Good thoughts on the backlash.. I am on Rack and pinion for X&Y and with all my testing have seriously never measured any backlash those axis... So that's a gap on my part to consider that.. Machines on belts or acme screws could absolutely have backlash.. (precision ground ball screws don't generally have much backlash either).. If you have already tuned backlash though, that would be compensated into the calculation... Kind of a chicken and egg thing though right.. Mach 3 and 4 have backlash compensation parameters.. So you could tune backlash, then run this process, then tune the backlash again.. back and forth and such until you are happy with both results..) As for the Z, I was thinking very similarly, but my z column gets in the way pretty quickly when trying to do that.. my max Z movement is only 8" and by the time I move one direction or another for the column to be out of the way, I can't trust that my bed is flat enough to be super precise (more precise then testing 1" and extrapolating I suppose... ).. These are challenges on large format routers that probably don't exist on mills.. (man I want a mill :-P)

deck a flat area say the width of your largest router bit. then stack your 4in gauge block on that, then ring your stack together on top of that. should give you enough room to clear your spindle. may have to remove the dust shoe. as to the backlash, best practice to eliminate any and all possible errors if you're shooting for accuracy. backlash comp is good to have, but i'd make my inital steps per in calcs with out having to rely on that if possible, that said, the 8in gauge block stack will give you way better results than the 1in stroke on the dial indicator alone.

Though has crossed my mind... Not sure if 4" throw is enough to not knock over the gauge block stack.. worth trying though.. when I was stacking things up and looking at the problem sunday evening I didn't think the 4 was long enough to act as a base.. (if only my surface was conductive and trammed in).. Yah, there is an inside probing options (right next to the outside one I used).. it would work for your approach.. guess you would only have four hits to possible backlash in that model... I've got 7 direction changes from this approach.. And calculating the ball... (though precision of the ball still has to be accounted for)... While an oversight on my part, its also a credit to CNCRouterParts Pro machines that my lack of ever finding any backlash meant I forgot it in my tuning process...

artisandice your comments about backlash and such made me go back and double check with a better indicator.. Instead of using a digital dial indicator, I switched over to an analog .0005 test indicator.. low and behold I DO have .001~ backlash on my Y axis... while my Dial indicator says .0005 resolution.. its +/- .001.. so the .001 wasn't visible.. . Sadly I cant fix it as my Ethernet Smoothstepper doesn't have Backlash Compensation support with Mach 4 (yet)... Warp 9's forum says they are working on that now.. so here's to hoping.. So odd numbered direction changes will be off by .001... for now... won't test the others until the Warp 9 driver upgrade comes out... then I'll tune them all again.. AND do a video on that :-)..

Yah, I thought about using my 123 blocks and bolting them togeather to get a 6" throw. But as you mentioned not as precise.. as for squareness... I think a 1/10ths dial indicator attached to the Z axis and a couple of 123 blocks or 234 blocks bolted togeather in an L might make for a great test... would have to tram the blocks in carefully, then lock them down.. but yah.. could work pretty well.

Legit concerns.. Because my machine is mostly aluminum and my spindle case is not magnetic using a dial indicator vs slotting out a spot wasn't really an option (and trust me this is an annoyance for me when I square up my vice... ). My goal with the slot was to make sure I am "square" to my machines axis, and best way to do that was to let it lay tell me what its axis were.. Also the cosine error for being off as long as I am close to the center in both directions is actually pretty tiny... Really, now that I am thinking about it, probably could have skipped the trench and just eyeballed that.. even a 1/10th of an inch of of square, over 8.5" thats .00059" off in cosine error.. Ringing the blocks togeather... again I did try that off camera but two things.. 1, they wouldn't stay rung when I placed them in the trench and 2 those larger blocks (with my low grade gage blocks) really didn't want to ring togeather and hold.. So I am balancing various issues when I came to my solution.. Its not great, but its far better than your standard half thou dial indicator and a 1" movement.

Bob Giggy the shank is 3/8. The ball is 1/2"

Bob Giggy oh and thanks! Glad you liked it :)