- Joined
- Mar 2, 2018
- Messages
- 232
After doing some precision work using the weird 20s and 2s dials, and bending my mind into a pretzel to manage backlash, I finally decided to shelve my dream of buying a really top shelf DRO setup, and just put something on there for now to get it done. I went with the iGaging absolute scales, used the readouts they came with, and I even took a hammer to the factory brackets to convince them to serve my purposes. It's a pretty ugly installation, but it seems solid enough.
I went looking for a standard procedure for dialing in scales, and didn't find anything. I decided to try loosening everything a touch, and running the axis back and forth several times to let it find its own center. After I did that, I decided to try checking the DRO against the dials. Over one inch, the DRO and dials are in total agreement. The larger the distance I measure, the more there seems to be a cumulative error. I hit 1.000, 1.9995, 2.9995, 3.001, 4.001, 5.0015 or thereabouts.
Like all hobby guys, I obsess over trying to work to the nearest millionth of a picometer whenever possible. My B grade gauge blocks might not be a sufficiently accurate reference measurement, and my work might be way off, by fractions of an atom, which would mean I am a failure. I mean damn it man, if I'm not slicing quarks exactly in half, how can I look at myself in the mirror and say that I am honestly doing my best work?
But seriously, I know these scales have issues due to a variety of considerations. The resolution isn't that fine. They measure natively in those whacky oh là là French units, and are subject to rounding errors. They are especially unreliable at the tenths level. I also know the dials on my mill are questionable, because everybody who built and designed the thing thinks in metric, and they're meant to read in Merikan units. I don't know about the g0704 specifically, but there is some crazy plan out there how to make a new acme screw to fix the error on the g0602 cross-feed, which reads 0.9956" per inch or something. I'm looking at a lot of sources of error, and I just have no concept of how happy I should be with how this is set up here and now. Certainly, it's a lot closer to being accurate than I ever was before. Honestly, on the mill, I've been working to 0.010" tolerances up until recently.
Finally, I've noticed that both scales have a dead zone on the order of 0.250" at all four limits of travel. Crank the axis to the limit, set a zero, start cranking in the other direction, and nothing happens well after the backlash is taken up. I marked with a sharpie, and here is where the table started moving, and over here at least 0.250" to sometimes even 1.750" away is where the DRO picked up movement. Uh? Did I do something wrong, or do these cheap scales just suck? Maybe I somehow messed up the magical PCB nonsense inside the scale when I cut it, but why is there a dead zone on the factory end too?
I went looking for a standard procedure for dialing in scales, and didn't find anything. I decided to try loosening everything a touch, and running the axis back and forth several times to let it find its own center. After I did that, I decided to try checking the DRO against the dials. Over one inch, the DRO and dials are in total agreement. The larger the distance I measure, the more there seems to be a cumulative error. I hit 1.000, 1.9995, 2.9995, 3.001, 4.001, 5.0015 or thereabouts.
Like all hobby guys, I obsess over trying to work to the nearest millionth of a picometer whenever possible. My B grade gauge blocks might not be a sufficiently accurate reference measurement, and my work might be way off, by fractions of an atom, which would mean I am a failure. I mean damn it man, if I'm not slicing quarks exactly in half, how can I look at myself in the mirror and say that I am honestly doing my best work?
But seriously, I know these scales have issues due to a variety of considerations. The resolution isn't that fine. They measure natively in those whacky oh là là French units, and are subject to rounding errors. They are especially unreliable at the tenths level. I also know the dials on my mill are questionable, because everybody who built and designed the thing thinks in metric, and they're meant to read in Merikan units. I don't know about the g0704 specifically, but there is some crazy plan out there how to make a new acme screw to fix the error on the g0602 cross-feed, which reads 0.9956" per inch or something. I'm looking at a lot of sources of error, and I just have no concept of how happy I should be with how this is set up here and now. Certainly, it's a lot closer to being accurate than I ever was before. Honestly, on the mill, I've been working to 0.010" tolerances up until recently.
Finally, I've noticed that both scales have a dead zone on the order of 0.250" at all four limits of travel. Crank the axis to the limit, set a zero, start cranking in the other direction, and nothing happens well after the backlash is taken up. I marked with a sharpie, and here is where the table started moving, and over here at least 0.250" to sometimes even 1.750" away is where the DRO picked up movement. Uh? Did I do something wrong, or do these cheap scales just suck? Maybe I somehow messed up the magical PCB nonsense inside the scale when I cut it, but why is there a dead zone on the factory end too?