On Sat, 2009-06-13 at 11:27 +0100, Andy Pugh wrote: > Most of the dimensions for the general geometry were +/- 0.2mm except > for the flexural element, which was 0.2mm +/- 0.05mm dimensioned from > a face with a stacked-up positional tolerance of about 0.4mm. > The machinist set up his CNC mill to the centre value of each > tolerance starting from a part edge and pressed "go". When the program > finished the flexural element was not even there. > > Who was at fault? I argued that the wider tolerances elsewhere in the > geometry were specifically so that they could get the flexure right, > they said "You always work to mid-tolerance, and the drawing should > assume that"
The machinist. I used to do job shop work in my shop. The feature that you described should be inspect-able, and the from dimension you quote it should have measured (by whatever method is appropriate) between .15mm and .25mm. Since it was not there, we can assume that it would measure 0.00mm and is therefore out of tolerance. I have heard this "middle of the tolerance" argument many times and it is as wrong today as it ever has been. And I was on the side (machinists) that tried to use it to our advantage! If the stack-up of multiple tolerances actually prevent the part from being made such that each individual tolerance limit can be observed, then the drawing should go back to drafting with that explanation. If you actually make the part, then it must pass inspection, even if some features must be created near their tolerance limits to allow other features to exist within their own tolerance limits. The "middle" of the tolerance band is no more valid or important than any other point within the tolerance limits. Wow, who would have thought I was this opinionated? :) Thanks, Matt ------------------------------------------------------------------------------ Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
