DV02 Metrology Issues
Last Update: JeffMangum - 16 September 2009
Metrology Trouble List
List of "issues" associated with current use and performance of the Vertex DV02 metrology system:
- Linear Sensor Range Concerns: There may be a problem with the range (+-1mm) of the linear sensors. It may be too small to accomodate measured temperature flexure of the fork arms. Or, there may be other problems with the linear sensors on DV02. See below.
- Need New Metrology Monitor Points: Need to add metrology system offsets to SET/GET_METR_COEFF_N monitor points. See below.
Linear Sensor Range Concerns
On 2009-09-13 NickEmerson
I have some concerns regarding the new solartron linear sensors. After looking at the data from DV02 I see that the yoke arms expand/contract by about 55um per degree C(with very good correlation between length and temperature). Looking at this in a very basic way, if we consider the operating temperature range to be ±20C, then we need a total range of the sensor of 0.055*40 = 2.2mm. This is for primary operating conditions. If we then move to secondary operating temperatures we add another 10 degrees which brings the required range to 2.75mm. The actual range of the sensors is 2mm (+/-1mm).
Now, one could argue that with the insulation around the antenna and heat source from the pedestal(which is warmed by the UPS) perhaps the yoke temperatures do not reach the extreme -20C ambient conditions. So, perhaps this might be acceptable if we align it well enough with a positive-temperature bias.
However, once we get into the secondary operating conditions (up to +30C) I think it quite likely one or more of the sensors will saturate. What happens when the sensors saturate is that they suddenly give an output value of about 88.8mm. So, as soon as it saturates it will instantaneously introduce an elevation pointing error equal to atan(88/1440) = 1.8 DEGREES and something similarly enormous in azimuth. Although the spec allows for "degraded" performance in secondary operating conditions, I think a 1.8 degree error is more than just degraded.
Another problem is that there are no monitor points to tell us if the sensor saturates, so we will just see a huge pointing error and have no idea why. I guess we can continuously monitor the linear sensors to look for the saturated value but that seems not a very clean way to do it. This saturation happened during one pointing run on DV02 (July 22nd Allsky1) although I think it was during a slew so no one noticed.
One solution might be to clip the saturated value to ±1mm rather than 88mm, but then we might not realize that we are saturated. But at least it might still let us have "reasonably degraded" performance. Anyway we should probably start a discussion with Vertex about this (I've had brief conversations with Lutz and Davide but maybe it needs to be expanded).
noted, with regards the question of how the LS should respond when it reaches it limits:
It is better for the sensor to produce a totally bogus value than to clip at its limits. I prefer the current performance (which allowed you to spot the fact that it saturated on July 22nd).
...and commented that Vertex possibly has a design flaw in their LS system:
Vertex clearly needs to be notified about this design error ASAP. Have they in fact received approval for the replacement (Solartron) sensors which replaced what ever their original design was? Is it the sensor described in DRD-45 (dated December 5, 2008)? The model described in DRD-45 is AX5/1/U, and indeed its range is +-1 mm. There is no way that we could have known that the yoke would thermally flex beyond this range, but Vertex should have.
I think I made the following overly-simple calculation : Steel CTE ~ 13ppm/C, Yoke arm height = 3m, temperature range = 50C. This would require a range of 13*3*50 = 1.95mm, which would have been enough as long as it was correctly aligned. But, obviously this was wrong.
Range May Not Be The Problem...
On 2009-09-15 NickEmerson
Following discussions with Lutz and some reanalysis of DV03 data, the situation may have changed slightly. Now it seems there may be a problem with the linear sensors on DV02, which is causing the coefficient of thermal expansion to appear much larger than it actually is. The linear sensors that were delivered with DV02 had to be modified to fit the antenna, and this modification might have caused a scaling problem. I'm not sure if this is a temperature-dependent scaling issue or a displacement-dependent scaling issue.
The thermal expansion of DV03 according to the new sensors (unmodified) is only about 35um/C, more consistent with a which would mean that we do have the required range of the sensors.
Obviously we need to figure out why there is such a discrepancy. I'll have a look at DV01 (which still has the BEI sensors) to see how that compares.
Tiltmeter Installation Concerns
On 2009-09-13 NickEmerson
noted, with regards the installation calibration of the tiltmeters:
Part of the problem with the AN0 and AW0 terms is that the antennas are initially aligned on different pads at the SEF to within the required tolerance, but then there is a large variation of tilt in the different pads down here so they are all being aligned to slightly different references. The pads were delivered by NRAO, so this I see this as really an ALMA problem. Even if the pads are within tolerance there will be a variation - If one antenna is aligned perfectly on a pad which is tilted 10 arcseconds in one direction, another is aligned perfectly on a pad tilted 10 arcseconds in the opposite direction, then the two will differ by 20 arcseconds. But I suspect the pads may not be perfectly in tolerance, and then Vertex only aligns them to within 10 arcseconds of that reference, so I can easily believe we end up with 30 arcsecond differences between antennas.
Need for Automated Metrology Calibration Capability
On 2009-09-14 NickEmerson
For the software calibration of linear sensors, I think the best would be to make the calibration constants available to the ABM(through the SET/GET_METR_COEFF_N points), and run the calibration through an ABM script. This will make it easier to keep track of the calibration values to see they are within acceptable limits. Obviously this would rely on the mechanical installation already being correct from the beginning. The monitor points that need to be added to the ACU/ABM interface are:
Amplitude of sin(2Az) term
Phase of sin(2Az) term
Amplitude of sin(3Az) term
Phase of sin(3Az) term