[BTeam] yet another XH/XY spreadsheet

[Michael Tiefenback]

Something is arising here that seems non-intuitive, to me at least.  The 
gap is not hideous.  The length is not hideous.  The field is not 
hideous.  But something wicked this way comes?  There should be a 
"clearly a problem" explanation, but it presently eludes me unless it is 
the pole width.  It would seem likely that we can alter the pole near 
the entrance/exit points with added material, making the magnet more 
banana-ish, breaking the symmetry, and increasing the flux through the 
return (exaggerating any saturation there).  But one could append flux 
return material, too.

Body sextupole should be addressable by pole edge contour/shim (I think 
this is so), and body gradient (while it is not what you reported as 
being a problem) by a little "H" steel.  End effects can be addressed in 
yet different ways.  Is it normal sextupole, or skew sextupole?  I have 
not yet opened the data files you provided, being at home and on a 
primitive Linux box.

Also, the turns count for XH/XY is 1:1.2 which is somewhat greater than 
the 1.010:1.160 gap ratio.  That would give me impression that the pole 
field of the XH is greater than that of the XY.  I haven't yet checked 
on the exact length of the magnets, but I was thinking they were 
similar.  Doesn't the XH bend 5.5 degrees vs a slightly lesser (few %) 
bend for the XY?  If that's due to the XY being an unnecessarily 
saturation-influenced magnet, then I guess everything you report becomes 
more reasonable.

But we also have a problem with downstream areas in the machine.  The 
sextupole can be viewed as enhancing the beam emittance, but if 
everything downstream were "co-pacetic" (I can only see Latin "pace" in 
there) then there should be no real problem.  A single nonlinear element 
can't be the whole problem.  Or so it seems to me.  The SR-driven 
emittance growth is by orders of magnitude the stronger influence than 
one or a pair of 2R magnets.

Michael

Jay Benesch wrote:
> The spreadsheet attached has Fourier components calculated from 60 
> points on 1 cm radius circles. Mesh between the pole faces is 0.25 cm 
> quadratic aka 10 nodes per tetrahedron, so node spacing is about 0.1 
> cm. For three of the sets, at 900, 1050 and 1090 MeV/linac,  the orbit 
> splits the sagitta as we normally do, with half on either side of the 
> middle of pole face.  In the fourth case, done only for 1050 
> MeV/linac, the beam enters the steel near the middle of the pole face 
> in the hope that getting it away from the corners at the ends would 
> change things. See summary worksheet for comparison of Fourier 
> components summed along the orbits.
>
> The XH and XY poles are nearly identical except for 1.01" gap in the 
> XH and 1.16" gap in the XY.  If you wish to compare them, pole drawings:
>
>  XH - RECONFIGURED CORE
> https://misportal.jlab.org/jlabDocs/docDownload/68451
>
> XY - (new) CORE
> https://misportal.jlab.org/jlabDocs/docDownload/43932
>
> XY RECONFIGURED CORE WITH POLE SHIMS
> https://misportal.jlab.org/jlabDocs/docDownload/72191
>
> I don't know how to design a shim to deal with the XY multipoles.  The 
> simplest routes I see are:
>
> a. buy new poles for the three XY magnets with 1.01" gap and a 
> separate power supply
>
> b. design/build three magnets of same cross-section and turns count as 
> the XH but 40" long vs 37.83", to get same BdL as existing XYs.
>
> And, of course, put 0.063" carbon steel beam pipes for first pass 
> through the trapezoidal BCOMs.
>
> Jay
> ------------------------------------------------------------------------
>
> _______________________________________________
> BTeam mailing list
> BTeam at jlab.org
> https://mailman.jlab.org/mailman/listinfo/bteam