Initial proposal at C&A Meeting: http://www.slac.stanford.edu/exp/glast/ground/LATSoft/vol1/geometry/notes/GeoApril2007.pdf See Slide 7
When Joanne modeled the tiles as depicted without changing anything else Joanne realized there would be a gap along the edge where the sides come together for rows 1 and 3 so she modified the design by making the width of these rows greater. You can see some more pictures at http://www.slac.stanford.edu/exp/glast/ground/LATSoft/geometry/pictures/ACD/newSides/
From Alex, for our information, in reality the corner gaps between the adjacent sides tiles are 4mm in nominal.
Checking Changes
Ran detCheck's dumpIds application to compare the output using xmlGeoDbs v1r39p3 against v1r40 (contains changes for side tile shingling). The initial output provides a count of materials, constants and detectors - the only difference between that portion of the output is in the number of constants. It is now 771 rather than 753.
The rest of the output is volumeIdentifiers - performing a diff on the output shows that the only changes are to the side tiles and demonstrates that the build order of the side tiles has changed from bottom to top, now to top to bottom. This is confirmed by Joanne. A visual scan of the dumped Ids suggests the volIds are as they should be and that the placement and dimensions of the side tiles has changed in a sensible way. Here is the old dumpId output and here is the new dumpId output.
Next up: Modifying side ribbons to follow contours of new side tiles. Here's a picture (see slide 3). There are additional pretty pictures in that link as well, including the reworking of the top ribbons.
Updates to the top and side Ribbons
Running detCheck on xmlGeo v1r39p3 and v1r42
IdMap from v1r39p3 and v1r42