Tagger field mapping analysis
D. Sober
Postings in reverse chronological order.
Results of Fall 2016 survey of beam elements - 30-Jan-2017
Calculations relating to possibility of scattering to counters in same plane - 02-November-2016
First results from tagger Monte Carlo simulation -- 13-Jul-2016
Derivatives files for high-energy rays (no quadrupole) -- added 03-Jun-2016
High-energy rays (6.69 GeV < E_e < 9.00 GeV) -- Calculated 07-Jan-2016
Note: These rays use a different SNAKE setup, with the
EXIT field box instead of the FOCAL field box.
[The derivatives for low-energy rays (0.18 GeV < E_e <6.99 GeV)
were posted on 26-June-2015.]
Tagger beam pipe collimation of low-energy electrons
-- rev. 18-May-2016
Effects of beam pipe and dipole vacuum entry pipe on low-energy
electron trajectories, with and without quadrupole
(Beamline_collimation.pdf)
Plot of limiting angles from above document
(Collimation_angles.pdf)
Beam optics derivatives with quadrupole -- 11-May-2016
Plots of derivatives with quad gradient = 0 and -62.5 G/mm
(From Quad_and_radiator_position.pdf, 26-Feb-2016)
- Derivatives w.r. to x0 and x'0 (derivs1-4qyn.pdf)
- Derivatives w.r. to z0 and z'0 (derivs5-8qyn.pdf)
Comparison of raytracing results at 12 GeV using measured map
and Tosca field --07-March-2016
Plot difference in x-intercept, energy, angle
(Data-Tosca.pdf)
Quadrupole: Comparison of using Tosca field map and simple analytic form -- 03 March 2016
Demonstrate that the analytic form employed for the results of my
note of 26-Feb-16 is just as good as using the Tosca field map
(Quad_Tosca_and_linear.pdf)
Raytracing with the quadrupole, and effects of radiator
position -- 26 February 2016
A first look at using the quadrupole magnet
(Quad_and_radiator_position.pdf)
Also discusses differences between goniometer and amorphous radiator
positions.
Presented at beam meeting of 29 February 2016
Efficiency of the tagger fixed array -- 02 February 2016
Efficiencies, gaps between counters, and effect of dipole magnet
poles
(Tagger_ratios_and_gaps.pdf)
Presented at beam meeting of 02 February 2016
Some input files for SNAKE at JLab -- 27 July 2015
Most of my SNAKE calculations have used an old version of the code
that I have modified for my convenience. The input "directive" files
are of slightly different format than those more recently used in
the JLab version. I have run selected calculations through the most
recent JLab SNAKE to which I have access (provided by John LeRose in
August 2013), and found no significant differences from my
calculations at CUA.
In case anyone is interested in using the tagger map files with
SNAKE, here are some sample input files that have been tested to
work with my map files at JLab:
- Directive files
These files contain the names of the map files, which can be
found on the GlueX docdb, and also on ~sober/HallD/maps at JLab.
(The parameters in lines 2 and 3 are relevant only for spin
rotation of protons, and are ignored here.)
- Trajectory files for E0 = 12 GeV
These files are requested during execution.
Some new analysis notes -- July 2015
Revised derivatives from raytracing after probe recalibration
-- 26 June 2015
-
ascii file of 8 derivatives d{x,x'}/d{x,x'} and
d{z,z'}/d{z,z'}
-
ascii file of the 4 non-zero "crossed" derivatives
d{x,x'}/d{z,z'}
-
plots of derivatives d{x,x'}/d{x,x'}
-
plots of derivatives d{z,z'}/d{z,z'}
-
plots of derivatives d{x,x'}/d{z,z'}
Tagger dipole contour plots -- 19 November 2014
[Note: See later posting "Analysis of the Field Maps (2 July
2015)" for revised contour plots]
I have made contour plots of the uniform-field region for the
three measured field maps, using contour levels spaced by 20 gauss
at 1.5T and equivalent fractions of full field at 0.75T and 1.7T.
Specifically,
0.75T: Bmax = 0.7506 T, contours =
.751,.749,.748,.747,.746,.745,.74,.73,.70,.60,.50,.40.,.30
1.5 T: Bmax = 1.5017 T, contours =
1.502,1.500,1.498,1.496,1.494,1.492,1.490,
1.48,1.46,1.4,1.2,1.0,.8,.6
1.7 T: Bmax = 1.7021 T, contours =
1.7023,1.7000,1.6977,1.6955,1.6932,1.6909,1.6887,1.6773,1.6547,1.5867,1.3600,1.1333,.9067,.6800
The structure with a period of 4 cm in X is a consequence of
imperfect probe calibration at the level of a few gauss. It should
not have any detectable effect on the raytracing calculations.
-
0.75 T contour plot (pdf file)
-
1.5 T contour plot (pdf file)
-
1.7 T contour plot (pdf file)
-
3 plots on one page for easy comparison (pdf file)
New field analysis results -- 23 October 2014
- B(at NMR probe)/E0 and B(at NMR
probe)/Current using scaled field
(Table and plots)
By ray-tracing through the measured and Tosca fields, I have
calculated the scale factor needed to multiply the field to give
a full-energy deflection of 13.400 degrees.
By interpolation in the field maps to the nominal position of
the NMR probe (x=-12.8 cm, y=-283.0 cm) I have calculated the
NMR value required to steer the full energy beam to the dump at
E0 = 12, 13.6 and 6 GeV, corresponding to the nominal
map fields of 1.5, 1.7 and 0.75T. Note that the
interpolated field at the probe position at 1.7T differs by
about 9 gauss from the actual NMR reading recorded in the
mapping data files -- not a big effect (0.05%), but bothersome.
The Tosca field results vary significantly (by more than 0.1%)
from the measured field. This is presumably related to the fact
(noted earlier) that the measured fields increase slightly with
x, while the Tosca fields do not -- probably because the poles
deflect slightly when the field is on.
- Comparison of B vs y at different x with average B
and B at NMR probe position
(plots)
The large difference in B(at NMR)/E0 at 6 and 12 GeV
seen in the previous plot is due to a substantial difference in
the field shape. The square data point on each plot shows the
interpolated field at the NMR probe position. At 0.75T this is
very close to the maximum field in the gap, while at 1.5T and
1.7T it is close to a minimum versus x. The horizontal line
shows the average field along the line x=0.
The breaks in B seen at the transition between Configurations 1
and 3 (near y=0) are not as big as they look at first glance --
typically 2 or 3 gauss on each curve. I have not made any
attempt to smooth things at that level.
The increased non-uniformity of B at 0.75T is surprising, but I
saw similar effects with the Hall B tagger magnet. My
interpretation is that it is due to the fact the permeability
begins to decrease at low excitation, so that the lowest-energy
field configuration is not necessarily the most uniform.
- Shifts of E/E0 at the focal plane for
13.6 GeV and 6 GeV relative to 12 GeV
(plot)
This plot shows that the energy calibration of the tagger focal
plane as a function of E/E0 is essentially
independent of E0. Using the 1.5T-derived map for all
E0 will give an error of less than 3 MeV, which is
less than the resolution of any microscope or fixed-array
channel.
Summary and directory of field maps for raytracing -- 21 July
2014
Note: The "main" maps indexed here have been replaced by the new
versions following the probe recalibration - see "Analysis of the
field maps" (2 July 2015).
Derivatives of beam optical quantities -- 21 July 2014
Note: This link references the revised file (after probe recalibration)
posted on 26 June 2015 -- see above.
At the request of Richard Jones, I have calculated first
derivatives of the focal plane quantities xFP, x-angle,
zFP and z-angle, with respect to x, x', z and z' at the
radiator, as a function of electron energy (assuming E0 = 12 GeV).
These derivatives were calculated by taking positive offsets of 1
mm in x and z and 1 electron characteristic angle in x' and z',
and comparing with the central ray.
I have checked that the dependence is essentially linear except
for one case: dxFP/dx'0 vanishes at the
point-to-point focal plane, which intersects the nominal focal
plane at Ee ~ 3.3 GeV. In this region, the calculated
first derivative is small and not very meaningful.
The table contains only 8 of the possible 16 derivatives. The
derivatives of z and z' with respect to x and x' are exactly zero.
The derivatives of x and x' with respect to z and z' are small and
randomly distributed due to precision uncertainties of order 0.5
ppm, and have also been omitted.
Using the new fixed-array counter positions from the mounting
plates -- 02-July-2014
The fixed-array counters were installed in the tagger hall on June
30. Only then did it come to my attention that the counter positions
on the mounting plates are not the same as those in the
table in the GlueX Wiki (based on our May 2013
calculations). Many of the counters were shifted along their nominal
electron trajectories either (a) to increase the spacing between
phototube assemblies or (b) to avoid other conflicts with mounting
hardware. The net effect of these changes on the energy calibration
is small but not negligible, as described below.
- Table of new counter positions
(text file)
(Excel file)
- Table of fixed-array counters with nominal and new energy
boundaries and angles
(text file)
- Plot of difference between actual (mounting-plate) and
nominal (Wiki table) fixed-array counter positions
(pdf file)
The X and Y values are in Focal Plane coordinates. The blue +
symbols show the nominal positions, and the red circles show the
actual positions. The diagonal dashed lines show the normal
Microscope region, in which the fixed-array counters are not
installed.
The principal differences are
- Counters 1-131, which originally alternated between y=-8
and -13 cm, now alternate between -8 and -18 cm (counters
1-80) and then cycle between the 3 planes -8/-13/-18 cm for
counters 81-131.
- Counters 194-274, in the "sampling" region below the
Microscope, now (mostly) alternate between -8 and -18 cm
instead of all being at -8 cm.
- Plot of revised energy boundaries of counters
(pdf file)
This is a revised version of the plot of 26-June-2014 posted
below. It shows the upper and lower energies relative to the
nominal central energy, for both nominal and the revised
calculations. The new calculations include both the new raytracing
results and the corrected counter positions.
The largest change is seen in part of the sampling region (E_gamma
between 5.6 and 7.3 GeV), where the alternation of the counters
between the -8 and -18 cm planes now gives a substantial
alternation in the energy limits. This is seen more clearly in the
following plot.
- Plot of energy width and energy spacing of counters
(pdf file)
The energy width (E_high - E_low) is plotted with blue squares,
and the energy spacing (E_high_[n] - E_high_[n+1]) with red
triangles. We see that the large variation in energy limits
between 5.6 and 7.3 GeV is due to a variation in the energy spacing,
which alternates between 57 and 63 MeV, rather than the nominal 60
MeV. Note that the alternation between the y=-8 and y=-18 cm
planes has little effect on the energy width.
A more disturbing effect is seen between 8 and 10 GeV, where for
some counters the energy spacing is smaller than the energy width,
i.e. there is a partial overlap between adjacent counters even for
central rays. For the full-coverage region above the Microscope
(9-10 GeV) this overlap never exceeds 0.3 MeV out of a total width
of 15 to 21 MeV, i.e. less than 2%. In the 8-9 GeV region, which
will be installed only if the microscope is moved to lower energy,
the overlap rises to 5%, plus one channel at 10%. We may wish to
consider re-making the mounting plate for that region.
Some useful files -- 23-June-2014
By popular request at this morning's beamline meeting:
- Position and angle of rays at tagger focal plane, 10 MeV steps
(Excel file)
- Position and angle of rays at tagger focal plane, 30 MeV steps
(Excel file)
- Table of fixed-array counters with nominal and new energy
boundaries and angles (text
file) [Note: This file should be
replaced with one with new counter positions -- see
2-July-2014 postings]
The limits in the counter table are for central (zero-angle)
rays only, and the counter thickness has been ignored. This
table includes fixed-array counters in the normal Microscope
position (Options 2 and 3). The counter widths for E_gamma
between ~7.9 and ~8.2 GeV are not entirely consistent. I will
update the table when I figure it out.
First raytracing results -- 23-June-2014
The processed field maps (described at the last meeting on
5-June-2014) have now been tested with SNAKE for the full range of
electron energies, and agree quite well with previous
calculations, as described below.
- Location and description of map files
For drawings and descriptions of the field boxes, endplanes and
trajectories, see the files I posted for the previous beamline
meeting (5-June-2014).
In case anyone is interested in using the map files, I have
copied them to the JLab CUE in my directory
/u/home/sober/HallD/maps .
A detailed description of the file formats and their intended
use is given by the file
Field_boxes_and_headers.pdf , which I have also copied to
the /maps directory.
- Normalization of field maps
To make direct comparisons possible, each field map must be scaled
so that the full-energy electron is deflected by 13.400 degrees.
The necessary factors are as follows:
Nominal field E0[GeV] Measured
map factor Tosca map factor
1.7 T
13.6 1.00114
 :
0.998556
1.5 T
12.0
1.00029
0.996795
0.75T
6.0
1.00097
0.999000
- Agreement of new raytracing with reference rays
Figure
1 shows some differences between the new rays using the
1.5T map and the "reference rays" (used in calculating the
fixed-array counter placement) at the same electron energy.
Except for the lowest energies (Ee < 1 GeV), the
angles agree to better than 0.1 degree, and the perpendicular
position xperp=xFPsin(theta) agrees to
better than 2.4 mm. In the Microscope region (3 < Ee
< 4 GeV), the xperp differences are between 1.6
and 1.9 mm, or slightly less than one Microscope energy channel.
- Consistency of the three maps (1.7T, 1.5T, 0.75T) as a
function of E/E0
The curves labeled "1.7T - 1.5T" and "0.75T - 1.5T" in
Figure
2 show the differences in xFP and angle at the
same value of Ee/E0 between the
corresponding pairs of maps. The angle differences are
negligible (<0.04 degree), and the largest xFP
difference is about 2.5mm, which corresponds to a small fraction
of a Microscope energy channel. The conclusion is that the
raytracing results using the 1.5T map may be scaled to any
incident energy without significant loss of accuracy.
- Consistency of field maps with Tosca fields
Figure 2 also shows the differences in xFP and angle
calculated using the measured field maps and using the Tosca
field at the same excitation. These differences are again small
(less than one Microscope channel). The results for the three
fields are quite consistent with one another, and can be
explained in sign and order of magnitude by the following
argument:
- The full-energy electrons travel mainly through the region x < 0. (see
figure from June 5 update)
- The low-energy electrons travel mainly through the region x
> 0.
- As previously noted, the measured field between the poles
increases slightly with x
(Figure 3) , probably because of a very slight
non-parallelism of the poles.
- The Tosca field shows no x-dependence.
- The field maps are normalized by the requirement that full-E
electrons deflect by 13.4 degrees.
- Therefore the low-E electrons deflect more in the measured
field than in the Tosca field
A difference of ~10 gauss (consistent with Figure 3) between the
average field seen by the full-E and low-E electrons is sufficient
to explain the observed differences.
The map-vs-Tosca differences shown here are about a factor of
two smaller than the differences from the "reference rays" shown
above. The reference rays were in principle calculated (not by
me) from a different version of the 1.5 T Tosca map (with the
grid in room coordinates.) I don't know the details. In any
case, the differences are not large, and I don't think that this
mystery is worth pursuing.
- Energy boundaries of fixed-array counters (plot
revised 26-Jun-2014)
Note: This plot was calculated using the old (2013) counter
position table, not the actual mounting-plate positions.
See revised plot of 01-Jul-2014.
Figure 4 shows the upper and lower energies relative
to the nominal central energy for each counter,
plotted vs electron energy. The dashed curves show the nominal
boundaries (symmetric about 0 by definition), and the solid
curves show the results of the new raytracing analysis. Note
that only on-axis rays have been traced for
this analysis -- the actual boundaries will be diffuse and
somewhat larger.
The energy regions have been labeled with the counter widths.
The region populated by the 4 mm counters is normally occupied
by the Microscope, whose fibers are 2 mm wide. Relative to the
nominal boundaries, it is seen that the new analysis shifts the
channel energy by no more than half a fixed-array counter (or
one Microscope channel.)
- Eliminating angle error
The angles of the fixed-array counters are determined by the
holes drilled in the mounting plates. Although the changes in
angle relative to the reference rays are small (< 0.1 degree
for Ee > 1.3 GeV, <0.05 degree in the
Microscope region and beyond), it is worth seeing whether the
angle error can be eliminated by shifting the x position of all
or some of the mounting plates.
Figure 5 shows the distance in cm by which each part of
the focal plane must be translated in order that the local
counters are aligned with the electron trajectory angles. (The
enhanced region on the plot shows the position of the
Microscope.) There is no single shift that would improve things
everywhere, but, since the required shifts are increasing in the
downstream direction, we may wish to consider inserting gaps
between the mounting plates to improve the average angular
alignment for each plate. However, given that our precision in
angular alignment of the counters is in most cases worse than
the error, this is probably a waste of effort.
Update on analysis of field maps -- 05-June-2014
I have completed the analysis and manipulation of the field maps,
and am ready to begin raytracing calculations.
The following tables and figures illustrate the results.
At each excitation I have produced 4 "field boxes" in the format
used by the SNAKE code. The boxes are named
- "Entry" - where all trajectories enter the field; based on
config. 6
- "Main" - where most of the deflection occurs - based on
config. 1-4
- "Exit" - where the highest energy electrons leave the field;
based on config. 5
- "Focal" - between the main field map and the focal plane;
calculated by Tosca
The Entry and Exit boxes were constructed using the measured fields
of configurations 5 and 6 respectively, extended by Tosca
calculations scaled to the measured field at points of overlap.
Depending on its energy, each trajectory is propagated through 3
field boxes:
Low energy (E < 7.5 GeV): Entry + Main + Focal
High energy (E > 7.5 GeV, including the full-energy electrons):
Entry + Main + Exit
The boxes are illustrated and described in the table and figures
below.
Field mapping -- Presentation to GlueX Collaboration Meeting, 21-Feb-14
(pdf file)
Progress summary (all regions completed 12-Feb-2014)
Entry beam pipe region (Configuration 6) (added 06-Feb-2014)
I have compared the measured points with the nominal beam line,
assuming that the pole center (origin of mapping coordinates) is at
the position X = -29.810 cm, Z =629.402 cm in room coordinates and
using an angle of 6.5 degrees. The measured points are all within 5
mm of the beam line, differing by about 2 mm at the pole root, and
the slope of the line relative to the beam line is about 7 mr (0.4
degrees).
The fields in this region at the three excitations (0.75, 1.5 and
1.7 T, all scaled to 1.5 T) agree very well with each other and also
with Tosca, if the Tosca calculation is shifted by -4 mm (upstream).
Configuration 3 (mapped Feb 3-4, 2014) (added 07-Feb-2014)
This configuration is nearly the mirror image (through Y=0) of
Config. 1, and the field looks very similar to Config. 1, except
that the variations between the probes (uncalibrated) are even
larger, easily visible in the plots versus X. Most of this effect
goes away when I apply the nominal calibration factors.
Here are some sample plots. All 3 excitations are very
similar.
- 1.5 T - uncalibrated
- 1.5 T - calibrated
Comparison with Tosca, and alignment tests (added 24-Jan-2014)
- General agreement with Tosca calculations
The field shape vs x near the long exit edge is in good
agreement with the Tosca calculations, from full field down to
about 300 gauss
I do not have actual power supply currents to compare with Tosca,
but the ratios of field to power supply "PPM" values (using the
prob calibration data at the center of the pole) are closely
proportional to the Tosca B/NI ratios at 0.75, 1.5 and 1.7 T,
which tells us that the saturation effects at high field are
predicted correctly by Tosca. For B < 0.75 T, there is
negligible saturation, and the ratio should be constant. The
ratios at 0.05 and 0.1 T do not lie on the expected line, implying
that there is a small (~2700 unit) offset in the "PPM" value, probably
due to hysteresis effects.
- Alignment of Configuration 2
Matching the decrease of the field from maximum to ~1/2 max. with
the Tosca calculation, the required offset in x is
22.44 cm at y=0,
22.48 cm at y=100cm,
22.49 cm at y=200cm
as compared with the nominal offset of 22.40 cm. These offsets are
the same for all 3 fields.
- Alignment of Configuration 1
The x alignment is hard to test, since the field falls by only a
fraction of a percent at the minimum and maximum x values of
Configuration 1, but this small drop is not in good agreement with
Configuration 2 (at high x) or Tosca (at high or low x).
A shift of the Configuration 1 x-coordinate by 0.7 cm
gives much better agreement at both ends (see figure.) This
effect is consistent at y=0, 100 and 200 cm, and at all 3 field
values.
This shift is not important, because the field is essentially
independent of x within Configuration 1 except for these
end points. We can ignore the highest 3 x-values
from Configuration 1, since we have the Configuration 2
measurements for those points. (There are no electron trajectories
in the region of the lowest x values.)
The Y alignment of Configuration 1 is easy to compare with Tosca,
as there is a significant falloff at the last two values, 306.4
and 308.9 cm. (The 308.9 cm setting is missing from the 1.5T map.)
Agreement with Tosca requires shifting the Y value of the measured
points by about +0.85 cm. Configuration 2 has no
points near a Y field boundary.
Configuration 2 (mapped Jan. 14-15, 2014)
This configuration consists of two files at each excitation: a
rectangular region and an appended triangular region.
I have combined the two files onto a single grid, and plot the
results below.
At the highest X and Y values, the field drops to 0 because the
measurements are made over a triangular region (see the "dotplots"
to visualize this.)
In the fringe field region, there is an interesting periodicity of
about 70 cm, which may be due to the magnetic effects of some
hardware like the coil clamps (or the mapping apparatus?)
Configuration 1 at 1.5 Tesla (mapped Jan 8, 2014)
Old information - 12 December 2013
Here are some files which may be helpful in analyzing the
mapping data.
Updated versions of these files are available at JLab on
~sober/HallD/tagger_mapping, in case you are able to read them
there.
map_analysis.txt
(Description of program mapsort and plotting procedures)
dmapsort.f
(Fortran program)
Configuration1.txt
(sample data file from Tim Whitlatch)
plotvsx.gpl
(sample gnuplot input file to plot output of dmapsort vs.
x)
plotvsy1-3.gpl
(sample gnuplot input file to plot output of dmapsort vs. y
using 3 output files)