Coming right along with the tool post build , and today I started cutting the dovetails .The ones on the post are a couple of thousandths different , but I don't think that's going to be a problem . I have one tool holder cut (and 2 more slotted) and it's an easy sliding fit on both of the post dovetails . I'm wondering though just how much clearance I should allow . I know they don't need to be sloppy loose , but how much clearance is enough ? And where to measure ? A thousandth between the dovetail faces is a whole lot different than the same between the flat faces ...
------------------
I've made three tools with dovetails, which means I know less about them than before I started and thought I did know something. They are a lathe milling vise, a boring head and a centering indicator holder that fits the ways of a small lathe.
The Machine Tool Reconditioning book describes measuring between cylinders pressed into the dovetail angles. You can use geometry to locate the angled faces, or just compare one with another to size them identically, then make the mating parts to fit. If you use drill bit shanks remember that they are undersized. Ground drill rod works well and may be useful for gaging or fixturing other jobs. If they are cut longer than the dovetail you can secure them in place with rubber bands.
At first I tediously machined them to a press fit that became looser with filing and stoning and use but leaving a gap for a spacer or gib is easier and allows adjusting for errors. Once two parts almost telescope together you can lightly bevel an edge, press them together and use the marked line of contact on the bevel to know how much more to remove, by measurement or short trial cuts.
I suppose the answer to how much clearance is: less than the throw of the cam that locks them together. Start small and try it, you can always remove more metal.
The trigonometric relationships of 45 and 60/30 triangles are worth remembering, because those angles are so common.
formatting link
I took a semester of college geometry in night school in which all the homework problems involved 45 and 30/60 triangles, so we practiced and memorized the relationships in our heads without needing a calculator. Then I applied it at work on phase angles in aerospace digital radio modulation schemes where it fit perfectly, again without a calculator. I did have to learn to think of angles in radians, as 2*pi is a full circle, pi/2 is a quadrant etc, but we didn't have to calculate their numerical values in degrees.