Some comments on modeling of ATF magnet strengths and the ATF control system
Some comments on modeling of ATF magnet strengths and the ATF control system
================================================================================
GENERAL REMARKS
---------------
ATF magnetic measurements (MMS) data consist of field measurements (Tesla) for
dipoles and gradient measurements (Tesla/m) for quadrupoles ... no integrated
strength data is available. The actual effective length of many magnets is,
therefore, a matter for 3D magnet code modeling or conjecture, and fudge
(correction) factors, determined by analysis of observed beam behavior, are
used by the control system in computing magnet setpoints and in determining
magnet strengths from measured currents for the online model. These fudge
factors reside in an ASCII file which is read by the control system and used
whenever strength -> current or current -> strength computations are performed.
(NOTE: as of ~ DEC-1998 only the DR magnets use fudge factors)
Note that the control system software computes beam rigidity using 100/3 to
approximate the quantity 1e10/c, where c is the speed of light. This
introduces a small computational error (~0.07%) in the conversion of magnet
currents to magnet strengths and vice versa.
In the following text, all files referred to reside on the ATFAXP computer
unless otherwise noted. Also, I will use the following acronyms:
MMS = Magnetic Measurements data
BT = Beam Transport line
DR = Damping Ring
DATABASE CHANNELS
-----------------
Magnet currents are found in the database by referring to the appropriate
channel name. These names are of the form:
'DB_MAGNET::<magnet name>:CURRENT.DAC.WRITE'
where magnet name is something like QM11R. These values are like "BDES" values
in the SLC control system, in that they are the current that the power supply
is being asked to deliver. The "readback" or "monitor" current (like BACT) is
considered less reliable by the control system gurus.
The energy of the Damping Ring (DR) is found via the following channel name:
'SAD_CONTROL::DR_ENERGY'
In my own (Matlab) code which performs the current/strength conversions (see
USER$DISK1:[MDW.DOC]MATLAB.DOC) I use this energy for scaling both DR, BT, and
EXT magnets.
DR COMBINED FUNCTION BENDS
--------------------------
MMS data for the DR combined function bend (field [T] vs current) may be found
in the file:
USER$DISK1:[KURODA.ATF.MAGNET_B_I.DATA]COMBINED_B_MITSUBISHI.DAT.
I have been unable to locate MMS data for the gradient of this magnet versus
current, so in my COMFORT models I use the design SAD value ... however, in
order to make my online COMFORT model agree with SAD on what the DR tunes are,
I use a multiplicative factor of 0.932712 for the quadrupole content of the
BH1R family (making the BH1R family weaker in COMFORT than in SAD).
(NOTE: see MDW logbook for the full set of availble BH1R MMS data)
DR QUADS
--------
Quadrupole magnet currents (amps) for the DR are converted to gradients (T/m)
by the control system using linear interpolation of MMS data found in files
corresponding to the eight different varieties of DR quads.
The conversion from quad current to inverse focal length made by the control
system proceeds as follows:
o B'(quad_type) [T/m] is linearly interpolated from the MMS data
o |K1| [1/m] = {leff(quad_family) [m] * 10 [kG/T] * B'(quad_type) [T/m]}
/ {(100/3) [kG-m/GeV] * E [GeV]}
o |K1| [1/m] -> |K1| / (1 + fudge(quad_family))
Note that MMS data is associated with quad type, while effective lengths and
fudge factors are assigned on a quad family basis. When all of the DR quad
trim windings are powered, the fudge factors may even be assigned on a
magnet-by-magnet basis.
(NOTE: as of ~ DEC-1998 DR fudge factors are assigned by magnet family)
The MMS data files may be found in the following directory:
USER$DISK:[KURODA.ATF.MAGNET_B_I.DATA];
the effective length used by the control system for each quad family may be
found embedded in the control system software in the following file:
USER$DISK1:[ATFOPR.SAD]DR_MAG_LIB.FOR subroutine DRMAG_GET_EFFECTIVE_LENGTH;
the fudge factors used by the control system for the DR quads, computed by
K. Kubo, may be found in the following file:
USER$DISK1:[ATFOPR.SADDATA]DR_QUAD_CORRECTION.DAT.
Note that, because of the way the fudge factors for members of a family are
averaged together and some creative counting, the actual fudge factor for the
DR QF1R and QF2R families used by the control system is (N-1)/N times what is
specified in the file, where N is the number of quads in that family ... this
"adjustment" of the fudge factor for the QF2R family (by the fraction 25/26)
represents an additional 0.07% error in current/strength conversions for that
family. This all occurs in:
USER$DISK1:[ATFOPR.SAD]SAD_CTRL_DR.FOR subroutine DRQUAD_GET_CORRECTION.
The DR quad types and their associated MMS data files names are:
type name file
---- ---------------- --------------------
1 Hitachi type 1 Q_HITACHI_TYPE1.DAT
2 Hitachi type 2 Q_HITACHI_TYPE2.DAT
3 Hitachi type 3 Q_HITACHI_TYPE3.DAT
4 Hitachi type 4 Q_HITACHI_TYPE4.DAT
5 Tokin 18 cm Q_TOKIN_18CM.DAT
6 Tokin 18 cm (42p) Q_TOKIN_18CM_42P.DAT
7 Tokin 6 cm Q_TOKIN_6CM.DAT
8 Tokin 6 cm (42p) Q_TOKIN_6CM_42P.DAT
The family names, types, control system effective lengths, and fudge factors
(as of 7-DEC-1998 18:32) for the DR quads are summarized here:
family type N leff fudge
------ ---- -- ------ ----------
QF1R 1 28 0.0788 0.0126096
QF2R 5 26 0.1986 0.0011024
QM1R 3 2 0.1985 -0.0089336
QM2R 2 2 0.079 0.0226221
QM3R 2 2 0.079 0.0226284
QM4R 3 2 0.1985 -0.0089336
QM5R 4 2 0.1987 -0.0103301
QM6R 4 2 0.1987 -0.0103299
QM7R 7 2 0.0789 -0.0193019
QM8R 7 2 0.0789 -0.0193022
QM9R 4 2 0.1987 -0.0103299
QM10R 4 2 0.1987 -0.0103301
QM11R 6 2 0.1987 0.0055920
QM12R 8 2 0.0855 0.0099402
QM13R 8 2 0.0855 0.0099401
QM14R 8 2 0.0855 0.0099402
QM15R 6 2 0.1987 0.0055920
QM16R 4 2 0.1987 -0.0103301
QM17R 4 2 0.1987 -0.0103299
QM18R 4 2 0.1987 -0.0103301
QM19R 4 2 0.1987 -0.0103299
QM20R 3 2 0.1985 -0.0089336
QM21R 2 2 0.079 0.0226285
QM22R 2 2 0.079 0.0226285
QM23R 3 2 0.1985 -0.0089336
Finally, in order to make my online COMFORT model agree with SAD on what the DR
tunes are, I use yet another multiplicative factor of 0.9984212 for the QF2R
family (making the QF2R family weaker in COMFORT than in SAD).
All of this information is contained in the Matlab function file:
USER$DISK1:[MDW.MATLAB]GET_DR_QUADS.M
(see USER$DISK1:[MDW.DOC]MATLAB.DOC for more on Matlab functions for ATF).
BT QUADS
--------
Quadrupole magnet currents (amps) for the BT are converted to gradients (T/m)
by the control system using linear interpolation of MMS data found in a SAD
script file corresponding to the three different varieties of BT quads.
The conversion of quad current to inverse focal length is done in SAD itself
using this script, and, as of ~ DEC-1998, no fudge factors are used.
The SAD script file is located on the acsad3 unix machine as:
acsad3:/users/atfopr/sad/atflib.n
The effective lengths of the BT quads have been taken from the (design?) SAD
input file:
acsad3:/users/atfopr/sad/btdaihon.sad
The BT quad types are:
type name
---- ----
1 QICA
2 QICB
3 QICC
The names, types, and effective lengths for the BT quads are summarized here:
name type leff comments
----- ---- ------ ------------------------------------------
QD10T 1 0.1980 powered in series with QD32T <-----------+
QF10T 1 0.1976 powered in series with QF31T <---------+ |
QD11T 2 0.1978 powered in series with QD31T <-------+ | |
QF11T 1 0.1976 powered in series with QF30T <-----+ | | |
QD12T 2 0.1978 powered in series with QD30T <---+ | | | |
QF20T 3 0.1970 powered in series with QF21T <-+ | | | | |
QF21T 3 0.1970 powered in series with QF20T <-+ | | | | |
QD30T 2 0.1978 powered in series with QD12T <---+ | | | |
QF30T 1 0.1976 powered in series with QF11T <-----+ | | |
QD31T 2 0.1978 powered in series with QD11T <-------+ | |
QF31T 1 0.1976 powered in series with QF10T <---------+ |
QD32T 1 0.1976 powered in series with QD10T <-----------+
QF40T 1 0.1976
QD40T 1 0.1976
QF41T 1 0.1976
QD41T 2 0.1978
QF42T 2 0.1978 powered in series with QF43T <-+
QD42T 2 0.1978 |
QF43T 2 0.1978 powered in series with QF42T <-+
QD50T 1 0.1976
QF50T 1 0.1976
QD51T 1 0.1976
QF51T 2 0.1978
QF52T 3 0.1970
QD52T 2 0.1978
QF53T 2 0.1978
All of this information is contained in the Matlab function file:
USER$DISK1:[MDW.MATLAB]GET_BT_QUADS.M
(see USER$DISK1:[MDW.DOC]MATLAB.DOC for more on Matlab functions for ATF).
STEERING CORRECTORS
-------------------
Corrector (ZH,ZV) magnet currents in both the BT and the DR are converted to
field (Tesla) using a single-point magnetic measurement relating field (T) to
current (amp):
B(ZH,ZV) [T] = Cmms(ZH,ZV) * I [amp]
where Cmms(ZH) = 0.0108 [T/amp] and Cmms(ZV) = 0.0112 [T/amp].
The conversion to radians of kick
theta(ZH,ZV) [rad] = {leff(ZH,ZV) [m] * 10 [kG/T] * B(ZH,ZV) [T]}
/ {(100/3) [kG-m/GeV] * E [GeV]}
involves a value for the effective length that may be found embedded in the
control system software in the file:
USER$DISK1:[ATFOPR.SAD]DR_MAG_LIB.FOR subroutine DRMAG_GET_EFFECTIVE_LENGTH.
The value for the effective lengths of ZH and ZV correctors from this file
(0.1232 m) is sufficiently different from their SAD deck effective length
values (0.06 m) that this must be represent another type of fudge factor.