Readout segmentation for multiple cylinders with offset axes

[henryg7vdj]

Could you explain how I can set up readout segmentation with CylindricalGridPhiZ for multiple cylindrical surfaces with displaced axes?

I am writing a geometry for the epic outer barrel detector, following this example: https://github.com/eic/epic/blob/main/src/MPGDCylinderBarrelTracker_geo.cpp. In this case, and the others I have seen, all the cylinders have the same axis (along the beamline). However, the outer barrel that I what to model is made from up to 70 staves, each with modules with a curved surface, where the radius of curvature is significantly less than the radius of the barrel. The axis for each cylinder will be parallel to the beamline, but offset from it.

I tried to do this (my code is at https://github.com/eic/epic/tree/curved_OB), but the reconstruction program does not see the hits. I am guessing the problem is with the readout segmentation.

Thanks

[MarkusFrankATcernch]

Hi,
we discussed this issue during the meeting. The radius gets into the game when converting back the CellID to a world-position. Would it be possible to parametrize the code in such a way that the radius of the sensitive element is identical to the
barrel radius? Just to conform that the source of the problem is in CylindricalGridPhiZ.cpp line 80

[henryg7vdj]

Thanks for looking into this. It sounds like we can't do the readout for our sub-detector using CylindricalGridPhiZ and we need to find another way - probably approximating the curve as a series of flat strips.

[MarkusFrankATcernch]

Well, there is always the possibility to either

• extend the functionality of CylindricalGridPhiZ or
• create a new "CylindricalGridPhiZ" type segmentation

which satisfies your needs.
If you know how to compute the transformation from local coordinates to world coordinates using the local coordinate frame of your sensitive volume, this should be do-able without too many head aches. As far as I could see the barrel radius only enters in this computation.

[MarkusFrankATcernch]

@henryg7vdj How would your segmentation look like:

Would it be equi-distant in dPhi or in dPhi* ?
Same: equi-distant in dZ or dZ* ?
"*" = on the surface of the sensitive volume.
(not that I have an idea yet how this transformation can be resolved analytically)

[henryg7vdj]

Equi-distant in dPhi*; dZ and dZ* are the same, aren't they? as the axes are parallel.

This looks like an interesting project, but I'm pushed for time right now. This week I am focussed on getting a simpler model with flat staves working. I will come back to this later.

[MarkusFrankATcernch]

No dZ and dZ* are only the same for theta= pi/2 (90 degrees).
Otherwise, unless you are at the edge of the sensitive element abs(dZ*) > abs(dz).