Performance of the 1-meter models of the 70 mm aperture quadrupole for the LHC low-beta insertions

Abstract

Following the successful testing of the first 1-metre model of the 70 mm aperture quadrupole for the LHC low-beta insertions, two further 1-metre magnets have been built. All magnets feature a four-layer coil wound from two 8.2 mm wide graded NbTi cables and a four-way split yoke supporting structure. In this paper we review the training history of the three magnets performed at 4.3 K and 1.9 K in several tests. All magnets surpassed the operating gradient required for the LHC, with the third magnet reaching 260 T/m, its short-sample gradient at 1.9 K. The peak temperatures in the superconductor at various operating conditions are reported and a summary of magnetic field measurements is given.

Full text

Page 1

EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH
European Laboratory for Particle Physics
Large Hadron Collider Project
LHC Project Report 240
Following the successful testing of the first 1-metre model of the 70 mm aperture quadrupole for
the LHC low-β insertions, two further 1-metre magnets have been built. All magnets feature a four-
layer coil wound from two 8.2 mm wide graded NbTi cables and a four-way split yoke supporting
structure. In this paper we review the training history of the three magnets performed at 4.3 K and
1.9 K in several tests. All magnets surpassed the operating gradient required for the LHC, with the
third magnet reaching 260 T/m, its short-sample gradient at 1.9 K. The peak temperatures in the
superconductor at various operating conditions are reported and a summary of magnetic field
measurements is given.
* CERN, LHC Division,
1 FNAL, Batavia, USA, ** Oxford Instruments Ltd, UK
Presented at EPAC’98, 22-26 June 1998, Stockholm, Sweden
Administrative Secretariat
LHC Division
CERN
CH - 1211 Geneva 23
Switzerland
Geneva, 28 September 1998
Performance of the 1-metre Models of the 70 mm Aperture Quadrupole
for the LHC Low-β β Insertions
*L. Bottura, Ch. Giloux, G.A. Kirby, M.J. Lamm
1, R.Ostojic, F. Rival, F. Rodriguez-Mateos,
S. Schloss, A. Siemko, T.M. Taylor
**S.R. Milward, K.D. Smith, M.C. Townsend, J.R. Treadgold, J.M. Wiatrzyk
Abstract

Page 2

2
Performance of the 1-metre Models of the 70 mm Aperture Quadrupole for
the LHC Low-β β Insertions
L. Bottura, Ch. Giloux, G.A. Kirby, M.J. Lamm
1, R.Ostojic, F. Rival, F. Rodriguez-Mateos,
S. Schloss, A. Siemko, T.M. Taylor
CERN, Geneva, Switzerland,
S.R. Milward, K.D. Smith, M.C. Townsend, J.R. Treadgold, J.M. Wiatrzyk,
Oxford Instruments Ltd, UK

1 Permanent address: FNAL, Batavia, USA
Abstract
Following the successful testing of the first 1-metre
model of the 70 mm aperture quadrupole for the LHC
low-β insertions, two further 1-metre magnets have been
built. All magnets feature a four-layer coil wound from
two 8.2 mm wide graded NbTi cables and a four-way
split yoke supporting structure. In this paper we review
the training history of the three magnets performed at 4.3
K and 1.9 K in several tests. All magnets surpassed the
operating gradient required for the LHC, with the third
magnet reaching 260 T/m, its short-sample gradient at 1.9
K. The peak temperatures in the superconductor at
various operating conditions are reported and a summary
of magnetic field measurements is given.
1 INTRODUCTION
The design of the LHC low-β triplets is based on a
70 mm aperture high gradient quadrupole operating at 1.9
K with a design gradient of 240 T/m [1]. Besides
accommodating fully separated beams at injection, the
quadrupoles must provide a high field gradient and low
multipole errors required for the 7 TeV colliding beams,
and sustain high heat load from the secondary particles.
During the past 5 years, CERN in collaboration with
Oxford Instruments, has designed, built and tested three
1-metre long single aperture quadrupoles fulfilling these
requirements. The quadrupole coils were also designed to
function in a twin aperture configuration, with a nominal
operating gradient of 160 T/m at 4.5 K [2]. A number of
such magnets will be required in the LHC insertions, in
all cases where the 56 mm nominal aperture of the
insertion quadrupoles may be a limiting factor.
In this paper we summarise the performance of the
three quadrupoles and present the training history, the
main results of the quench protection studies and of the
magnetic field measurements.
2 MAGNET DESIGN
The design and construction of the first low-β
quadrupole model (Q1) for the LHC has been described
in [3]. To summarise, the magnet has a four layer coil
wound with two 8.2 mm wide graded NbTi cables, with
the transition between the cables in the middle of the
second innermost layer. The coil pre-stress is
accomplished with aluminium force rings, which are
placed around the four-way split yoke structure, which in
turn presses against 7 mm wide stainless steel collars.
As a result of the tests of Q1 [4], in which the magnet
was found to have a peak temperature close to 450 K, the
copper-to-superconductor ratio for the next two magnets
(Q2 and Q3) was modified from 1.3 to 1.2 for the inner,
and from 1.3 to 1.7 for the outer cable. The other
parameters of the cables are identical. As a consequence
of these modifications the peak field in the coil is shifted
from the inner cable (second layer) to the outer cable
(third layer), with a slight reduction of the short sample
gradient, from 265 to 260 T/m. On the other hand, the
peak temperature in the coil is expected to decrease to
around 300 K. In all other respects, Q2 and Q3 are
identical to Q1, with some minor modifications in the
assembly procedure.
The major difference between Q2 and Q3 is in the
scope of instrumentation. While Q2 is instrumented with
the basic set of spot heaters for magnet protection
(located on the pole turns of layers 1 and 3 and in the
middle of the outer conductor in layer 2), Q3 has also a
full set of accompanying voltage taps for quench
propagation studies. In addition, it is equipped with strip
heaters located in two radial positions: one set is located
between layers 2 and 3, and the other on the outside of
layer 4. In both positions, there are a total of four strip
heaters, each one covering approximately 1/2 of each of
the two adjacent coil octants. This system is proposed for
the protection of the full length quadrupoles.

Page 3

3
3 TRAINING HISTORY
The training of the magnets was performed at Oxford
Instruments. Q1 was tested on two occasions in 1996 [4],
while Q2 and Q3 were tested in June and November
1997. Finally, an extensive program of power tests and
magnetic measurements was performed on Q3 in CERN
in May and June 1998.
120
140
160
180
200
220
240
260
05101520 253035404550
Field Gradient at Quench (T/m)

4.3 K1.9 K
4.3 K
1.8K1.8K4.3 K
Test 1
Test 2
Test 3
Quench Number
Figure 1: Training history of the three 70 mm single
aperture quadrupoles: Q1 (diamonds), Q2 (triangles), Q3
(squares). Quenches were performed at the nominal ramp
rate of 5 A/s. Starting from quench no. 4 of test 3, the
fraction of energy dissipated in Q3 was increased to
100%. Horizontal dashed lines represent nominal
operating gradients in the LHC: 160 T/m at 4.5 K and
220 T/m at 1.9 K. The short sample gradients are 200
T/m at 4.3 K and 260 T/m at 1.9 K.
The training history of the three magnets is
summarised in Figure 1. On the first 4.3 K training test,
Q1 reached a plateau of 193 T/m (3980 A), which
corresponds to 98% of the short sample limit. The magnet
was subsequently tested at 1.9 K, where it reached 238
T/m (4850 A). After a long thermal cycle, the magnet
was retested and the previous quench levels were reached
after one quench both at 4.3 K and 1.9 K.
Q2 and Q3 were first trained to 3750 A, required for
the 4.3 K magnetic measurements. In both cases this was
achieved in 7-8 quenches. The somewhat erratic training
of Q2 both at 4.3 K and 1.9 K was traced to inadequate
shimming of the coil, which was later corrected in Q3. At
1.9 K, Q3 trained well, with a monotonic increase in the
quench current, reaching 251 T/m, 95% of the short
sample gradient.
In the second test performed at CERN, Q3 was first
trained at 4.3 K reaching the short sample gradient of 200
T/m in four training quenches, a significant improvement
over the previous test. At 1.9 K, the magnet started
training at the same level as in the preceding thermal
cycle, but reached the previously achieved gradient of
250 T/m in only three quenches. In the subsequent
quenches it went up to 260 T/m, the short sample gradient
at 1.9 K. After a thermal cycle, where the magnet was
removed from the cryostat to repair faulty
instrumentation wires, it was cooled to 1.9 K and
continued training at the same level, retaining its trained
state. In the following quenches, the fraction of energy
dissipated into the magnet was increased from about 50%
to the full energy deposition. As the fraction of energy
dissipation increased, fluctuations in the quench field of
2-3% were observed, with subsequent regaining of the
short sample gradient. This behaviour may be related to
the fact that apart from the few initial training quenches,
which occurred in layers one and three, all quenches
close to short sample were traced to the transition region
between the inner and outer cables in layer two of pole C.
Finally, at the end of the training sequence, the magnet
was quenched twice at 4.3 K to confirm its quench
plateau.
4 QUENCH PROTECTION STUDIES
One of the important tests for Q3 was to determine the
efficiency of the strip heater protection scheme at 4.3 K
and 1.9 K, and to establish experimentally the effect of
the modification of the copper-to-superconductor ratio of
the cables.
A summary of peak temperatures achieved in Q1 and
Q3 at 1.9 K and 4.3 K is shown in Fig. 2. In all cases, the
quenches were induced by firing spot heaters (in layer 2
for Q1, and in layer 1 or layer 3 for Q3). In case of Q1,
the power supply was turned off upon quench detection,
the terminals effectively shorted and the remaining spot
heaters fired. For Q3, the same power supply sequence
was followed, but instead of the spot heaters, the strip
heaters were fired. In both cases over 99% of the energy
was dissipated into the magnet.
The voltage across the cable segment adjacent to the
heater was monitored throughout the current decay. After
accounting for magneto-resistance and conductor
geometry, the copper resistivity was determined. The
peak temperature then followed from its peak value.
0
50
100
150
200
250
300
350
400
450
500
01000 20003000400050006000
Current (A)
Temperature (K)
4.3 K Q1
1.9K Q1
1.9K Q3
4.3K Q3
Figure 2: Peak temperatures in Q1 and Q3 at 4.3 K and
1.9 K. The overall error in peak temperature is ? 15 K.
As shown, Q3 exhibited a much improved behaviour
over Q1, as the peak temperature decreased from 450 K
to 250 K at 4400 A (220 T/m). At 4.3 K, the peak

Page 4

4
temperatures in Q3 in the case of firing layer four strip
heaters only, shown in Fig.2, were only slightly higher
than those when all strip heaters were used. For both 4.3
K and 1.9 K the Q3 peak temperature did not depend
whether the quench was initiated in the inner or outer
cable. Furthermore, it increased monotonically with
current, indicating that the magnet protection was
dominated by the strip heater-induced resistance and not
from quench propagation. The effectiveness of the
heaters is due to their wide area of coverage and short
time delay between firing and onset of resistive voltage
growth (20 ms at 4400 amps). A complete study of the
quench velocities and heater delay times will be presented
in a forthcoming report.
5 MAGNETIC FIELD MEASUREMENTS
Magnetic field measurements were performed using cold
radial measurement coils designed at CERN. For Q3, the
measurements were performed using coils sensitive to
both the body and end fields, while for Q1 they were
done with a coil set sensitive to the body field only. For
Q2, the measurement coils did not function properly and
the field was not measured.
The multipole errors in the body of Q1 and Q3 at 220
T/m are summarised in Table 2. In Q1, a 10% thicker
insulation system was used in one of the poles [4], and
the quadrupole symmetry of the magnet was partially
lost, as seen by the expected non-zero multipoles of order
n<6. In general, the difference between the measured and
expected multipoles may be attributed to the errors in
positioning of the coil blocks. For Q1, it was found that
the displacements that reproduce the residual errors obey
a Gaussian distribution with rms of 0.02 mm and 0.04
mm for the radial and azimuthal directions. When
compared to the random multipole errors based on these
rms displacements, the measured non-allowed multipoles,
except for b5, are within 2σ.
Table 2 Relative multipole errors of Q1 and Q3 at 220
T/m, at reference radius of 17 mm (in units of 10
expected random errors from coil construction are given
in the last column.
-4). The
Q1
measured
-0.472
0.836
-1.136
1.728
3.179
0.349
-3.867
-0.234
-0.418
-0.070
Q1
expected
0.214
0.214
0.
0.569
-0.200
0.200
-3.458
Q3 measured
2σ
b3
a3
b4
a4
b5
a5
b6
a6
b10
a10
-3.334
-1.589
0.439
0.326
-0.290
-2.447
-4.543
0.384
-0.349
0.007
2.90
2.90
1.04
1.04
0.74
0.74
0.
-0.418
0.
In Q3, only the allowed b6 and b10 multipoles are
expected. The measured values of -4.54 and -0.35 units
are in excellent agreement with the predictions of -4.30
and -0.35 units. Except for the sextupole, the non-allowed
multipoles are in general smaller than for Q1. Similarly to
b5 in Q1, the a5 multipole in Q3 is outside the 2σ interval.
Further investigation is needed to determine whether this
is related to a systematic feature in coil construction.
The multipoles show weak dependence on current: the
saturation contributes 0.12 units to b6 and 0.035 units to
b10. The width of the b6 hysteresis curve due to persistent
currents at 4.4 K is 4.68 units at 250 A.
The measurements were repeated before and after the
training and thermal cycles. None of these changes of
state seems to have an influence on the field quality of the
magnet. For example, the b6 multipole is reproduced at
the level of 10
measurements, including the bath temperature and power
supply control, were at the level of 2 10
-3 units, while the repeatability of the
-4 units.
3 CONCLUSIONS
As part of the LHC magnet development program,
three 1-metre long 70 mm single aperture quadrupoles
were built and extensively tested. All magnets surpassed
the operating gradient required for the LHC of 220 T/m,
with the third magnet reaching 260 T/m, its short-sample
gradient at 1.9 K. Peak conductor temperatures have been
measured, confirming that the magnets safely absorb their
own energy. The measured field harmonics are in good
correspondence with the design values, confirming the
coil construction technique.
Following the tests in single aperture configuration, Q2
and Q3 quadrupoles will be disassembled and the coils
used for building a two-in-one wide aperture quadrupole,
to be tested by the end of 1998.
ACKNOWLEDGMENTS
The authors wish to thank the technicians responsible
for the magnet assembly for their excellent work, and the
LHC test team for the power test and magnetic field
measurements.
REFERENCES
[1] R. Ostojic, T.M. Taylor and S. Weisz, “Systems
Layout of the Low-β Insertions for the LHC
Experiments”, PAC’97, Vancouver, May 1997, LHC
Project Report 129.
[2] G.A. Kirby, R. Ostojic, T.M. Taylor, G. Trinquart,
“Design of the 70 mm Aperture Superconducting
Quadrupole for the LHC Dump Insertions”, IEEE
Trans. App. Superconductivity, 7 (1997), No. 2,
pp.562-565.
[3] R. Ostojic, T.M. Taylor and G.A. Kirby, “Design and
Construction of a One-Metre Model of the 70 mm
Aperture Quadrupole
Insertions”, IEEE Trans. Magn., 30 (1994), pp. 1750-
1753.
[4] R. Ostojic et al., “Quench Performance and Field
Quality Measurements of the First LHC Low-β
Quadrupole Model”, PAC’97, Vancouver, May
1997, LHC Project Report 130.
for the LHC Low-β