This 80 cell-per-turn lattice was created using the FFAG lengthening scaling law with constant radius of curvature.  For A=1.25=100/80, lengths and angles were multiplied by A, gradients divided by A^2, dipoles constant (or alternatively quad offsets multiplied by A^2), orbit offsets multiplied by A^2.  Fringe fields lengths also had to be expanded by A^2 so I had to further add 5% to gradients to keep smallest vertical tune above zero.  The tunes cover a larger range due to longer overlapping fringes.

Beam radius 1mm
Halo radius shown an additional 5mm (6mm total from beam centres)
Elliptical vacuum chamber thickness 3mm, inner max half-height 8mm

The two pictures are at the same scale as each other and the 100 cell pictures

B_r=1.07T in the mu=1 approximation, corresponding to B_r~1.12T because real mu=1.05, Shin-Etsu grade N36SH for example (H_cJ=21000 Oe, i.e. reverse flux of up to 2.1 Tesla without demagnetising)

Comparisons in brackets are shown relative to 100 cell lattice values

QF (quad)
goal_quad = -28.58688 T/m (-33%)
GFR = 29mm (+53%)
Inner radius of pieces = 38mm (+36%) 
Pole tip field 1.19T --> 1.09T (-9%)

BD (combined function)
goal_dipole = 0.502760009 T (+3%)
goal_quad = 18.475296 T/m (-33%)
GFR = 18.5mm (+48%)
Inner radius of pieces = 27.5mm (+28%)
Pole tip field 1.08T --> 1.01T (-6%)

Pole tip field has improved because the 9mm margin between closed orbit centre and pole tip has not been increased, while everything else has been geometrically scaled up with the same fields on beams.