These are the results of a global linear fit for beam cavity monitors 1 and 2. The fit is "global" in the sense that data from all three BCM calibration runs are used. Each datum was derived using local unser monitor zeroes.

The final BCM calibrations are here:
BCM2 calibration from global fit Chi^2/DOF = 1.08
BCM1 calibration for first part (runs 46996 to 47335) Chi^2/DOF = 0.57
BCM1 calibration for second part (runs 47336 to 47764) Chi^2/DOF = 0.95
These values should be loaded for the next replay. The charge derived from BCM2 will change only of order 0.1%, so anyone not doing a luminosity scan won't even notice. The two calibrations for BCM1 may be helpful, however, except for the day or so that it wasn't working at all.
About errors: For charge derived from BCM2, I assign a point-to-point error of +-200 nA to be added in quadrature with the 0.2% systematic error of the Unser wire calibration. Hence, for a cross-section determination at 20 microA the charge error would be 1%, while at 100 microA it would be 0.28%. For the contribution of the charge error to an L-T separation, only the point-to-point error is significant.

The figure shows the residuals of the calibrated beam cavity monitor output versus current. One can quickly see that the BCM2 residuals looks great. There is no evidence of nonlinearity (within the errors!) between 10-110 microA. The Chi^2/DOF is 1.08. This is what we're going to use for charge determination in FpiII.
The best fit to BCM1 would require at least two different calibrations. (This is better seen by plotting the residuals as percentage of the current: alternative figure. ) This is no surprise since we know its gain shifted over 3% in the middle of the run. For what it's worth, for a global fit the Chi^2/DOF is 24.7. The time-dependent calibrations for BCM1 are provided above.

Here the BCM residuals are plotted versus sample number (ie, time). The long breaks between the various calibration runs are demarcated by green vertical lines. The BCM2 calibration is consistent with a constant calibration hypothesis at the level of +-200 nA. However, BCM1 looks particularly confusing when plotted this way! (In the presence of gain shifts, the residuals make more sense expressed as percentage of the current: alternative figure. Here you can clearly see the gain shift evident at the start of the 2nd BCM calibration run.)

These are the hclog links from the FpiII run relevent to the Unser wire calibration.
Unser wire calibration data
Unser wire calibration results

These are the hclog links from the FpiII run relevent to the BCM's. Any BCM calibrations found here are now obsolete and superseded by the above global analysis:
BCM calibration data run 47053 (July 11, 2003)
Initial FpiII BCM calibration
Updated FpiII BCM Calibrations (local zero analysis)
BCM1 gain is shifting
Summary of BCM1 work
BCM calibration data run 47522 (July 31, 2003)
Updated BCM1 calibration from run 47522
Final BCM calibration data run 47739 (August 11, 2003)

Sept 5, 2003, by Dave Mack