View Full Version : Scraping a lathe bed

David S Newman
03-15-2007, 04:14 PM
I can and have many times hand scraped many flat surfaces, i'm about to hand scrape a lathe bed with dovetail ways, how do I keep the dovetails parallel using a micrometer? Must be a simple answer but I can't think of it. Hope someone can help . David

03-15-2007, 04:34 PM
i'll take a shot, but don't start removing metal on this recomendation alone! starting with the longest of the two mating surfaces, get the horizontal sections in the same plane with a reference flat and indicator on a surface guage. then get one angled surface flat (exact angle doesn't matter, you scrape the mating part to fit) now you've only got one angle dovetail surface left. you scrape it flat same as the first one, with a reference surface and get distance between the two angled planes constant by putting a piece of ground shafting in each V and measuring between them.

David, like you I've done scraping and have some experience on dovetails (mill) and think the above will work, but take it for what its worth - opinion of the semi experienced. you could always get Connelly's book - Machine Tool Reconditioning or others here may know more. one issue is the above technique does nothing to keep the dovetails aligned with the headstock and would require realigning the headstock

03-15-2007, 04:53 PM
I second the recommendation to get Edward F. Connelly's book. Its $90some well spent if you are going to tackle a lathe. I am trying to finish up my vertical mill. While it has more surfaces, it ends up being less difficult to fixture for than does the lathe with its long bed ways.

Having just read a bit about lathe scraping in the above book and a few others, I can tell you that levelling will be more critical at this point than ever again. You don't want to find yourself scraping out a wind (twist) that is only there because the whole machine was twisted at the time the work was done. You would scrape a twist into the ways.

One over-riding principle in Connelly's book is that a machine wants to go back to its correct architecture. That is, they may wear low in places, but the minimal scraping will take it back to flat without changing angles etc. provided the wear is not very bad. You use unworn portions to help show you what to do and to help maintain the proper planes to the flat surfaces. I found personally, that establishing just how a way surface is out of kilter and knowing that in advance is pretty important to not heading off on a tangent.


David S Newman
03-15-2007, 04:58 PM
Magyver, I understand what your saying but it's the male side i'll be scraping first , could measure the female as you suggest but how would I accurately place rollers on the male dovetails to measure ? David

03-15-2007, 05:29 PM
Dovetail ways...dovetail ways....this just sunk in.

The one and only means I used for checking parallel on the dovetail knee ways on my milling machine was to use a pair of ground dowel pins that were sized to contact the faces of the dovetails about mid point (doesn't have to be exact).

As I worked, a couple of times...and especially at the end to verify things...I mic'd across the pins using a large micrometer to see that they were becoming more parallel and not less. I plotted the variation with a marker along their length so I knew what needed to come off.

I used this method which assumes you have good, minimally worn areas near the ends of travel and scraped until I brought flat on each dovetail face down until the "belly" that was worn in disappeared. You don't want to spend much time scraping for bearing at this point, just make sure you have spots of bearing along the full length. In essence, I made each dovetail flat in what I believed to be the correct plane based on the minimally worn portion, getting things pretty close and then began to compare for parallelism as follows:

I measured across the pins plotting the variation at several points maybe 2" apart along the length of the knee. I then scraped to change the end to end parallelism of the dovetails that were previously made planar until I had no more than about .0002 variation along the roughly 24" of length.

Connelly's method is better than this as my method relied on the adjacent flat ways to be truly flat or they would influence the measurement with the dowel pins. Connelly makes a sort of wire frame that has something that rides on the top, presumably unworn surface and just holds a pin or rubbing block in under the dovetail against the face of the dovetail. The other end has an indicator attached that rides directly on about the midpoint of the dovetail face. Connelly's method assumes that the top surfaces of the dovetails which are unused in the case of my mill were factory flat....and mine weren't as verified by marking out.

The actual angle of the dovetails as you do this is somewhat less than critical as Mcgyver mentioned...as you will simply scrape the mate to match, but do be careful as you don't want it to vary along the length. Dovetails are hard to work with and you would do well to get the bed on something that will allow you to bring it up to waist height or higher and tip it such that the face of the dovetail is roughly horizontal. Odd positions tend to make for inadvertant uneven scraping.

I hope this helps and was not too confusing. Its very hard to describe some of this in words.


03-15-2007, 07:02 PM
Paul's describing the same thing that i meant with the ground shafting, shafting was what i had, but dowel pins are probably easier to come by (although they'd have to be big enough dia)


Paul makes a very good point about leveling the lathe, then again if the bed is worn down, is there a surface you have confidence in to level?

re the reference, I should have explained it needs to be dovetail reference. it doesn't matter whether its 61 or 59 degrees, but Paul is right has to be consistent. here are mine (they are perfect and no i didn't make them, a friend did). I've also seem camel backs that one side at 60. Come to think of it, I'm not sure how how you'd make the reference regarding keeping the consistent angle, easy way would be to have access to new machine in good condition. maybe a sine plate and indicator? geez would that be tedious. anyway it makes life a lot easier if you just plunk the one surface down on the horizontal and snuggle the angle surface with blue on it up the dovetail


so with the horizontals in the same plane (first order of business) and a reference with a consistent angle, you can first get one dovetail flat and at a consistent angle, then with an iteration of pins/mic and reference bring the second one in.

03-15-2007, 07:28 PM
I strongly recommend that the first thing you do is get Edward Connelly's book "Machine Tool Reconditioning", it is advertised in HSM. Connelly discusses the tools needed, how to make them, scraping, tolerances, etc., etc., and takes you through the process of 5 different machines.
To get an idea of what is involved, read "Reconditioning a Lathe-Revisited", starting in Sep/Oct '04 HSM. The lathe discussed has prismatic and flat bed ways, and dovetail ways on the cross slide and compound rest.