S beams and C channels have a flange slope of 16-2/3% which is about 9.462 degrees. The flange thicknesses on these shapes is always listed as an average flange thickness.
I have an old Bethlehem Steel Structural Shapes Handbook from 1978 that I have scanned to PDF. It contains approximate detailing dimensions (commonly used by structural steel detailers) and theoretical dimensions used for design work. It provides gage and grip dimensions for the flanges as well as "I" values in X & Y directions along with the flange to web radii. Unfortunately it does not list the flange toe radii that you find on most rolled sections, including angles. I expect those are listed in the ASTM standards that the shapes are rolled to.
Structural steel detailers typically don’t really care about the exact shape of these sections because of the methods used to join them in that type of work. They are only concerned with the straight section of the web and the gage and grip dimensions of the flanges for bolting.
I often need to cope the ends of channels to closely match up with the inside profile of an intersecting channel for welding, so I always model these sections as accurately as possible. I have attached a model of a C6 x 8.2 channel illustrating how I model a channel using the published dimensional data.
Using this method my section areas and I values from SE match very closely with the listing in the Bethlehem Steel Handbook so I figure I am close. Additionally, my joints always seem to work out within normal manufacturing tolerances.
detailing is not the reason to get the channels exactly right (S shape very un-common these days). It's all about the weight. What I do now is edit the shape until I get the exact same weight per foot as is listed in steel books (err the spreadsheet).
good fabrication drawings include the weight of everything. so production can plan out each move of each part.
I too, had to be concerned with the actual design dimensions and not the drafting dimensions for steel shapes. Way back in my NX days I generated a set of part families using the AISC dimensions. If anyone is interested I might be able to pull the dimensions and spreadsheet data into seperate files.