A couple of weeks ago I bought one of these ER32 collet chucks which I thought in mounting directly on the spindle of the myford would probably be stiffer than the flange/backplate type:
When it arrived I tested it with a range of different size collets by clamping milling cutters in collets and clocking the run-out. I came to the conclusion that it was parallel to the spindle (to beyond what I can measure), but has a concentricity error of ~0.0025", which is a shame because I had hoped for better - but then it is a "budget" item. I've contatced the supplier and they are sending a second one (which is nice of them - no complaints) but the cynic in me feels unless I'm very lucky it will be much the same. I have a dividing head with a Myford spindle, and for anything *I* am likely to do this chuck will be more than accurate enough for that, but as it looks like I'll end up with two of them I was wondering...
As it's only two and a half thou there is more than enough material to re-cut the collet seat, so I'm strongly tempted to see if I can re-cut the taper in-situ (which should essentially guarantee it's concentric, obviously). Three approaches spring to mind, all using the taper of the existing bore to set the topslide angle:
1. Bore it with a tipped boring bar, making absolutely sure the tip height is as closely aligned with the spindle axis as I can achieve (I'm aware that taper-turning is the one case where very accurate tool height really matters). The only thing that might be a problem here would be if the part has been hardened (I'll test this with a file on the front face somewhere).
2. Similar to (1), but using a dremel as a toolpost grinder, looking to take (say) a quarter to half a thou cut per pass with the lathe turning at (say) 100rpm. I'm not really keen on grinding in the lathe, but this once I guess I can accept it and ensure the ways are all covered up.
3. Turn up a lap to mount in the tailstock, using blue and the current taper to ensure the angle is correct, and then try lapping it back to concetricity. This would be messy but more gentle than (1) or (2), but I've never heard of anyone using lapping to correct a concentricity error.
Does anyone have any thoughts/guidance/derisive-laughter to offer?
AS