A New Ballscrew for the Z-axis; or
How to simulate a cylindrical grinder with a lathe

5 Bears Home Homebrew CNC bench mill

Please Wait
for the large page to load...
Tools Required:
  • Lathe
  • Adjustable 3/6 jaw or 4 jaw chuck
  • A quality micrometer with 0.0001" resolution.
  • Assortment of 3M Wet/Dry paper, 220, 320, 480, 600, 800 grits
  • Steel flats

This is a necessarily long installment because it contains some good information, and will hopefully help a lot of guys with their own projects. In it, I will detail how to take a hardened and ground ballscrew, chop it to a shorter length, and finish it for a ball-bearing journal to standards as good as grinding, all with a lathe and some basic machinist's tools.

While I was messing about with the column as layed out in the previous installment, I became more and more worried about both the length and size of the 12mm THK ballscrew. I was asking what is basically an instrument ballscrew with a single circuit nut to both hold and translate vertically a spindle, and do so under cutting pressures. The same type of ballscrew will be moving the table on the X and Y axes, but they won't be fighting gravity to do so!

I did have another top-quality ballscrew set aside; it is an NSK 16mm dia x 5 mm pitch, with a dual-circuit nut. Note the two return tubes on the NSK ballnut in the foreground, next to the THK 12mm ballscrew. This 16mm ballscrew is a significantly stronger setup. The only problem was the length. It was 6 to 8 inches too long.

I had cut and finished rolled, cheap ballscrews before, and it was always a bit intimidating, because one rarely knows what kind of steel you are cutting, the hardness level, how it will behave, etc. And to chop and turn this beautiful ground ballscrew simply seemed like a crime. But I was honest enough with myself to realize that shortening it, and actually using it, was a better idea than leaving it in a box. So with trepidation, I applied a hacksaw with a high-speed blade to the appropriate spot, and found the teeth slid off the screw like it was sapphire. Uh oh! This is HARD steel.

The only way to cut it with any chance of success is in a chop saw, and that's just what I did, after wrapping the good portion of the ballscrew in plastic.

This is nothing more than a 10" miter saw with an abrasive blade. On hand is a water bottle to cool the ballscrew, and a hand diamond truing tool to refresh the cutoff wheel edge.

The chop is made in short increments, perhaps 8 to 10 "passes" of 5 seconds or so, to avoid heating the steel excessively. As the wheel is lifted, I squirted a bit of water into the cut to draw the heat. You want to avoid drawing the temper excessively, or even worse, causing an uneven heating which might warp the ballscrew. So basically its chop... squirt... sizzle... chop... again and again. Get through it without forcing it or trying to rip through quickly. Remember you are whacking a very expensive, precision assembly with a whirling abrasive disk.

After you are through, clean up the burr with a belt sander. Keep most of the ballscrew under plastic wrap; above all, avoid letting any abrasive dust get into the ballnut. Obviously, retain the ballnut on the side of the screw that you want to keep!!

If you haven't figured it out yet, we are going to turn a ball bearing journal for the end of the ballscrew opposite the "driven" end. This will require a precise fit of the turned ballscrew into the ID of the selected radial bearing. And the journal must be as close to 0 TIR relative to the working portion of the ballscrew as we can make it. Simply chucking it in a 3-jaw chuck will not work. It must be indicated to run dead-true relative to the axis of the lathe, so the turned journal will not run eccentric. Excess eccentricity will ruin the job.

You must use a 4-jaw chuck, an adjust-tru 3 or 6 jaw chuck, or a 16mm collet. Of the three, the first two are the better choice... more accuracy. Apply a .0001" test indicator to a ground steel held in a collet. You will see what I mean. The collet has runout, whereas the chucks can be tweaked to 0.

So just how the heck to you center a "threaded" ballscrew in the lathe? It is impossible to apply a DTI to the ballscrew because of the grooves. To correctly center requires a full rotation of the work to check the TIR.

Secret #1: I "invented" this method when I was working on rolled ballscrews. Try it with confidence, it works great. Chuck a delrin plastic round of 20mm or so, and bore it for 1" deep to an ID that is 0.001" smaller than the carefully measured OD of the ballscrew. So if the ballscrew mikes .6288", bore .6278".

  The bore is now .001" less than the OD of the ballscrew. Take a skim cut on the OD of the plastic. Don't skip this step! The OD and ID of the plastic must be concentric. Part off the plastic a "tube" of perhaps 0.75" long. The walls of the tube should be ~ 1 to 2 mm.

Grease the end of the ballscrew, and under strong hand pressure, slide the delrin plastic onto the end of the ballscrew. Being undersized, but thin-walled, it will deform slightly by expanding to fit the screw. It will also present to your DTI a perfect surface for centring the screw! If the plastic won't go on, make another, with a slightly greater ID. Likewise, if the pressure to install the tube is too light, make another a bit firmer.

Consider the tube sacrificial. Once on, it will probably need to be turned off. If it easily slides off with hand-power, it's probably not grabbing hard enough. Go ahead and install the ballscrew in the lathe, and using the plastic tube as the indicating surface, adjust the position for 0 runout, or as close as you can get. Be satisfied with nothing less than 0.0005" TIR.


Shift now your entire mindset for turning, and become the hyper-precision machinist you were meant to be. You must begin working in tenths here - double-check your work and get the measurements down to levels you probably never worked with before. We are simulating a grinding operation! Put away your calipers, which are worthless for diameter measurements of this class.

The whole idea is to avoid getting the journal too "skinny". Go as slow as you need to. Understand your machine. If it's a Monarch 10EE or a Hardinge, it should hold .0005" all day over such a short run, but perhaps with a smidge of taper. If your lathe is +/- 0.002", or has evil taper, then keep your cuts up a bit to take these into account. The goal is to get the ballscrew turned as closely as possible to ball bearing ID, and sneak up on the final dimension with abrasives.

The ballscrew was installed in my favorite chuck, a 5" Buck 6-jaw. To the left of this photo is the ballnut, snuggled right up against the chuck jaws. The ballnut is protected with a baggie and a lot of masking tape. The plastic indicator contact pad, and a 0.0001" clock, was used to get the ballscrew to within 0.0002" or so of dead truth. I couldn't get it better than that because the plastic was close to round, but no longer exactly round. Just do the best that you can.


Be sure your tailstock is dead-on, and adjust laterally if necessary. This journal will be 10mm OD by 10mm long, for a sealed, radial bearing of 10mmID, 28mmOD, 8mm thick.

With a lot of scary thoughts, I applied a HSS center-drill to the end of the shaft. To my great relief, it cut without too much difficulty. The core of the ballscrew is not hardened quite as much as the grooves, but it is still tough, stringy, high-carbon steel. This is another reason not to heat the metal too much during the chop operation. If you get it above transformation temp in local areas near the cut, and then apply water, you have just hardened it to some scary Rc number, and made it almost unturnable.

In this photo, I begin roughing with a TPG222 indexible carbide tool. Even with tailstock support, the lack of rake in this tool (like most indexible carbide tools) and the pressures, cause the work to spring away from the tool, up to .003" or more! Go ahead and hog, but remember this is not an accurate cut. Take it no further than final diameter of the journal, + 0.010" or 0.25mm.

Note in this photo the surface finish of the journal... not good.

It is time for another "secret". Secret #2: Every shop needs a bit like this for carbon steel finish turning. It has extreme top and side rakes for a steel-turning bit, and looks more like an aluminum-turning tool than one for steel. The tip has a miniscule radius, just a few swipes with a diamond hone to knock the burr. It is designed to cut no more than 0.004" or so per pass, no more. The rake, being steep, will allow it to cut with very low pressures. The work will not spring away, and being razor sharp, it does not skip, rub, or otherwise cause grief when taking these shaving cuts. If you have a DRO, you will find that it will easily cut with an infeed of 0.0002 or less. So we can really sneak up on our final dimension.

The radius, being small, will also generate the square shoulder at the root of the journal.

Put the tool to work, measure often and repeatedly. You can tell if your setup is accurate if your measurements are repeatable, i.e. mike the journal, write it down, mike it again. The two values should be identical. If not, your micrometer is suspect.

Turn the journal down to final diameter + 0.0007" or 0.02mm. Be especially careful on the last pass. It is easy to get rushed mentally and knock off too much metal. If you do "go small" by accident, consider a ball bearing with a smaller ID.

Note the finish delivered by this bit... better than the TPG, no rubs or skips.

Time to break out the wet/dry paper. In our case, we will use it wet with a bit of oil. Use some steel backing, like some small pieces of O1 stock, a blank lathe tool bit, anything flat.

By backing the paper with a flat, it is less likely to take the journal out of round, and by "leaning" on one edge or the other, we can correct (or create) any slight tapers in the journal. For example, this journal as finish-turned, tapered ~ 0.0001". Not bad, in fact excellent, but that is because the length of the turning was so short. If this was a journal for the driven end, rather than only 12mm long, it might be 20 to 30mm in length, and this can exaggerate any tapers which your lathe may produce. And as you will see, you want a slight taper from big to small, left to right.
To start: If your finish-turning was 0.001" oversized, start with 220 grit. 0.0008" = 320 grit. 0.0006" = 400, anything less go to 600 grit right away. With 320 grit paper, an application with modest pressure of 15 seconds will remove ~ 0.0002" With 600 paper, 15 seconds will remove about 0.0001". Remember, these are rules of thumb, and your results may vary. Simply realize that when we are dealing with 0.0001", it doesn't take much to remove this amount. You can go from bearing = uninstallable, to slop, in seconds. Grease or oil the fit before attempting to install the BB. An ungreased/dry interface will make you think the fit is tighter than it actually is.

Secret Trick#4: Intentinally induce a minor taper so that the end of the journal is smaller than the root. We want the BB to install partway, then stick. Look for a taper of ~0.0003"/in., induced with the abrasive paper.

Why? It is possible to remove too much at the root (left side in this pic), and have the bearing fit nicely on the far end, only to suddenly have it "pop" into a sloppy, loose area next to the base of the journal during installation.

Here, the BB slides with modest finger pressure as far as shown, then gently binds up. Don't force it, or it may be very difficult to remove! Measure the taper with careful manipulation of the micrometer. If you can't induce a taper, or simply don't want to, it is certainly optional, but just be careful and don't cut the journal undersized next to the working portion of the ballscrew.

As the average diameter approaches the target, switch to finer grits, ultimately using 800 to 2000 grit. 2000 grit paper (lacquer finish paper from auto-parts store) will really put a shine on the journal!

Be sure you apply a bevel on the end of the shaft to ease test fits. When you remove the BB after a test fit, you will see where it contacted the shaft by longitudinal, very minute, scoring.

With scissors, cut some tiny strips. Abandon the steel backing, and use the paper as shown, to reduce the diameter to the left of the scored areas. You also want to blend in the scoring, as by definition, that is where the BB is grabbing the journal.

Work the strip longitudinally a distance about the width of the strip itself, plenty of oil, back and forth...

The job is complete when the bearing can be installed with finger pressure, yet maintains a distinctive sliding friction all the way onto the shaft. There should be minimal "stick-slip", which is where the BB encounters varying diameters along the installation path. The BB should easily remain in place under its own weight when vertical, and the inner race should grab and be retained when the shaft is rotated, and the outer race held by hand.

Finish the final thread by hand with a good file, as it will be badly buggered and burred.

Whew! I hope this gives everyone a good idea on how to work a shaft without a grinder for an excellent fit. This is how all of the shafts and bearing journals were done for the turboprop, and they all came out nicely. The entire process is easier than it sounds. Give it a try!

When complete, all that I managed to do was remove about 6" of the shaft, but it was necessary, and now I have a strong 16mm ballscrew in place of the whimpy 12mm job. Anyone need a THK ballscrew? It's for sale!