To: Distribution 22 Jan 96
From: Martin Nordby
Subject: Minutes of Near IR Engineering Meeting of 19 Jan 96
Q1 Magnet Status Report
Much of the recent work on the Q1 design has focussed on providing sufficient
magnetic shimming to cancel out the n=3-6 harmonics expected. The design
presented at the Conceptual Design Review in December had an outer shim
ring, using cylindrical shims which could be rotated. However, these shims
were at such a large radius that they could not tune out expected harmonics in
the 1E-3 range above n=3.
The next attempt put the shim ring between the quad and dipole rings. This
improved the ability of the shims to tune out higher harmonics, but the shims still
did not have the needed strength to remove all harmonics in the n=3-6 range.
Recently, this mid-ring was thickened to maximize the shim strength. The shim
cylinders were doubled in volume (from 0.3125" to 0.442" diameter), and the
dipole blocks pushed outward at the expense of thinning the outer ring.
This design is clearly superior magnetically, but makes the assembly and
support of Q1 more difficult. The main concern is that the brittle P.M. material
should not bear any structural load of the magnet. Thus, all load paths must
avoid the P.M. material. Another option being investigated is to lump the tuning
in 2 or 3 discrete tuning rings, such as being planned for Q2 (see below).
Andy has also being moving forward on developing the Technical Specification
for procurement of the P.M. prototype material. Initial feedback from vendors is
that the tolerance on magnet strength of +/- 1% is very tight, and they would
prefer doubling that. Also, the elevated-temperature coercivity of 12,000 kOe at
100 degC is also tough to meet. Andy is working with vendors to understand
the reasons for this, and seeing if--and how much--these values can be relaxed.
Shimming Q1
Stan Ecklund and Andy have been working to develop a shimming plan for Q1.
Stan found that, using the expected error harmonics of magnets with "real"
fabrication tolerances, the old mid-ring shims were 50% of the strength needed
to tune out the n=3-6 harmonics. However, a thicker ring with 0.44" diameter
shims (32) is sufficient. The design incorporating these shims still produces a
magnet with 2% stronger field than needed.
As a departure from the approach of tuning each slice individually, Stan looked
at using discrete shim rings along the length of the magnet (possibly 2 or 3).
This had previously been out-of-favor because of the anticipated effect on the
LEB. However, optics analysis has since shown that Q1 can be locally tuned,
without adversely affecting the beam.
With no shims in a slice, it could be made 11% stronger than needed, allowing
3 slices along the length (3/24-ths of the magnet) to be replaced with correction
rings. Because the errors which produce the harmonics are (assumed to be )
random, they average over the magnet, so tuning the magnet as a whole
actually requires less strength than for individual slices. For a 5 cm long
correction ring, 32 15-cm diameter rings at a mean radius of 97 mm could tune
out all harmonics up to n=16. Stan and Andy are continuing to look into this
option, and expect to decide on a direction for the prototype next week.
Effect of External Fields on Q1 Harmonics
Stan also investigated the effects of the dipole and radial solenoid field on
harmonics of the quad field in Q1. He divided each trapezoidal block into 40
sub-blocks, calculated the quad field, then modified each sub-block strength as
a function of applied field on the sub-block, and iterated. The mean field on a
block is 3 kG, but the maximum produces harmonics on the order of 5 E-3 of the
main quad field (at 74 mm inside radius of quad).
Presumably, the effect of fields from neighboring quads and the dipole ring will
be tuned out during production, but the solenoid field can not. The radial
solenoid field produces a 5 E-3 octupole harmonic, which is 100 times less with
just the dipole field. However, since the raidial solenoid field only affects a few
slices of the backward Q1, the total magnitude of the harmonic is not large.
Q2 Update
Dave Humphries reported on progress in the design of the Q2 shielding for the
HEB. Past analysis showed that with a 1 mm thick vanadium permendur shield
around Q2, the stray field at the HEB centerline is still 50 G, which is 50 times
the desired number. However, by extending the shield beyond the end of the
P.M. material by 2 cm, the stray field is cut by 50%.
Also, Dave looked at increasing the segmentation of the Halbach quad from 16
blocks to 24. This decreased the stray field to only 2-5 G for the un-shielded
case. Adding a shield should bring the field down to the target value of 1G.
The 24-block segmentation also produces a 3% stronger magnet. This is
needed to increase the safety margin of the magnet Finally, it eliminates the
intrinsic n=18 harmonic of the 16-block design. Because of these advantages,
this is the new working design for Q2.
Dave also worked up a rough design for Q2, assuming the beam asymmetry
changed from 9-on-3.1 GeV to 9.175-on-3.05 GeV. This eliminates the dipole
blocks from Q2, and allows the center of the quad to move in-board. Also, since
the beam separation is greater, 1 mm of extra space is available for HEB
shielding in the septum region.
Q2 Analysis
Mike Sullivan investigated the impact of the Q2 design with changed energy
asymmetry. The design has a center which is only 4.4 cm from the "original",
down from the 7.85 cm of Dave's working design (version 2.1). This allows the
quad field to be weakened an additional 2.5%, and marginally increases the
BSC through Q2 by 0.7 mm.
The 24-block Halbach array produces quad field harmonics in the 2 E-3 range
for n=2-5. This is better than that expected with the 16-block configuration,
using tight assembly tolerances, and a strength tolerance of +/- 2%. Of the
many errors/tolerances Mike can introduce into his assembly program, he sees
that the magnetization orientation and strength errors produce the largest
magnet errors. He is now trying to understand the sensitivity of field quality to
the other errors, so we can hold tight tolerances where needed, but try to loosen
up any tolerances which do not significantly impact field quality.
These minutes, and agenda for future meetings, are available on the Web at:
http://www.slac.stanford.edu/accel/pepii/near-ir/home.html