i once had a decent and fairly long email exchange with dazza on this and here is my short answer based on that experience -
if i start with a rectified shell, make a frame (or main triangle...), and am pleased with the results (read: it passes my smell test), that's that.

when i add the rear assembly, clean it all up, chase threads, replace it on a table - and then see that all of the sudden (as an example) the seat tube is now leaning to/from the table a C hair more than before, i do in fact measure the shell and notice a slight change (shrinkage, warp, whatever...) but rather than use a facing tool i take some A/O cloth on the flat side of an 8" half-round file, and i finesse the one area of the shell face that needs a shave. it's usually 2-3 soft strokes and the frame, back on the whipping post, again shows the same reading that it did when it was a main triangle without more parts added. i view it as the frame staying the same but the reading surface varying from time to time, and my fix is planing one part of it by hand rather than using a cutting tool.

Quote Originally Posted by Dave Kirk View Post
Cool.

My experience with shells is that they are faced very well from the casting house and are fine without facing to start but it's after they have been heated asymmetrically (all the tubes being on one side) that the top side shrinks and the faces go wonky.

Have you measured the width/parallelness of the shell after brazing? I would expect that you would find that the top side (where most of the heat is) would get narrow. I usually see the top being about .005" narrower that the bottom and when projecting this angle over the length of the seat tube you end up with the top of the seat tube being low which would result in the top of the seat tube being about 2 mm low on the plate. This is one reason I face post brazing. The other is that I want the faces that the BB bearings snug up to to be parallel so that the bearings will have the easiest life possible.


dave