Hello, everyone, 
I am trying to implement a spectrometer in the EUV spectral range in the Harada configuration, i.e. a variable-line space spherical grating.
To do this, I am using the 'Elliptical Grating' surface; however, when I start the optimiser I get that 'beta' is of the opposite sign and rescaled by exactly a factor of 3 compared to what I calculated using the formula:
beta = 3*b3/(R^2 * T). The formula was taken from the article “Design of spherical varied line—space gratings for a high-resolution EUV spectrometer”.
Zemax, on the other hand, provides the value via this equation: beta = -b3/(R^2 * T).

I am using the following parameters:
1. working wavelength: 105 nm
2. Incidence angle: 8.985 degree
2. system at unit magnification
3. input arm: 406.23 mm
4. Grating radius of curvature: -402.8 mm, so grating semi-axes: 2.4826e-3
5. Grating frequency: 2400 lines/mm.
6. Ruling parameter: b2 = 0.066, b3 = 0.0087. So, alpha = -1.3654e-4 and beta = 2.2323e-8.

Can you help me understand this aspect?
Thanks
Andrea