I have a question about lathe alignment

[verbotenwhisky]

I'm writing this from my sick bed. It seems I have a lot of time to think right now.
I have always had about .004" to .006" taper over a 6" long cut. I have leveled my lathe but I have a theory why it may not stay level. This is a South Bend 9" lathe. It is mounted on a nice shop built steel cabinet bench with a top made of .125" steel. Where I think the problem may be is, this lathe is in a shed beside my home. The floor is 3/4" particle board sub-flooring on 2 X 8 joists 8 ft long. With the heavy machines, this floor has some give or "bounce". If I lift one corner of the lathe bench even .050" it throws the level out. I don't believe it is possible to keep things level on this floor. Now to my question. The lathe only need to be level to itself ( not the world. They put them on ships in a shop and the ship is not always level to the earth). Short of rebuilding the shop, I am wondering if placing a 40" X 20" steel plate 1/2" or thicker on the bench top and then mounting the lathe on the plate if that would help solve the problem of constant changing of the leveling. The lathe could be aligned to the very stiff plate and remain constant. Am I way out in left field in this thinking?

I suspect you have a twist in your ways/bed, it doesn't take much of a twist to produce those results. are you using a machinist level to level the machine with? Keep in mind your not worried about the level of the earth, your concerned with the level from one end of the ways to the other from front top back.

 

[tq60]

Started from the beginning and quickly looked through this, no correct alignment of tailstock took place and testing is between centers.

Remember, any error in the tailstock alignment is doubled, so if off by 0.005 the error is 0.010.

Eyeballs are not that good.

If metal cannot be found, use a chunk of wood or pvc or just about anything handy.

Until you get the 2 collar test to be accurate, and between centers work will be bad.

Most unsupported work from the chuck is close in, most headstock rest on v- ways or has good alignment, most of the time.

For close in work a twisted bed will not matter much


Sent from my SM-G781V using Tapatalk

 

[verbotenwhisky]

Started from the beginning and quickly looked through this, no correct alignment of tailstock took place and testing is between centers.

Remember, any error in the tailstock alignment is doubled, so if off by 0.005 the error is 0.010.

Eyeballs are not that good.

If metal cannot be found, use a chunk of wood or pvc or just about anything handy.

Until you get the 2 collar test to be accurate, and between centers work will be bad.

Most unsupported work from the chuck is close in, most headstock rest on v- ways or has good alignment, most of the time.

For close in work a twisted bed will not matter much


Sent from my SM-G781V using Tapatalk

Based upon my limited experience I disagree with that assessment.