Linear non-scaling FFAG
Let us consider the operation of an AG synchrotron without ramping magnets again. Remember there
were two problems if we did this. One was that a beam would hit the wall due to the dispersion
function and the other was that the focusing force would effectively decrease. In a scaling FFAG, we
introduced the radial field profile so that the tune was constant for the entire momentum range during
acceleration. For the orbit shift, we simply widen the aperture in the horizontal direction.
The orbit excursion in a scaling FFAG is smaller than in a cyclotron, but still not negligible, for
example around 0.7 m compared with a 5 m radius for a few hundred megaelectronvolts [10]. This is
because of the upper limit of the field index k. To squeeze the orbit shift during acceleration, the field
index k should be as large as possible. On the other hand, it risks losing the stability inherent in AG
focusing because too large a k leads to over-focusing. Also one may notice that most of the orbit shift
happens in the lower momentum region where the field gradient is relatively small.
There is another way of designing a FFAG accelerator, which reduces the orbit shift as much as
possible, without paying much attention to the tune excursion during acceleration. This is effectively a
synchrotron lattice with as small a dispersion function as possible without chromaticity correction. If
the dispersion function is small enough, the orbit shift caused by momentum changes can be
! = " # tan$ % ln
r
r0
accommodated in a reasonably sized vacuum chamber. If we eliminate multipoles higher than
quadrupole in Eq. (9), the dynamic aperture is expected to be large as well. We choose a quadrupole
field gradient that gives a phase advance per focusing unit below 180° at the injection momentum and
let it decrease when the beam is accelerated. This is called a linear non-scaling FFAG