Calibration Status

This page is currently just a rough guide to the current state of calbrations.

Last modified: May 3, 2005, Nathaniel

Comments welcome! This is meant as a rough guide to what is going on and what works needs doing. Experts feel free to send comments to Nathaniel or Lisa for inclusion on the page.

Far Detector

Timing Calibration

Timing calibrations at the FD have 3 parts: strip-by-strip correction, correction for hardware swaps, and correction for time-walk.

Both Andy and Brian have provided strip-by-strip corrections, but only Andy's is probably compatable with the jump corrections. This hasn't been looked at in detail.

Andy has supplied a by-hand list of jump corrections (due to hardware swaps) that is now in the DB and valid up through about October 2004. After this time, the LI timing summaries were turned on, so that further changes can be tracked that way. Timing summaries are in the default generation model used by Simona when creating drift calibrations, so the status will be the same as those.

Andy has also supplied a time-walk correction. It is parameterized in ADCs, generic across all channels. This could probably be improved, but the Cambridge guys seem satisfied with it's performance.

PE (Gain) Calibration

Jeff Hartnell has provided a table of the FD gains, inserted into the DB on 2003-11-26. This table uses the mean/rms method to calcluate gains; only the most obviously wrong values were thrown out.

We know that substantial changes have happened since then (i.e. the newer PMTs at the north end have changed over time). Also, these numbers were never validated in any substantial way. Therefore, the PE numbers are to be treated with caution.

VA Nonlinearity Calibration

In principle, all nonlinearity is taken out with the light injection (SigLin) but in order to do so the PIN diode channels have to be linearized.

Giles Barr has written substantial code that can read charge injection data and fit the VA response curves to a kicker function. His code now runs semi-autonimously and can be used to generate constants to put into the DB. It's intended that this operate automatically at the FD near-time processing. To date, however, these values have not been committed to the DB.

When complete, he will commit only PIN diode channels to the public database, to keep the size down. However, he will privately archive the linearity of all tables, for hardware debugging.

Drift Calibration (SigDrift)

Drift calibrations are taken every 3 hours. Each of 20 LEDs is pulsed 1000 times in each of 16 LI boxes. At present, the mean response of each pulsed strip is compared to the mean response of the PIN diode to get the drift correction.

Data from LI is put into LI summary blocks. These blocks are pulled out of the data by code written by Phil Adamson and maintained by Simona Murgia. At present, much of the FD running has been processed, but not all tables are publicly accessable.

The drift tables are huge, and probably not amenable to distribution to remote (non-FNAL) sites. This represents a long-term problem for the calibration group.

Also troubling is the fact that the ADC/PIN ratio is not stable under hardware changes. In particular, it has been seen that wiggling the LI injection fibres can cause a change in the light level to be seen, destroying the ADC/PIN ratio. This makes creating a consistent drift correction very difficult.

As of May, 2005, it is the calibration group's top priority to investigate improved ways to do this calibration. See the plenary talk of the

Linearity Calibration (SigLin)

Linearity is corrected by taking the ADC/PIN for each channel for varied light levels, then correcting to the slope at low light levels to get "linearized ADCs (SigLins)". Data is called a "Gain Curve" and is scheduled in the LI system to run every 30 days.

Data is processed as the Drift calibration. Again, the software for applying the curves is written by Phil Adamson.

These procedures may change slightly; in particular, high-statistics running to do PE calibrations is important.

Strip-to-strip Calibration (SigCorr)

Most of the work on strip-to-strip has been done by Phil Symes, with help from Chris Smith and heavy (but constructive) critizism from Jeff Hartnell. Strip-to-strip calibrations are done by comparing the mean ADC output of through-going muons on a strip-by-strip basis, after correcting for path length, attenuation, and zeroes. Full calibration of the Far Detector requires ~3 months of cosmic ray data.

At present, a single calibration has been performed, using 2004 data. This calibration has gone through numerous updates as bugs were found in the analysis. It is envisioned that eventually we will go to a rolling calibration, but work has not really begun on such a project in earnest.

Attenuation Correction (SigMap)

Attenuation correction is done using the Mapper data. This calibration now only attempts to correct for variations along a strip, leaving the strip-to-strip component entirely to the SigCorr calibration. Therefore, the mapper data is normalized to the center of each strip.

Mapper data is NOT stored in the database; this information remains hidden away by the Mapper Czar (Leon Muralem). Leon has provided fits to the mapper data in the form of a double-exponential.

These tables were updated in January 05 to complete the Far Detector. These calibrations can probably be considered final.

Detector-to-Detector Correction (SigMIP)

Jeff Hartell is performing this calibration via stopping muons. This calibration returns a single number for each detector. This number is used, among other things, to tune the light level of the detector in MC. For more info, see Jeff's thesis (currently being written).

Temperature Measurement

Temperatures at the FD are measured in several ways: off the VFBs directly, using the DAQ at the beginning of every run, and via thermocouples in the cavern, read out by the DCP.

FIXME: where are the thermometers? How many? Which two are accessible by the DCP DB tables?

Near Detector

Timing Calibration

At present, we do not know if we need a timing calibration at the ND. By design, all front-end channels recieve the TCLK (53MHz) clock syncronously, to within a few nanoseconds. Cable lengths in the detector vary by only a few more nanoseconds, so the total timing shift from channel to channel is probably not more than 1 53MHz tick (i.e. 1 bucket).

A timing calibration may be an improvement, but this analysis requires a volunteer.

PE (Gain) Calibration

A one-time gain calibraiton was done, using the mean/rms method, by Tobi Raufer, on 2004-11-14. This calibration has not been verified, and may contain flaws. In particular, the front-end hardware has been considerably improved since that time. As above, this calibration should be taken only as indicative, and the number of PEs per channel is only approximate.

Drift Calibration (SigDrift)

Drifts are done as in the FD, in principle.

Currently, no database tables have been generated.

Linearity Calibration (SigLin)

Linearity is done as in the FD, in principle.

Currently, no database tables have been generated.

Strip-to-strip Calibration (SigCorr)

Phil Symes has generated a strip-to-strip calibration for the ND as for the FD. The time taken to generate such a calibration is approximately 10 days of cosmic muon data.

However, many problem still exist with this analysis. Proper batch-processed ntuples have not been used, due to lack of availablity. Strips in the spectrometer pose difficulties due to multiplexing. Strips in the full-instrumented planes are difficult to get statistics for, and may be prone to bad attenuation corrections.

This analysis is still awaiting a chamion to validate these issues.

Attenuation Correction (SigMap)

As the Far Detector, fits to the mapper data are used. These data are all in the database and can be considrered production-quality.

However, because ND strips are read out only on one end, this correction is much more important in this detector. (Two-ended readout in the FD ensures that an error in reconstucted position results in only a small shift in total energy, but at the ND this error could be large.)

Because the statistics are large, it's envisioned that cosmic ray muons could be used to calibrate the attenuation maps, or at least verify them. Such an analysis is awaiting a champion.

Detector-to-Detector Correction (SigMIP)

In principle, the techniques used by Jeff Hartnell at the FD and CalDet can simply be applied here. In practice, however, the important part of the job is the analysis of systematic errors, which is a major undertaking.

It is hoped that Jeff can be persuaded to work on this after he graduates. Otherwise, this is a major hole in our calibration chain.

Temperature Measurement

FIXME: No idea. Probably not important.

CalDet

Timing Calibration

FIXME.

PE (Gain) Calibration

FIXME. Exists, probably for all run modes.

VA Nonlinearity Calibration

FIXME. Probably not done.

Drift Calibration (SigDrift)

Drift calibration is processed differently for CalDet data. Only Phil Adamson has the know-how to process this data.

Linearity Calibration (SigLin)

As Drift.

Strip-to-strip Calibration (SigCorr)

The procedure for this calibration step was developed by Ryan Nichol and Chris Smith. CalDet strip-to-strip is complicated by the fact that the strips are oriented in x and y rather than u and v. This means that the average cosmic ray muon pathlength is different in each view. The pathlength differences can be corrected for using the track angle and simple geometry considerations. However the zero reconstruction is not straight-forward and requires knowledge of the light level. The procedure adopted is therefore iterative and is described in numi-971 and numi-972.

Attenuation Correction (SigMap)

Ad-hoc at the moment.

For the strip-to-strip calibration, the attenuation pattern was mapped out using PS muons while the detector was in the T7 beamline. In this location, the muons were approximately perpendicular to the z-axis of the detector.

Detector-to-Detector Correction (SigMIP)

Jeff Hartnell has devoted much work. See his thesis.

Temperature Measurement

The dependence of detector response with light output was measured using cosmic ray muons while the CalDet was in an enclosure and the PS was off. A set of calibration constants was calculated every hour or so over a 1-2 week period and the average fractional difference between the sets of constants was calculated and plotted versus air temperature (as measured by a radio shack probe). The dependence was observed to be -0.3%/deg C. This is partly accounted for by the VA dependence of the PIN response during LI drift point data taking, which was independently measured at Oxford to be -0.22%/deg C. Thus the scintillator is assumed to be responsible for the remainder (-0.08%/deg C). Independent measurements of the scintillator light output with temperature at Caltech showed a -0.3%/deg C dependence, which appears inconsistent with the CalDet measurements. However, in the Caltech studies the scintillator was allowed to reach thermal equilibrium before the data was taken. For CalDet, the temperature reported is simply the ambient air temperature.


Fermi National Accelerator Laboratory Magnet Logo