1. James Bellinger, December 2006 1 CMS Week Muon Alignment James N. Bellinger University of Wisconsin at Madison 5-December-2006 DCOPS Data from MTCC2

2. James Bellinger, December 2006 2 CMS Week Muon Alignment General notes With a better understanding of disk deformation and more time available we were able to align the lasers better for MTCC2. Although disk deformation results in a laser position change larger than the DCOPS aperture, COCOA can use lasers from opposite sides to reconstruct positions.

3. James Bellinger, December 2006 3 CMS Week Muon Alignment Review of the structure of single SLM A typical SLM has 10 DCOPS along its line, with a crosshair laser pointing into the line from each end. This line is not perfectly radial. Each DCOPS has 4 CCDs arranged in a square. CCD numbers 1 and 3 measure something like “Rphi” and 2 and 4 measure “CMS Z.”

4. James Bellinger, December 2006 4 CMS Week Muon Alignment What happens when the field is on The center of the disk is pulled inward and the rim of the disk pushed outward. This tilts the lasers backwards, so that the beam is redirected about 2 mrad inward towards the solenoid. For the DCOPS near the laser, the pull of the disk inward approximately compensates for the laser tilt, and we don’t see large deflections in Z. We see a large effect in DCOPS far from the laser.

5. James Bellinger, December 2006 5 CMS Week Muon Alignment When the field is on Because the laser lines don’t go through the center of the disk, the laser tilt causes deflections in the “Rphi” as well as the “Z” measurements, so both are required to reconstruct the true rphi and z. Because the DCOPS are now in new locations with respect to each other and the laser, the CCD for one can shadow a downstream CCD and reduce the signal it sees.

6. James Bellinger, December 2006 6 CMS Week Muon Alignment Laser CCD3 Rphi CCD4 Z This reflection is unique: rarely a problem Background of glare from laser 12 345 678910 The laser position does not change much from position 1 to 5, but once past the midpoint it quickly goes offscale in Z. The “Rphi” variation is well contained. Field Off data: laser positions in 2 CCDs in DCOPS along an SLM 9 5

7. James Bellinger, December 2006 7 CMS Week Muon Alignment Data Quality observations The double peak is a unique local reflection. The DCOPS next to the laser sees a substantial background due to glare. Some CCDs show a reasonable profile but have low signal strength (CCD 3 set, #9). Some show no signal (CCD 3 set, #5). Most profiles are perfectly useable.

8. James Bellinger, December 2006 8 CMS Week Muon Alignment CCD3 Rphi CCD4 Z Field On Saturation Position offscale in other direction 6 Same SLM as before The changes in “Z” are easy to see, and there are changes in “Rphi” also.

9. James Bellinger, December 2006 9 CMS Week Muon Alignment Relative position along the SLM in mm Beam off – Beam on “Rphi” position in mm Deviations in Rphi ME+2/SLM1 The trend of the deviations is consistent with the tilting of the laser. CCD has bad signal. Fit should have failed.

10. James Bellinger, December 2006 10 CMS Week Muon Alignment Laser CCD3 RPhi CCD4 Z 10 98 76 5 4 3 2 1 This is the same SLM as before, but with the laser at the opposite end. The profiles are generally clean, and COCOA will be able to link together the measurements. Field Off

11. James Bellinger, December 2006 11 CMS Week Muon Alignment CCD3 RPhi CCD4 Z Laser Field On Same SLM as before. There appears to be some shadowing in CCD3 data, but the peaks are still detectable. 1 2 3 45 6 7 89 10

12. James Bellinger, December 2006 12 CMS Week Muon Alignment As you can see here the fit result is decent even when the profile has some saturation, as in CCD3 (lower left plot). I restrict the fit to a range around the peak. This is the DCOPS I numbered ’10’ in the previous sets of plots.

13. James Bellinger, December 2006 13 CMS Week Muon Alignment This shows the Z-position of the laser peak as a function of field for 4 different Layer 2 chambers at ME+1. The overall deflection is O(.5mm): small because the disk bending and laser deflection are in the same direction.

14. James Bellinger, December 2006 14 CMS Week Muon Alignment COCOA reconstruction of CSC center positions, from Gyongyi Baksay at FIT CMS Z position of the CSC centerIn mm, arbitrary origin Position along the SLM in mm

15. James Bellinger, December 2006 15 CMS Week Muon Alignment COCOA reconstruction of CSC center positions, from Gyongyi Baksay at FIT The error bars are smaller than the symbols. Position along SLM in mm CMS RPhi position of CSC CenterIn mm, arbitrary origin

16. James Bellinger, December 2006 16 CMS Week Muon Alignment Running Experience At irregular intervals a random terminal server port would lose connection with the DCOPS, and remain offline until the power was cycled at the LV crate. (This required a reboot of the analog system, as we belatedly discovered). 3 times during the month a telnet access to a terminal server port wedged, and ran in an infinite loop. We did not recover any instances of data corruption.

17. James Bellinger, December 2006 17 CMS Week Muon Alignment http://www.hep.wisc.edu/~jnb/cms/data/index.html The majority of the DCOPS data is available online in text form via the address above. It includes the fit information and the raw profiles.

18. James Bellinger, December 2006 18 CMS Week Muon Alignment Near-term Plans Finish rewriting the core of the DAQ to get better speed. It isn’t complete, but I estimate that a full readout cycle will take less than 15 minutes. It addresses the hung port problem. Integrate the DAQ into the DCS/PVSS system. I think I can use DIM protocols, but need advice. Finish the code to retrieve and assemble DCOPS data from the database.

19. James Bellinger, December 2006 19 CMS Week Muon Alignment Old Readout Scheme A bash script looped over the laser lines and submitted scripts which would telnet to the port for each DCOPS, issue the commands, and read back the raw profile. Another short program fit the profiles, if possible, and another shell script accumulated the resulting fit information and committed it to the omds database. Since this was sequential, the time required to complete a cycle for the +side was over an hour and a half, and a hanging process could stall the readout.

20. James Bellinger, December 2006 20 CMS Week Muon Alignment New Readout Scheme A C++ program loops over the SLM Lines, forking a job to manage each Line and fork off subtasks to read out particular ports. The lowest level forked task is still a telnet invocation. I estimate that this part of the readout will take 9 minutes for all the SLMs, not just the +side ones. When the SLM jobs have completed or timed out, the Transfer Lines will be read. Then the profiles will be fit, data committed to the database, and the DIM service updated with the new results.