Nikolay (and all on the T474X team...)

   My appologies for the long delay in replying to your message with your
simulations of the 10D45 magnets for the chicane in the test experiment
T474X planned for next year for the ILC-ESA project.  We had a run going
here and only in the last weeks could I turn my attention to the magnets
for the next stage of T474.

   The bottom line of this message is I want to use a different set of
10D37 magnets (assembly drawing attached), not the 10D45 we originally
planned to use.  This note is mostly to Nikolay about magnet details and a
list of questions. Mike Woods is away until next week on holiday and he
hasn't heard about the 10D37 yet, so this decision will be preliminary
until he returns, and until I hear from you about the simulations of field
uniformity. I will be on vacation from Aug 17 until Sept 4, but I look
forward to further discussions and comments from you or any others of the
T474X team in the next while.


  The main reason for not using the 10D45 are that they need a lot of work
to bring them into operation.  There was water condensation from the beam
pipes for many years at SPEAR, and after they were decomissioned the
magnets were not stored properly and the pole faces have thick corrosion
which would have to be machined off.  To achieve the desired field
accuracy would require careful work, new drawings, and significant time in
the machine shops. Also the magnets are not mechanically identical.  They
were made with many irregularities in construction (holes in odd places,
not all the same length, etc) and I think the quality of the field, and
the similarities of the magnets that we intend to run in series on a
singel power supply would be not as good as for the 10D37.  The 10D37 are
in pristine shape, and can be used after only routine cleaning and
maintanence of covers, etc.  We can make field measurements starting soon
with no delays for modifications.  They are also smaller and lighter (2500
lbs instead of 5000 lbs) making the support stands easier to engineer for
rigidity and seismic bracing.

   The 10D37 (12 original) were made in 1971 (engineered by Dieter Walz,
who is still here at SLAC and a member of T474X) to transport beam to the
SPEAR ring.  They were removed and put in good storage more than 10 years
ago.  I have located 4 of them and requested and expect to get permission
to use them. They are carefully made and all mechanically identical.
A directory of the drawings scanned to PDF files from the SLAC
archives is at http://www.slac.stanford.edu/~arnold/10D37/

When considering magnets for T474X we have to keep in mind some basic
requirements for this experiment that are similar to, but not identical to
the requirements for the eventual new magnets that would be built for the
ILC.  The exact layout of the ILC chicane is not set yet, but it will
require multiple magnets with deflections of 0.5 to 1 mr giving beam
offset of ~5 mm and needs to operate over a range perhaps 45 to 500 GeV.

  The plan for T474X is to build a 4-magnet chicane and measure beam
offset in the middle with new BPM's.  The length of the chicane, strength
of magnets, deflection angles, and beam offset in the middle are all
similar to, but not identical to parameters at ILC, mainly because we will
be operating at a maximum energy of only 28.5 GeV.  For T474X we have
selected chicane parameters to give similar offset (+/- 5mm) to that
proposed for the ILC so as to test the performance of new BPM's and movers
etc.  The chicane is sized to use magnets of ~1 m length with fields of
~0.1 T with bend angles of ~1.25 mr.  We plan to operate the 4 magnets in
series, so having them be nearly identical is important.  The operational
range range will be small, a few GeV perhaps around 28 GeV. The operation
plan envisioned for the ILC is to cycle the chicane magnets back and forth
frequently between +B and -B to extend the energy resolution for a given
bend angle and to help measure and control systematics.  We want to test
the cycling operation in T474X.


Some features of the 10D37 magnets that are important for us:

  1) They are H style with saddle coils, with dimensions:
 pole length 37 inches (93.98 cm)
 gap width   10 inches (25.4 cm)
 gap height 3 inches (7.62 cm) This gives adequate space for beam pipe and
field monitors (NMR probes, moving wires, etc. ), and provides appropriate
Bdl at field values similar to that in the lower energy range of the
future ILC chicane.

  The steel is specified as C1010 with not more that 0.13% Carbon. This is
a standard spec for reasonably good magnet iron.  There is no record of
the exact steel lots, and Dieter Walz said there was no attempt to keep
track of different lots of steel from the manufacturerer, so within the
uncertainties of the spec there could be small variations in magnetic
properties bewteen magnets.

  2) Measurements of other magnets in this family in 1991 using flip coils
show uniformity of Bdl vrs x of few parts in 10**4 over x = +-1 inch in
the central region.

  3) For operation in T474X to bend 28.5 GeV beam by +/- 1.25 mr giving
offset in mid chicane of +/- 5mm would require ~185 A to produce ~0.118
Tesla-m.  For use at SPEAR they were routinely operated at 1200 Amperes
and Bdl of 0.75 Tesla-m.  We would be operating them at 1/10 of the B
value where saturation begins so saturation effects should be very small.

  4) While many features are ideal for our use, there are a few details
that make the magnets less than perfect for a precision energy
measurement.  Some simulations like the ones you have done for the TESLA
spectrometer and for the 10D45 should tell us what level of field
irregularities to expect.

(Note to the T474X team:  Nikolay has done excellent work on magnet design
and simulations and must-reading can be found at
http://www-zeuthen.desy.de/e_spec/Morozov.pdf
http://www-zeuthen.desy.de/e_spec/morozov_magnets.pdf

Those who have not yet should also study the proposal by the TESLA team
for a beam energy spectrometer described in the DESY print
LC-DET-2004-031.
)
     - due to cost constraints in 1971 the return yokes were made somewhat
thin (2 inches wide).  When these magnets are excited at large Bdl there
is significant leakage of flux outside the yoke.  When running way below
saturation the yokes will probably be adequate, but this needs to be
verified by simulations and measurements.  Inadequate size of return yoke
might cause variation of Bdl in the middle region.  If significant flux
leakes out it could possibly cause trouble by connecting to the other iron
in the area (rebar, lifting fixtures, and seismic bracing of the concrete
support girder and blocks) and complicate operation of the chicane to
achieve precise reproducibilty of field settings.  If necessary and if it
would make a significant improvement it would be possible to add steel to
the sides to increase the return yoke thickness.

     - there are 5 bolts on each side through the return yokes connecting
the core togeather.  The studs are specified as "mild steel".  Your
similations for the TESLA maspectrometer show that such bolts can cause
perturbations of the gap field to unacceptible levels.  There are also
other holes and voids in the core for installation of dowel pins and
lifting fixtures that might also cause field irregularities. The effects
of these bolts and holes needs to be investigated in simulations and
measurements, but there is not much we can do about them.

     - mirror plates are used to terminate the fringe field on the ends.
This is probably a good idea and your simulations will show how effective
they are.  I assume for now that we will use them.  However, the current
design has a feature that may not be good.  The plate on the end with coil
terminations is farther from the core than the one the other end.  This
will (I think) cause a small shift in the location of the magnetic center
from the location of the mechanical center.  We need to measure where the
magnetic center is in each magnet anyway, because the distance between
magnet centers enters into the energy measurement. In principle as long
as we know where the centers are, it doesn't matter if they are at the
mechanical center.  However it may be better to keep the magnets symetric.
If we continue to use mirror plates we might want to put them at the same
distance from the core.

     - solid core magnets require careful proceedures for standardization
to prevent variations of magnetic field from eddy currents.  You probably
know this story well, but I only recently understood this problem when
looking into the issues of how to standardize precision magnets to achieve
reproducible fields.  In solid core magnets the inevitable large eddy
currents cause various portions of the steel to go through different paths
in the B-H curve, such that the field in the middle of the magnet where
the NMR probe monitors would naturally be placed can be significantly
different (fractions of percent) than the field at the ends of the magnet
depending on the speed of the excitation or dexcitation (from power supply
crashes for example).  In the scheme where Bdl is monitored using only one
or a few NMR probes, such dependence on exitation speed would require
careful and consistent standardization proceedures to ensure that point
NMR measurements are precisely related to Bdl. The best data I could find
on this is in:
  J. K. Cobb, D. R. Jensen, SLAC-PUB-321.  The theoretical understanding
of this phenomena is explained well in:
  Klaus Halbach, NIM 107, 529-540 (1973).

We may find that it is difficult to obtain the desired precision with
solid core magnets.  For the ILC, if we continue with the plan of cycled
+/-B operation, it would certainly be possible to build high quality
laminated core magnets that would not have problems from eddy currents.

As you are probably well aware, there will also be time dependence of the
fields after excitation due to time dependence of iron domain
reorientation.  Magnetic after effects, as they are called, are discussed
in text books (Bozorth) and explored by workers long ago (G. Richter, Ann.
Physik 29, 605(1937)).  There is probably lots of recent data, but I didnt
find it yet. I have seen data (SLAC-PUB-137) on some SLAC magnets that
show slow settling of the magnet field with changes as large parts in
0.01% after 5 mins.  The natural settling time of magnet steel, solid or
laminated core, may make cycled operation of any magnet with cycles
shorter than 10's of minutes difficult.


Here are a few comments on other issues related to the magnets, our
planned installation, and the magnetic measurements:

  1) We plan to make the beam pipe with Al.  This completely eliminates
the issues you have studied about the small but significant magnetic
properties of various kinds of stainless steel.  We regularly make vacuum
chambers from Al and make connections to SST flanges outside the magnetic
field region using special pieces of material with SST explosion bonded to
Al.  Typically a short pipe or flange is machined from the explosion
bonded material and welded on one side to Al and to SST on the other.

  2) The stands and supports are yet to be designed, but I plan to make
them mostly with Al to keep magnetic material away from the magnets.  It
will be interesting if possible from simulations like you have done for
support girders of the TESLA magnet to see how much flux reaches out to
material in the stands and supports.

  3) The water temperature of coils and magnet iron needs to be measured
and stablized. You recommended stablization to 0.1 deg C and monitoring to
0.05 deg C for the TESLA magnets.  Presumably we need to do something
similar. Equipment and techniqes for that are not yet specified.  This
needs some work by our collaboration.

  4) High quality magnetic measurements will be required.

     (Note to the T474X team: Nikolay has a nice summary of precision
magnetic measurements in
http://www-zeuthen.desy.de/e_spec/morozov_field_meas.pdf )

Measurement techniques and capability at SLAC are described in:
 Levi et al, NIM A281: 265-276(1989), SLAC-PUB-4654.

In principle we should measure all 4 magnets precisely, but as a minimum,
the first two must be accurately known and monitored.  The magnetic
measurement campaign will be undertaken at SLAC in the fall, as I
understand it, by Michele Viti, Victor Duginov and Serge Kostromin.
This will be a big job, and is very important for the experiment.  The
main parameters, as I currently see it are:

   - SLAC technicians are prepared to make point measurements with NMR and
Hall probes, and Bdl measurements with moving wires and moving NMR and
Hall probes.  Past measurements have achieved ~0.7 10**-4 precision for
Bdl (Levi et al).  We may want to work to see if we can improve this
error. Flip coils would (probably) not be used.  I'm not sure if we will
measure with vibrating wires.  It would be useful to establish procedures
for zeroing the field, and vibrating wires seems to be a good way to
measure "zero" fields.

   - We have not yet identified a power supply, but we need to find (or
have the collaboration buy) a precision (parts in 10**5 stability) bipolar
supply with current capability to perhaps 300 A.  This would be adequate
for conditioning but would not drive the magnets to saturation.  Perhaps
we may want to hook the magnets up to a large supply one time before the
measurements to drive them all into saturation for uniformity.

    - The magnet cores and coil temperatures need to be stablized and
accurately measured during the magnetic measurement campaign.  We need to
set the specs for this and arrange the necessary equipment.

    - We plan to monitor magnets in situ with NMR probes.  So far we have
not planned to use moving wires or flip coils in situ.  If we think
either of these is essential, we need to begin now to plan for the design
(space on beamline, arrangement of pipes, wires, movers etc) and make or
buy the necessary equipment.

    - The measurement campaign would be:

    measure excitation curves in a region up to 0.15 T or so
    measure B(x,y,z) in the central region x,y = +/- 1 cm
    measure Bdl(x,y) in the central region
    measure the temperature dependence (needed for corrections)
    measure location of magnetic centers
    determine procedures (ramp speed, settling times, maximum conditioning
field etc) for magnet conditioning (running at some value B above the
maximum operating point) and standardization at the +/-B operating points.
This has to take into account operation in a cycling mode, and should be
designed so the point measurement from NMR probes gives a precise relation
to the Bdl (considering eddy currents and hysteresis effects).

This will be a lot of work.  Talking with Dieter and folks here they tell
me that the measurements for the SLC spectrometer magnet in Levi et al
required months of work, and then additional time in situ to verify and
measure stability in beamline conditions.


Finally for Nikolay, a summary list of questions for simulations of the
10D37 magnets:

   1) what is the expected field uniformity Bdl(x,y) in the central region
   2) where can NMR probes be placed
   3) should we use mirror plates, should they be symetric wrt magnet
center, and how accurately do they have to be placed
   4) what is the relative contribution of fringe field to the Bdl (sets
scale for error from Hall probes).
   5) are there significant problems caused by the relatively thin return
yokes (field irregularities, problems with excitatin of iron in the
region), and would it be worth the effort to add steel to increase the
yoke.
   6) what are the expected field irregularities from the various bolts,
holes and voids in the core. There is nothing we can do about them, but
its good to know, and may influence where we put NMR probes.
   7) what is the coefficient for variation of Bdl with temperature (sets
precison needed for temperature measurements).

   I look forward to your reply.  This is my first time to work on a
precision magnet problem, and I've got lots to learn.  I welcome any
comments or criticism from you or any of the collaborators.



       Ray


Raymond G. Arnold                        arnold@slac.stanford.edu
M.S. 42 SLAC                             voice (650) 926-2755
P.O. Box 4349                            fax   (650) 926-2407
Stanford, CA 94309