Can you believe it? I actually had time to play with the mill.
First thing is/was to take some measurements and see how in tram it is.
I used the Rollie Dad method. Many of you know this but I'd thought I'd show my take on it for anyone who doesn't.
And
it's a way to get someone to reply and double-check me.
The approach is to 1st measure how true the head is to the column and make adjustments as needed.
Then to measure how true the column is to the table and make adjustments as needed.
Tools include a round straight rod, measuring device, and some means to attach the measuring device.
I used a 1/2" diameter roller out of a printer.
I could not use my dial test indicator as it doesn't provide sufficient range before the tip bottoms or tops out.
I used a dial indicator (.001) and did my best eye-balling I could.
The dial indicator has to be mounted from the column.
Mounting from the table gives erroneous results if the column is not true to the table.
The left diagram shows the ideal case (measuring in Y). The column is the large rectangle, the head the smaller rectangle, with the rod sticking down from it.
Just to the right shows the indicator when it's at the top or bottom position.
Take a measurement near the head. If there's any swing then move the 0 marker such that the needle swings the same distance either side.
Run the head up, and take another measurement, without changing the indicator setting.
In the ideal case, the indicator would not move or at least move the same amount either side of 0.
If the rod is truly round and straight, any movement would indicate the rod or collet is not perfectly central to the spindle.
The right diagram shows the effect if the rod (again, it must be straight) happens to be installed crookedly (or the collet is not true to the spindle).
Again, the top reading is made and the indicator adjusted such that any swing of the needle is the same to either side of 0.
The second reading will swing (because the rod is out of kilter) but in the ideal case would swing the same to either side of 0.
The 2nd attachment shows the head tilted and not true to the column. Repeating the measurements would show the 2nd reading (at the bottom of the rod) to swing a different distance to either side of 0. For example, if the needle swings left by 7 (that is, -7) but only right by 3 (that is, +3) then you add the two then divide by two to get the amount the head is out by. In this case -7+3 = -4, -4/2 = -2. The sign shows in what direction the head is tilted. (But which direction I haven't looked into yet.)
The 3rd attachment shows the effect against the table (assuming the table is true to the column).
The angle at the top is the same as the angle you see below the table (the arcs).
The purpose was to see how much the error in the Y head tilt would add to the tilt of the table to the column.
To find the angle at the top...
y = h*sin(angle)
sin(angle) = y / h (we will need this)
angle = arcsin(y/h) (we won't need this)
y, in my example is -2 (or just 2).
I made my measurements 4.75-ish apart.
The angle is so small (and the -ish in 4.75 so large) I can use this for h.
sin(angle) = 2/4.75
Assuming a 24" table, and therefore 12" from center to end, the distance at the far end from flat is again y = h * sin(angle).
h is 12" and the angle is as above.
y = 12 * sin(angle)
y = 12 * 2/4.75
Anyway...
Here's what I got on my machine...
In the Y direction, I got a swing of -.0025 to .0019 (ish) giving a tilt of (.0019 - .0025) / 2 = -.0003.
Additional measurements still came below .0006.
I don't think I'll touch that.
In the X direction, I got a swing of -.003 to .0015 (ish) giving a tilt of (.0015 - .003) / 2 = -.00075.
A bit worse but there's not a lot of travel in X. I don't think I'll touch that.
Then I inserted a short rod in the spindle, attached a long rod to it at an angle, and a dial test indicator to the end.
I lowered the indicator on one corner of the table until it just touched.
I set the fine feed control on the quill?spindle? to 0, lifted off, moved the indicator to the other corner on the same end.
I moved the indicator down (I already knew I was at the highest corner) until the indicator changed. The fine feed then told me how much lower that corner is.
Just under .001. That may be okay. Basically .00017 per inch of travel in X.
Then I swung the indicator to the other end of the table.
.026.
That's probably not good. That's better than .001 per inch along the table.
Going back to the discussion of the table and the tilt of .0003...
The contribution to the distance one end of the table changes from center is
y = 12 * (.0003 / 4.75) = .00076. Not a lot. Less than .000063 per inch.
The column is not true to the table in Y.
The only recourse is to shim the column.
But before I do that...I will repeat the measurements several times.
Sorry for the length of this post. I hope it helps someone.