Inconsistent sign convention of dipole etilt and quadrupole tilt

[duanz]

Hi Michael,

In the attached example, I tried to add an ETILT error of 0.01 rad to a CSBEND, and a TILT error to a KQUAD. I assume elegant follows the same local reference coordinate system as in MAD. Then tracking results seem to indicate a positive ETILT rotates a CSBEND in the counter-clockwise direction around the s-axis (particle orbit direction), while a positive TILT rotates a KQUAD in the clockwise direction around the s-axis (particle orbit direction). Which appears not self-consistent. Would you please help clarify this issue? Thanks!

Best,


Zhe