I decided to go with just 3" long blocks for the risers, so I cut 8 pieces of MDF and drilled holes in them.
Then I bolted down the front end plate - squaring it up to the face and re-bolted together the machine. (See earlier post.)
Then I made all four ends of the rails the same height again (see earlier post again.) It went a lot faster this time since I had a good method for doing it. All four corners are within about 0.002" in height from the table top.
Then it was time to calibrate the steps/mm. The default value (40.020) will get you close, but I changed the belts to 9mm and made them pretty tight and discovered that it was off quite a bit.
The first step is to measure your 1-2-3 blocks. I happen to know that mine are not exactly 3.000" long, so I measured each one and added all four together to get 11.995" (three measure 2.999" and one was 2.998".)
Next I stacked up my four 1-2-3 blocks end to end with a square clamped to the end for a stop. I bolted this to the table just to keep it from moving.
I attached my test indicator to the machine using the 1/4" dowel I added to the spindle mount and set it to zero on the end of the stacked up blocks. Next I lifted up an inch to make sure I cleared the bolt, moved down 11.995" and looked at the indicator. It took me a couple of tries before I realized that it was moving so much too far that it was maxing out the indicator!
Once I figured that out I tried again and told it to only move 11.900. Then I jogged it over 0.010" at a time until it moved the needle past zero. The programmed travel was 11.965 for an actual movement of 11.995.
So I took the programmed travel and divided it by the actual travel to get the correction factor (0.9975) Then I multiplied the current steps per mm by the correction factor to get the new steps per mm (39.919). I entered this in as $101 and then ran the test again to confirm. It was close, but not quite right, so I ran the correction calculation again and came up with a new steps per mm of 39.902. This time the test was within 0.001" and I called it good.
Then I rotated the stack of blocks 90 degrees and repeated the process on the X axis. The new steps per mm ended up being 39.905 - pretty close to the Y value (not really a surprise.)
Finally I moved on to the Z. For the Z, since I only have about 4" of travel, I just stood a 1-2-3 block up on top of another on for a 4" height. I had to move the test indicator to a 1/4" dowel fastened into the collet. Then I set the test indicator to zero on the top and moved down 4.000" The correction was not a much as the X and Y - I assume because the belt is not as tight. The new steps per mm ended up being 40.012.
Tomorrow I'll run the diamond-circle-square test again - maybe slow the feed down a little. I expect it will come out just about on the money this time. I might make some square pockets too just to see how close they come.
I might also do some speed tests milling holes. I have a job I currently do by hand using Forstner bits on my drill press. It takes about 12 minutes per part to do 50 holes by hand. My Shapeoko 1 was not rigid enough to do that job faster but I suspect the Shapeoko 3 will be.