Different results comparing on-trajectory and off-trajectory chicane tracking

[hbiss]

Dear all,

I have some issues when comparing transport matrices where the electron beam is on trajectory and off-trajectory midway. I wrote a small minimal example with to different approaches for demonstrating my problem (I attached both cases in a zip archive):
1. Case: Chicane of 4 rectangular dipoles with |angle|=100mrad and fse=0. The centroids are as expected zero and the R21 term in the transport matrix is as expected zero.
2. Case: Chicane of 4 rectangular dipoles with |angle|=99mrad and fse=0.01010.... The effective angles are thus (1+fse)*99mrad=100mrad. As expected the centroids are not zero in the chicane. But now I am bit confused: The final centroid is not zero (80nm) and there is non-vanishing focusing in the horizontal plane (R21 not zero). (There are also significant differences in R51,R52 and R56). Although this seems to have no effect, I enabled to make off-trajectory tracking reasonable the second order matrices in the run files and split the dipoles in order to incorporate the combining effects of nonzero centroids and 2nd order matrices onto the effective first order matrices.

I would expect the transport matrices from both cases to be equal as the effective trajectory in floor coordinates should be equal and the reference trajectories (should) overlap at the end of section again. I am wondering now if I forgot something to take into consideration? Any help is appreciated

Thanks,
Hauke

[michael_borland]

Hauke,

I think the reason for this fairly small discrepancy is that the dipoles are implemented as sector (wedge) bends with impulse kicks for the edge effects. This is only good to second order, so there is a small error in the trajectory inside and at the exit of the chicane.

An improved result can be obtained using the CSBEND element and K. Hwang's edge model. See attached.

--Michael

[hbiss]

Hi Michael,

thanks for your improvement! It helped me a lot and solved my problem. Beside the more advanced edge model I was also not aware of the major command analyze_map for calculating the map.
When I added drifts I had to enable there as expected the second order matrices for the off-axis tracking but still then everything is OK (negligble differences between both approaches).

Thanks,
Hauke