Sent a part off to be measured properly... ahem...

Hello all,

I recently moved my gear into a new unit and had to strip and rebuild both my mini-mill and the gantry router. I've long known that both machines have been out both in terms of skew and travel, but the kind of work I do (non-precision) and the scales at which I work (small) means it hasn't really been a problem - where I've needed a bit more accuracy I'd compensate at the drawing stage for the skew I've managed to measure (using high resolution scans of, say, a matrix of spot drill holes).

After cleaning and re-building the mill, I finally got around to installing the millkins module which allows me to tell LinuxCNC how much to compensate for skew in the table during commanded moves (ie. when running g-code or executing an MDI command, but not during manual moves via the cursor keys). Since I was doing this and wanted to do it as precisely as I could, I thought I'd drop a line to a professional metrology service (who have some pretty hefty clients, from automotive through to aerospace) to see how much it would cost to measure up a simple aluminium plate that I'd machined. They got back to me and very kindly said that as a one-off they'd do it for free. I got the results back today. The following pics are a scan of the report and a quick visual of how much I'm out by (programmed dimensions in black, measured in red...)

Attachment 26673 Attachment 26674

Using the skew value of 90.06 - this works out as the axis running out by around 1.5mm over a metre..!

Notice that in the report there's also mention of a bit of deviation in parallelism between the pairs of planes. Not sure what to make of that (or how to correct for it).

As a matter of interest, when measuring my high res scans, I got the angles as 90.06 and 89.90 - the X and Y length and width, however, were over by several hundredths, probably due to scanner optics, interpolation etc. - but as you can see, the scanner's not a bad place to start working out for skew - and now I know how much the X and Y dimensions are off by I can always work out the 'real' figures from there.

Anyway, it was really interesting to get an insight into how far out the mini-mill was out by, and of course to now be able to dial in the corrections with some level of accuracy. Not as good as having a properly trued up machine, but that's life when you don't have a foolproof way of setting perfect perpendicularity between your X and Y axis..! EDIT: Also quite an eye opener to see how far off the simple X/Y travel is - I'm assuming this is mainly down to the lower accuracy standards of the cheaper ballscrews..?

Router gets set up next...