values in the fifth column (‘lc’) correspond to the load cell vaues and are supposed to be near zero all the time. The sixth column (‘limit’) shows a message in the proper cell when a limit switch (inductive proximity sensor) has been triggered. The override button (below the sixth column) allows the user to override the related limit switch in order to move the motor of the hardpoint and get it off the limit position. On the MMT, they have added a hardware limit switch along with each proximity sensor as a second safety measure, and they have installed override temporary switches to allow the movement described above (no software override for these). The ‘avg’ cell indicates the readings being averaged for having a screen reading. This is for making the readings smoother, since the lvdts and the load cells readings are jumpy. C. Position window (xcell3): The lvdt values showed in the third column (‘lvdt’) come out of ‘the matrix’, which calculates the X, Y and Z incr values and the Rx, Ry and Rz incr values form the hardpoints’ lvdts values. These values are also important to figure out when something weird is going on in the cell (they should be around zero all the time, while in ‘magic’). The first column has cell for entering values where the hardpoints are supposed to take the cell. These cells are not used anymore since the needed values are always the same, necessary for taking the mirror to the ‘magic’ position, and have been included (automated) in the software. D. Actuator window (xcell4): The only ‘new’ feature here is the first column. It allows the user to ask for an individual actuator to apply a force in Z or Y directions, or to apply force with its main or auxiliary cylinders. No more that 100 PSI should be applied using this feature and always with the mirror down. It was intended for diagnose purposes only. E. Air window (xcell6): The important thing here is to realize that the ‘elevation’ valued is absolutely necessary for moving the telescope in elevation, because ‘the matrix’ needs this value to calculate the forces (sin(el), cos(el)) which need to be applied in order to support the mirror.
