9x20 CNC conversion architecture questions

I converted a 9x20 and it did good work. However I eventually decided that it fundamentally wasn't a very good lathe and swapped it for a Holbrook Minor, 10x20 (which I also converted)
(As an aside, the difference between the two lathes of almost identical capacity can be seen in this photo which shows the 9x20 tailstock on the Holbrook bed)
photos.app.goo.gl/dZ8BMdKZeZhoju3T9
Anyway: The X axis is tricky, as there isn't much room for a ballscrew. I managed to squeeze an 8mm one in by separating the bearing functions: A needle bearing near the pulley to take the belt tension and a pair of angular contacts in the handwheel/micrometer area to take the thrust forces.
I mounted the X servo under the saddle with the shaft pointing out towards the operator. I don't know if your servos are short enough to do that?
(Incidentally, the setup on the Holbrook is similar and I actually shortened a servo to make it fit)

The Z is, in contrast, very easy indeed. But don't do what I did and try to economise on the ballscrew length; allow enough length to fit a spiral-spring cover over the screw. You probably want some gearing between the servo and the screw, such as a toothed belt, to get the servo speed up a bit.

The spindle can be configured to swap between position and speed mode, but you might be disappointed with the stiffness in position mode if you also have adequate spindle speed. I would aim for at least 2000rpm spindle max speed (and even my 1960s Holbrook was capable of 3000). You might even want to look at some sort of 2-speed arrangement. (if your lathe is a variant with a back-gear then it's probably worth keeping it)
I think that some (most?) of the commercial mill-turn machines actually drop-in an indexing motor into mesh with a worm wheel for C-axis work.

I retained the manual controls on my lathe, and you might as well. But I never once used them and so didn't bother with the Holbrook conversion (but that does have a pair of MPGs on the saddle, and I definitely like that. In fact is also has a small push-button encoder to select manual mode and to choose the jog increment too.

Some pictures of mine here:
photos.app.goo.gl/nwM9Ze59j3zi5TB77

Also, a video of it dong work with my lathe macros, which turned out to be _far_ more useful than the manual controls for quick no-coding jobs:

RotarySMP wrote:

My suggestion to anyone wanting a CNC machine is to start with CNC machine. Used industrial CNC machines are often very cheap, and are engineered with enough space for decent stiffness ball screws in each axis. 

Good points, but there are sometimes reasons to convert a manual machine. 
It rather depends on the type of work you expect to do. If it is one-off projects with hand-working and wierd workholding / fixturing (classic model engineering or prototyping work, for example) then the better access available with a conventionally laid-out manual lathe can be advantageous. 

There are CNC lathes with a conventional layout (Schaublin, obviously. At a very small size the Denford Orac, and the some of the Hardinge ones. (though be wary of the spindle nose fitting on the Hardinge, some are collet-biased with a rather odd taper and peg for chucks and faceplates) 

CNC lathes do also tend to be rather bulky, with cabinets and full-enclosure coolant guards. These can be removed / shrunk but compact size rarely seems to be an important criterion in CNC lathe design. 

That said, I would suggest going ahead with your conversion, just be aware that it might not be your last lathe project.