Leveling lathe on uneven floor before a CNC retrofit. Suggestions?

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I've got good 3D thinking skills and good making skills, but still such problems challenge me. I'm sure I'm not alone. It would be interesting to distill this process into a FAQ or how-to. Or maybe someone knows of a set of really good "How to level a lathe" instructions somewhere?

The Myford ML7 User's Manual, widely available in free PDF form on the net, has about the best run through on setting a small lathe bed level.

But trying to get the bench level to the same sort of precision is a waste of time. In fact, many deliberately set the bench off-level so the coolant drains to to one corner for collection.

Yet others just sit the lathe on the bench, loosely bolted down at the headstock end only and let it assume its natural position. If the lathe bed is not warped, it will work fine like this.

My own ML7 is on a steel bench that is not bolted to the not-flat concrete floor and the lathe itself is packed up on two pieces of 20mm x 40mm soft pine wood with the mounting bolts running through them to the steel bench top below. It turns parallel to within less than two tenths of a thou as near as I can measure. (This was a temporary set up when I was overhauling the lathe but it works so well I left it like that.)

So there really is no definitive way to set the whole thing up. As long as it turns parallel at the end of the day, all is good.
As Baron says the word level is not being correctly used the removing of bed twist is the critical factor
The twist can be measured/removed easier if the lathe is level.If the lathe is raised say 50mm at the tailstock
it wont affect anything.Someone posted on tilting the mini lathe forward at say 30o to improve working
visibility among other things just like a cnc machining centre.It does not affect the lathe accuracy
If you take a 150mm thk slab of granite which is dead flat and you level it you can check the flatness
If you put a 25mm spacer under one corner the granite is still flat but NOT level and how do you then check for flatness
Generally in the context of lathe bed setting up, to "level" the bed means to set the bed straight by using a precision level. I think that is what we are talking about in relation to the lathe itself.
It doesn't have to be actually level. For instance, if the bubble sits two marks off level at both ends of the bed and in the middle, the bed can be said to be straight, or "level". As long as it's the same all along.

It's just easier to make it dead level rather than remember how many marks this way or that way the last reading was.
I have a lathe, and I did have a CNC.
I have found by far the best material for levelling is to use CORIAN - a Dupont plastic used for kitchen worktops. Ofcuts are very cheap, and are redilly available from any Kitchen installer. The part they cut out for the sink is never needed by them again. It is very easily sawn, and equally easy to file. It is approx 15/32 thick.

If you want more than one thickness just fine sand it, coat it in suuperglue, and clamp overnight.

It is also a really good material to use for making formers.
Hope this helps you
thanks to all who replied to my partial hijack of this thread. Thanks for the tip about Corian. I know the materials but hadn't thought of using it this way. As far as gluing Corian, as well as wood and others materialism I have found Loctite PL Premium to be an excellent adhesive, FYI.
Generally in the context of lathe bed setting up, to "level" the bed means to set the bed straight by using a precision level. I think that is what we are talking about in relation to the lathe itself.

Yes "but" terms can morph and lose some of the primary meaning. The intent is for the machine to cut straight. "Straight" has come to equal "level" but is that accurate to the intent? While a precision level is one way to determine if a bed is straight, no droop or rise, and I have one, it isn't the only way. Remember my example of the 10,000s of lathes etc. mounted on ships. A precision level could only be used while a ship was in dry-dock. Hence the taper cut check. My lathe cut straight long before I happened to get the level only because a guy was bailing out of his hobby after a flood and life change and having a "flood" sale.

I'm curious how stiff your 13x37 bench style lathe is. I have a 10x24 that used to 'dance on the floor' with any heavy cuts, and especially cut-off operations. This pretty much caused the lathe to always be out of level, and with a twisted bed.
I fixed this by purchasing a piece of 2 inch thick hot rolled steel and drilling/tapping the existing lathe base flange bolt pattern into this base. As my lathe stand was purchased for a different lathe (the harbor freight 12x36), I copied the bolt pattern of that base to the bottom side of the 2 inch slab. I mounted that on the sheet steel base, leveled it, then placed the lathe on that with its chip pan. I used shim stock to then level the lathe and remove any bed twist.
Now I can hear the frying bacon sounds when doing any cutoff operations, and can do DOC of more then 0.125 inch with turning or facing.
If your 13x37 is stiff enough you may have never experienced my issues, and if so this post is not germane to helping your leveling issue.
From your description of the concrete floor level, I'm guessing the original contractor did not compact the fill properly before the concrete pour, and it settled in the middle. I see one crack in your photo that hints of this problem.
I saw in a post once a sub frame made from pallet racking. The lathe was bolted to this and the subframe had leveling feet attached. I have my lathe sitting on leveling feet at the moment but will be going down this sub frame path in the future as I think it makes the whole structure much more rigid. The extra height can be an advantage too. At present I use a tall bar stool to sit on. Pallets are free and can be used to make a floor area around machinery.
This thread (and many like them) has prompted me to experiment with something I've been wanting to do for some time - a 3D representation of a typical twisted lathe bed, just to get a feel for what kind of deviation measurements could be expected. Consider this just a preliminary rough idea. If there is some interest or suggestions I can develop it further.

The basic notion is to have the ability to input parameters into a Cad drawn lathe bed model that results in a specific amount of twist & then take measurements off the resultant surfaces as though I had a real life dial test indicator against a perfectly cylindrical test bar. The resultant math wont lie as long as the inputs are reasonable (which I suspect is more in depth than first meets the eye). Here is where I got so far. It is simplified in that:
- the headstock plane defines the spindle center and the rail section relative to that
- the cylindrical test bar (right now) extends perpendicular from head stock plane
- In this first example I 'raise' the lathe stand foot a fixed amount at the TS end. This displaces the TS rail section both upward but also at some resultant slight angle because the 2 rails are joined via the bed webbing. So kind of like 2 people holding the ends of a ladder, but one person raises only one hand slightly. I then extrude the fixed HS section to the displaced TS section along a smooth minimal path to replicate this twist deviation.along the full length
- For now I have only drawn the front rail (way), because that's what I'm taking my artificial measurement on.
- I take a reference perpendicular DTI-like measurement on the headstock end (orange lines). So starting from the corner of the rail section, up Y-amount & over X-amount to contact the test bar quadrant. Reference these 2 values.
- Then at some defined distance away from HS I make a new measurement plane. I superimpose the deviated partially twisted rail section on this plane & take another measurement relative to cylindrical test bar. I preserve the Y-amount from original HS value, extend perpendicular to the deviated section (like a real life DTI) & compare the distance to test bar.
Hopefully this makes sense with a sketch


  • 2018-08-02_20-54-47.jpg
    51.3 KB · Views: 439
Section view of TS end. Raises the foot 0.25", assumes the lathe stand is 28" wide so 14" half width. The rail section displaces 0.071"


  • SNAG-8-2-2018 0000.jpg
    SNAG-8-2-2018 0000.jpg
    20 KB · Views: 420
Measurement at the outboard plane (arbitrary placed 13" from HS plane) encompassing the partial rail twist at this station. Measurement is 2.725".
So HS reference 2.750 - @13" measured 2.725 = resultant 0.025" DTI type observation by raising the front TS foot 0.250" on a 40" long lathe bed.


  • SNAG-8-2-2018 0005.jpg
    SNAG-8-2-2018 0005.jpg
    25.4 KB · Views: 417
Assuming I did this right, even this 1-leg up + 3 legs pinned it is a simplification. What I wonder about is if the front & rear rails are deviated (gently bowed) but essentially parallel. We typically measure across the bed at close intervals to map for lathe twist, but I'm almost thinking its just as important to measure incrementally along the Long axis from HS to TS or vise-versa. And would the cross bed measurements see much difference with this going on? But the net effect would also show up as test bar mis-measurement down the lathe bed depending on the hump arc position, no?


  • SNAG-8-2-2018 0007.jpg
    SNAG-8-2-2018 0007.jpg
    19.7 KB · Views: 436
  • SNAG-8-2-2018 0008.jpg
    SNAG-8-2-2018 0008.jpg
    10 KB · Views: 416
  • 2018-08-02_22-27-02.jpg
    18.3 KB · Views: 440
There's not much you can do about lack of bed straightness longitudinally. It is what it is. Shimming the feet at the tailstock end can take up twist but it won't take out a longitudinal bow. What levelling is doing in essence is taking the bed back to having the feet sitting in the same plane they were in when the bed was originally machined. If the casting has bowed longitudinally since then, regrinding is probably the only cure.
Apologies but my odd thoughts are different.

The FIRST test is establish whether the unbolted lathe is in twist or not. There are tests for a prismatic lathe which differ rather than the flat bed Myford. I'm not at issue with Hopper and Co.
However as Hopper might agree, opinions are getting very close to mine.
I would probably use a laser through the to see where it ends up. Will it hit the poppet in the headstock spindle- or not?

Again, the taught wire system between the centers is an established system.

What has to be recalled is that the so called Myford system using a test bar or cutting two equal or unequal rings only really applies when the lathe bed is new- and not in twist.

I read a bit on Geometers Microscope on the Lathe articles which were in Model Engineer and seem to have appeared now on the 'net.

And 'yes', I wrote up an article moons ago in Model Engineer when I had a friends Myford ML7 Blancharded which-as Hopper suggests) starts with flipping the lathe bed onto it bed and packing up the worn hollows and doing the feet first, then flipping it back and getting a reference- which has yet to be discussed, to look at the rest of the possible problems.

I hope that the foregoing does not cause upset

What levelling is doing in essence is taking the bed back to having the feet sitting in the same plane they were in when the bed was originally machined. If the casting has bowed longitudinally since then, regrinding is probably the only cure.

And this is the point in the evaluation process I think I fumbled my previous words. If the bed twist is restored to factory setup (which I thought the OP had done by shimming) and presuming no lathe bed wear issues and the DTI is STILL measuring deflection down the spindle supported test bar and the lathe is of the type with detachable headstock, then this should be checked before committing to the re-grind. Using the same arbitrary example above, a measured deviation of 0.025" at a point 13" down the test bar works out to just over 1/10 degree of headstock rotation. That's a pretty tiny amount for a chunk of assembly held in place with 4 bolts & some adjustment setscrews & was subsequently added to the freshly ground bed surface. I've often wondered how easy it would be to displace the headstock like when moving a lathe with straps around the headstock area.
I have heard that too, but cant't find any documented, compelling reason. This link comes somewhat close.

If the HS was pointing towards front of lathe off bed axis, you would theoretically cut a taper in longitudinal turning mode, with reduced diameter on TS end. Also you would cut a concave dish profile in cross feed mode, for example stock mounted on a face plate. If HS was pointing towards rear of lathe off bed axis, the opposite would occur.

I've also heard another story, maybe true, maybe not. For example someone had a zero degree HS alignment, turned some bar stock which extended from chuck unsupported by TS & measured the outboard diameter to be slightly larger than the inboard. Thus concluded the HS should be yawed slightly towards front of lathe & repeat the cut until the bar is cylindrical; equal diameter on both ends. The only problem is the source of this cutting effect is because the tool is not getting full depth penetration on the TS end. It is deflecting the stock slightly. As the tool gets closer to the HS end, the stock deflects less (cantilever beam principle) & cuts truer to the actual depth of feed. So less to do with alignment & more about material deflection. That's why personally I prefer an accurately ground test bar set in the HS taper socket. No chuck, no stock, no cutting forces, no surface variations. Eliminate the middle men LOL.

Latest posts