ATTENDEES: G. Biallas, J. Bisognano, C. Bohn, D. Douglas, L. Harwood

Decision:

The preferred method of controlling the magnets on the Reverse Bend & Pi Bend circuit is Case 3 below. It requires two shunts on the circuit, each controlling one set of four Reverse Bend Dipoles at each end of the machine.

Shunt Properties for 1 µ upgraded machine:

1. 50 V.

2. 4 A (D. Douglas would prefer 6 A for +- 1 -1/2 % adjustment, if available)

3. Regulation: 1 part in 10⁴.

The shunt for the four chicane bends will be the same and has lesser requirements.

Discussion:

Case 1. The initial concept of individual shunts on the Pi Bends requires shunts working at regulation of 1 part in 10⁵ and 60 V. and entails a new shunt design . This concept was rejected because of the perceived lower ripple driven beam motion and ease of accomplishment of the Case 3 and Assumption 1 below.

Case 2. Apply one shunt to gang all the reverse bend dipoles. From an elementary calculation, ripple driven beam motion from a system shunting the Reverse Bends was found, in the back leg, to be 3 times less than Case 1. Less motion makes the Coherent Synchrotron Radiation investigation easier. In addition, when the dipoles are ganged, ripple driven motion is suppressed below background at the cryomodules upon recirculation by virtue of the 1/2 wavelength Betatron phase advance across the back leg. However, the voltage requirements of this case are 100 V, requiring a new shunt design, non standard components and expensive cabling.

Case 3. Split Case 2 into two shunts, each in parallel with the four reverse bends at each end of the machine. This case reduces the shunt requirement to a replacement or upgrade of components in our existing design. It also saves costly cable runs. As noted above, ripple driven motion in the back leg is three times less than Case 1. However, when the two ends are split, the ripple driven motion at the cryomodules may be as much as doubled that in the back leg. This fault was judged acceptable.

Assumptions:

1. The two Pi Bends will be matched to each other at all levels of excitation to better than 1 part 10³ by the process of shimming or grinding the back legs.

2. The family of the Reverse Bends will match to the Pi Bends by better than 1 part in 10² by similar processes.

Action Items:

D. Douglas will verify and publish the calculation of ripple motion contrasting Case 1 and 2.

G. Biallas will summarize the meeting and coordinate it with L. Harwood, D. Douglas and W. Merz & E. Martin.