What's new
What's new

What extremes do people go to to get their machine level?

RC99

Diamond
Joined
Mar 26, 2005
Location
near Rockhampton, Queensland, Australia
Just wondering what is the general consensus of when you consider something is level enough...

I have been levelling my lathe with my newly acquired Starrett 199.. I have tried to get it near perfect within the sensitivity of the level but can only seem to manage to get it all within 1/2 a graduation so it is about 0.00025/12" which I assume is plenty good enough... One issue with the lathe is it has a centre support in the bed that does not touch the ground but is part of the cast base. If that support is not perfectly in line with either end supports it means the bed cannot be straight. but raised or lower in the middle..

But what sort of accuracy is the general consensus on what is a good standard on getting guide ways straight...
 
Just wondering what is the general consensus of when you consider something is level enough...

I have been levelling my lathe with my newly acquired Starrett 199.. I have tried to get it near perfect within the sensitivity of the level but can only seem to manage to get it all within 1/2 a graduation so it is about 0.00025/12" which I assume is plenty good enough... One issue with the lathe is it has a centre support in the bed that does not touch the ground but is part of the cast base. If that support is not perfectly in line with either end supports it means the bed cannot be straight. but raised or lower in the middle..

But what sort of accuracy is the general consensus on what is a good standard on getting guide ways straight...

What you have done is more than good enough, any better is just wasting time.
 
WHY? Machine tools operate quite satisfactorily on ships where they are probably level about .0001% of the time.

Yes you are dead correct, but I have no other means to determine when a plane is straight other then using a level to detect curvature in the plane..

You can use laser systems or autocollimator or wires to determine when a plane is straight, but I use a level..

It makes no difference if the level is even calibrated correctly as we are using it as a comparator, comparing one section to another, then adjusting so each sections are in the same plane..
 
WHY? Machine tools operate quite satisfactorily on ships where they are probably level about .0001% of the time.

RC is not talking about level as such, he is using the level to check for twist in the bed, it just so happens you need the machine to be level to enable checking for that twist.
 
I have South Bend publication #6514 and it states "Use a precision level of a least 12 " in length (don't ask me why, maybe because South Bend sold a 12" level) and sufficiently sensitive to show a distinct movement of the bubble when a .003" shim is place under one end of the level."

Ed P
 
I remember reading, i think it was in Australian Model Engineering Magazine, using a parrallel strip with a laser pointer attached. And pointing it to a dot (or line ?) on a distant wall. Due to the distance the error on the wall is a fraction of the error on a bed. Cant remember all the details now, i think it was like that.

Could be bashed into a neat usefull idea though.
 
I don't put any effort into leveling a lathe at all. I'm more concerned about the work it produces.:D

Two collars and between centers tests. Headstock in line with the ways and with the tailstock -- all three at once. Being level has nothing to do with it. Absolutely nothing.
 
Forget about home brew laser gimmicks. Many are ingenious enough but they rely on excruciatingly careful manipulation and human senses to even get within the range possible with a carpenter's level.

The reason for leveling a machine tools is that a level is by far the simplest and quickest way to put the machine into the alignment employed when the machine was first erected at the factory. If the machine is in correct alignment and rigidly supported it will produce satisfactory work whether bolted to a wall or on a ship in a storm.

A Starrett 199 with its ten arc second per graduation sensitivity is probably overkill for leveling an engine lathe but better than strictly necessary doesn't hurt. You don't HAVE to split graduations even though it's tempting to do so. A general purpose engine lathe needs to be within probably two graduations - 0.001"/10" - for satisfactory operation.

Using reveral technique, careful observation, regard for heat input when handling the level etc, a man with good vision can work a Starrett 199 level to 1/10 graduation fairly consistantly. However that lind of care and attention to detail is required only for more sensitive machine tools like planers and floor mills where very large element move on very long way bearings. Such machines are installed on large isolated concrete froundatons and their very success and longevity depend on tenths of thouandth over tens of feet linearity.

A three pedestal lathe poses a special problem. The middle pedestal might level satisfactorily but there remains the problem where a third point on a line can just as well define a curve as a straight line. The middle pedestal suggests the engineers who designed the lathe were concerned about the flexibility of the bed and they solved the problem with a third pedestal. It's easy to assume the middle pedestal when leveled holds the bed so it ways are linear but that assumption neglects linearity in a vertcal plane. The trace of the tool may turn a perfect cylinder if you are lucky but it not a slight hourglass or barrel shape is very possible.

Thus when larger lathes are installed in commercal shops they are bolted to their foundtation and among the hardware cast into the foundation are "kickers" - lateral jacks - placed at every lleveling point to nudge the machine sideways or endways a trifle to ensure the bedway aligns with the spindle axis and the carriage motor is linear in the horizontal and vertical planes. There are many ways to check a machine tool's linearity in a vertical plane. The classic way is toi run a reticuled microscope along a wire stretched parallel to the spindle axis but few have the necessary goodies.

A staightedge running the length of the max center distance supported on jacks or wood blocks with its reference face vertical and dialed in to the carriage motion will validate linearity just as well. The straight edge doesn't have to be that straight. A length of rectangular cold rolled steel or square tubing will work as well provided you map the readings. Once a firt et of readings are recorded and the exact reading points are marked you flip the reference and map them again using the same points. Symmetric readings demontrates linearity. Departures from symmetry represeent linearity error. Naturally you flip the straight edge back to its original orientation to validate foregoing readings by "repeat zero" error technique.

Adding: I neglected to mention proof bars and actual turning tests but I see other have brought them up. I'm a great fan of the two collar method and the long part turning test myself. Also to settle controversy, leveling is merely the simplest way to ensure alignments for lathes. Tools and techniques besides the level are required for other machine tools however the OP is about a three pedestal lathe.

A Starrett #98 level is quite satisfactory for leveling a small (less than 16" swing) lathe provided care is taken obtainng the readings. At 0.005" per foot graduations the #98 is almost not quite sensitve enough for the job but with care, reveral technique, and keen eyes it can be made to level surfaces within 0.001" per foot.

Many lathes can be satisfactorly intalled on a home shop's concrete floor. For that matter very small lathes can be mounted on a wood bench. Just remember that an annual check of the machine's alignment is necesary and if found to drift with time corrected and put on a shorter schedule.
 
Last edited:
I don't put any effort into leveling a lathe at all. I'm more concerned about the work it produces.:D

Two collars and between centers tests. Headstock in line with the ways and with the tailstock -- all three at once. Being level has nothing to do with it. Absolutely nothing.

This point is made time and again, and it always skirts the point: The goal is to remove twisting, and the only handy external reference is the earth's gravitational field and the local perpendicular vector. Starrett and others make a cheap (well, not really) sensor to determine that vector. You are correct in saying that level is unimportant, but its faster to get the machine aligned with the level than without it.
 
"The goal is to remove twisting, and the only handy external reference is the earth's gravitational field "

Wrong. The single best indicator that a lathe is correctly set up, is the
fact that it turns and bores true.

The missing point here, is that for used machinery, the level test can be
fulfilled perfectly - and yet the machine will not turn and bore true.
 
The missing point here, is that for used machinery, the level test can be fulfilled perfectly - and yet the machine will not turn and bore true.

I concur with jim rozen. I recently leveled my well-used 15" Leblond lathe with a master precision level. After the leveling, test cuts on a two collar workpiece showed a 0.0008" difference in diameters.
 
However that lind of care and attention to detail is required only for more sensitive machine tools like planers and floor mills where very large element move on very long way bearings. Such machines are installed on large isolated concrete froundatons and their very success and longevity depend on tenths of thouandth over tens of feet linearity.

Thanks Forrest, I was also after what sort of accuracy is required to align other machine tools, not just lathes..

I was looking at the thread in the CNC forum of the installation of a floor mill and was also wondering to what extreme they go to getting alignments correct there.. As shown here http://www.practicalmachinist.com/vb/cnc-machining/updated-cnc-floor-mill-install-pics-194142/
 
Try and error or measure?
If you align a lathe properly from bottom to top, you will have much less work tweaking it to turn dead cylindrical.
Get the twist and bow out of the bed, then align the spindle and then align the tailstock.
Only then, you'll be able to tune cylindricity by shifting the tailstock back and forth in a predictable manner.

Interesting sidenote:
DIN8605 (accuracy for lathes) says, that the spindle and tailstock may only point upwards and toward the tool. This is to compensate for cutting force and weight of work. Tolerances are quite tight. IIRC 0.01 mm over 300 mm.

Nick
 
"After the leveling, test cuts on a two collar workpiece showed a 0.0008" difference in diameters."

Heck, that's pretty good for an worn machine. Of course, you have to spec how
far apart the collars are, and if you could have gotten there without the level
anyway.

For a worn machine, you could be right in at the headstock, and also a foot
away from the headstock, but be out a thou right betrween the two.

Jim
 
"The goal is to remove twisting, and the only handy external reference is the earth's gravitational field "

Wrong. The single best indicator that a lathe is correctly set up, is the
fact that it turns and bores true.

The missing point here, is that for used machinery, the level test can be
fulfilled perfectly - and yet the machine will not turn and bore true.

Wrong. Read it again, Jim. "External reference" is the point that I was making. An untwisted condition is much easier to achieve if you start with a level. Tweaking it using turned collars? Of course.

Greg
 
"Tweaking it using turned collars? Of course."

Yes. This works just fine. For a used machine, you can chase
bubbles all over the place. But when you're done, it may or may not
still work. Then tweak it using turned collars. This gives solid
information about the real state of the ways on the machine.
 
"After the leveling, test cuts on a two collar workpiece showed a 0.0008" difference in diameters."

Heck, that's pretty good for an worn machine. Of course, you have to spec how
far apart the collars are, and if you could have gotten there without the level
anyway.

For a worn machine, you could be right in at the headstock, and also a foot
away from the headstock, but be out a thou right betrween the two.

Jim

The collars for the test cuts were about 2" in diameter and about 6" apart. Tweaking the leveling screws on the tailstock end yielded a 0.0004" difference in diameters. Prior to all of this tuning the lathe would cut a taper of 0.003" on about 6" of length.
 
Oh, here I go again. You start with the level and take the twist out of the bed to the best degree you can. Then you cut, two collar, or use a precision test bar. Now you have a piece of data- what sort of taper is the machine producing? There is some limit where you have to decide if the problem can be fixed with a very slight twist of the bed, or if something else is wrong that needs to be fixed first. If the bed isn't seriously worn and you put a twist in it, both the bed and carriage will wear prematurely, not to mention possible instability and chatter if the carriage isn't seating properly. Turning the bed into a pretzel is rarely the answer to making a lathe perform at its best.

CH
 








 
Back
Top