Dispersion in PARTRAN

[emo]

Dear Didier,

I am experiencing issues with the dispersion function since it was introduced in PARTRAN mode.
The problem is related to the different estimations of D(s) in envelope and PARTRAN modes when high beam currents are involved.
Here is an example for a double achromat tuning in envelope mode with a minimal dispersion after the second dipole.

Envelopes and dispersion_env mode.png (46.04 KiB) Viewed 1255 times

And here is the plot from PARTRAN using exactly the same tuning.

Envelopes and dispersion_PARTRAN mode.png (59.1 KiB) Viewed 1255 times

As you can see, in PARTRAN, there is a large increase of the dispersion after the second dipole (see black arrow).
But in my opinion this should not be real because the beam energy spread would induce an increase of the horizontal envelope, but this is not observed in the plot.

To explain better I added a second beam with a higher initial energy.

Envelopes and dispersion_PARTRAN mode_E+1%.png (61.94 KiB) Viewed 1255 times

If the dispersion was realistic then it should cause an offset of the centroid at 11.6 m, similar to the one seen in the deviation section between the two dipoles (see green arrows).
But the observed centroid shift at this position is many times smaller than expected (around 40 times).

This issue with the dispersion is seen only when the beam current is very high and space charge is involved.
Could it be that the space charge contribution to the dispersion is overestimated in PARTRAN?

Thank you in advance for your assistance.

Best regards,
Emil

[Didier]

Dear Emil,

As you can read in the manual, the Partran dispersion is calculated using second-order moments and is therefore probably highly dependent on a few particles away from the beam core, making this value irrelevant.
But it is clear that using the envelope dispersion to set your achromatic line is sufficient, because increasing the input energy has no impact on the beam position in the envelope and partan mode.

It might be interesting to plot the emittance [Y-dp/p] to check my hypothesis
Regards,

Didier

[emo]

Dear Didier,

Thank you for your answer and the explanation.
For now, I will continue using the envelope dispersion for deviation sections.

In the coming days I will try your suggestion for the emittance plot and I will write back after.

Best regards,
Emil