Being kinda slow the last month or so I've spent my days running experiments with sanitary methods, and according the the forty something dollar pamphlet AWS calls D18.1:1999 the joint is prepared with no more bevel than 10% of wall thickness, so really it's a square face butted up tight. I've had great success with pulse and a positioner. No filler. Trick is to watch so carefully that the cap isn't above flush. If it is then there's LOP. If I try this in a fixed position I can't do it. It's the devil to follow the tiny crack, for one thing, and the pulse just gets in the way because you need to vary the heat as the joint gets hot. Usually I get undercut or concavity on the outside if I run hot enough to get a root. I've taken some 3/4 samples, shined 'em up and wear as a ring. People can't believe it's done on one pass and looks like friction welding. Of course, they've never seen orbital, either.