Face mill finish

[Edward]

Let me explain:

If I swipe the X axis from extreme to extreme, I have 0 deflection now, the indicator (a bit preloaded) stays on zero throughout. This is because I shimmed the column to be square to the table. That works perfectly and I've never had it so good. I used a very thin aluminium shim, the kind of sheet that you used to get from lithographic plates.

Now for the spindle: I attach the indicator to the spindle by means of one of these annoying articulated bars. So the indicator is touching the table 15cm away from the spindle. Then I swivel the spindle 180 degrees and measure the opposite side of the table. There is a difference of 0.4mm. There should be no difference in an ideal world, the reading should be exactly the same at both ends. I suspect this is due partly to a bit of spindle runout and partly to the spindle not being exactly square to the table.

Because this covers a total distance of 30cm. I don't think it's too bad.

I also did a test with a 10mm bar in the spindle, and measured the Z axis moving it up and down with an indicator just touching along the bar. Depending on how the spindle was turned, I got different readings, from perfect to some deflection, again to be expected due to spindle runout. All this measuring is enough to drive you crazy. The important thing is that the pieces are square using an engineers square and the face mill doesn't leave any ridges, just a few tooling marks which I would rather not have, but they polish off nicely anyway.

I guess I have to live with some imperfections. Next mill will certainly have a possibility to adjust both the column and the head. In hindsight maybe I should've bought a different mill, however, a different mill will have other problems. Nothing is perfect, not for that amount of money anyway.


Edward

[Robin Hewitt]

I don't understand the first test. It seems to measure bed flatness, it does not depend on the column being vertical to the bed.

The rotating DTI is good, but you want to do North South, East and West to get a vertical.

To compare sensitive quill movement to Z slide movement is trickier. How about putting a bar in a collet, putting your DTI against it from the side and from the front, then comparing quill movement to slide movement. How's that?

[Edward]

I don't understand the first test. It seems to measure bed flatness, it does not depend on the column being vertical to the bed.

The indicator holding prong is actually in the spindle, held inside an 8mm collet. The measuring lever is just touching the table. Therefore any readings of the table are relative to the spindle/head/column assembly. Since the only thing that moves is the swipe east to west of the table, , the reading will change if the table is not horizontal.

Then repeat for north to south travel, using a long parallel across the table to serve as a flat surface for the indicator to slide on.

[magicniner]

Doesn't the column bolt down to the base with 4 bolts?
If so you can shim front to back, then shim side to side.

[Edward]

Yes, as I said in post No. 4, that's how I did it, and it's perfect. But aligning the column is not enough, you also need to align the spindle axis.

To give you an example, if you tilt the whole column 20 degrees to the right, your spindle will also tilt 20 degrees to the right because it is attached to the column, but you want to be able to tilt both independently.

And the reason is that column and spindle don't come from factory 100 percent parallel to each other, unfortunately, if they did, then shimming the column would be all that is needed.

If there is no way of adjusting for this parallelism between column and spindle, well, you are buggered basically.

Edward

[magicniner]

Yes, as I said in post No. 4, that's how I did it, and it's perfect. But aligning the column is not enough, you also need to align the spindle axis.

To give you an example, if you tilt the whole column 20 degrees to the right, your spindle will also tilt 20 degrees to the right because it is attached to the column, but you want to be able to tilt both independently.

Edward

You don't tram by measuring the column with a non-angle adjustable head, you tram by measuring the error with a spindle mounted indicator.

Calculate the error in parts which will result, if it's not acceptable strip the head off and scrape the dovetail to correct.

[Edward]

I have a feeling we are talking at cross purposes:)

Forget the word "tram" for a minute.

Basically you align the column first so it is square to the table.

Then, and only then, you align the spindle to the table. In that order. That's provided you can do it. Not possible with a fixed head.

Alternatively, you make sure that column and head run parallel to each other, then you just square them to the table.

I guess modifying the dovetails would do it, if properly done, but that's way beyond my scope.

Edward

[magicniner]

The indicator holding prong is actually in the spindle, held inside an 8mm collet. The measuring lever is just touching the table. Therefore any readings of the table are relative to the spindle/head/column assembly. Since the only thing that moves is the swipe east to west of the table, , the reading will change if the table is not horizontal.

Incorrect, what you will get is the error between the plane of the table ways for the moving axis and the table surface, this will remain constant with your indicator mounted on any fixed point on the machine chassis, the column, and if clamped still, the head, are simply extensions from the fixed chassis of the machine.
If there is an angular error between the plane of the table surface and the ways then no amount of jiggling your column, head or spindle angle will "adjust" this away, the only way to correct this is by removing material ;-)