* GadgetBuilder.com *
This page is for those who fail to recognize that a Homier is not a Hardinge and insist on tweaking their mini-lathe for improved accuracy, especially if this accuracy is beyond anything actually needed in practice. Surprisingly, the methods needed to improve a lathe's alignment are simple and the tools required are generally already available in the home shop. In fact, my little lathe as delivered was amazingly accurate, easily accurate enough for most uses; the only significant error in my minilathe concerned the tailstock. Read on for details of converting a minilathe from a Homier to a Hominge.
There are four major parts to aligning the mini-lathe:
1. Align the tailstock parallel to the ways
2. Align the spindle parallel to the ways
3. Align the tailstock to the spindle horizontally and vertically
4. Align the cross slide perpendicular to the ways (not covered here)
What follows is an outline of how I went about aligning my minilathe. I did not proceed in the optimum order so read this complete section before deciding how to proceed with your machine and its unique alignment errors. I'm still learning but I believe the order given above is a better way to proceed than the one I muddled through. Overall alignment takes a while but it isn't an all or nothing thing -- do it one piece at a time and do the next part when time permits. To avoid surprises, correct the horizontal alignment of the tailstock to the headstock after each session since most every correction affects this; it is quick and easy to keep this correct, especially if you fit an alignment bar.
Prior to aligning anything, the carriage "V" should be checked for "ridges" since this can adversely affect alignment of the carriage with the prism and the prism (via the carriage) is the main reference for alignment.
Minilathes are inspected prior to leaving the factory and some vendors include an inspection report with the lathe. This sample report shows the standards this vendor required be met; the procedure given below can improve accuracy considerably ("Radial pulsation" is also known as runout; error values in the report are in mm). Note that Item 5 in the report requires the tailstock be higher than the headstock by 0.03mm to 0.08mm(about 1 to 3 thou).
Some machinists turn a test bar to align their lathe, follow the link on the Wrathall site. The J-Latta variation of the two collars method works well according to that site (I haven't tried it); a 2" micrometer is needed to get full accuracy from this technique. If the test bar is bumped accidentally while installing shims then the test bar may need to be re-trued for which the 7x12 must be reassembled (motor, motor controller, etc.) or you could use RDM (see below)to accommodate any modest eccentricity introduced. J-Latta's underlying measurements are similar to RDM but the test bar is turned down to run true and be identical size at the two measurement points. Some time is required to true the test bar and get the collars to identical diameter, where RDM uses calculations to remove these artifacts from the measured values without machining the test bar. There are variations of the two collar method that work by re-truing both collars without moving the cross slide, then evaluating the difference in diameters with a micrometer; these variations are difficult to use on the 7x12 when adding shims because multiple trials are generally necessary to select the correct shims and the lathe must be reassembled (motor, controller, etc.) to re-true the bar for each trial. Plus the bar must be re-trued when fine tuning the alignment via headstock bolt torque. RDM can be used to evaluate the alignment of a lathe which cannot be powered up, a major advantage in some circumstances.
I used Rollie's Dad's Method (the RDM paper is also found in the 7x12 Yahoo Group's files) of aligning a lathe's spindle axis to the ways but stacked the deck to simplify its application. Study and compare the original RDM document to the less general scheme outlined below to understand how and why the "simplification" works.
At right is my setup: a piece of 1/2" drill rod extending 10.5" from the chuck, the compound removed and a DTI mounted in its place (a toolpost mounted indicator works nicely too). In order to achieve 10" between the left and right measurement points (not really necessary) I removed the threading dial. The bar holding the DTI is set parallel to the test rod and oriented so that the DTI finger can be easily adjusted in height without changing the horizontal position. The horizontal position of the DTI is adjusted via the cross slide.
NOTE: The 1/2" x 10" steel test bar is not rigid -- it bends slightly under its own weight. The error due to sag of a 1/2" test bar is small so if your goal is simply to eliminate taper due to a headstock/bed alignment error you can safely ignore this error. I calculate that my test bar sags about 7 tenths at the far end. This won't affect horizontal measurements noticeably but it will affect vertical measurements, i.e. the headstock will appear to tilt down by about this amount in addition to any actual tilt. This error from sag can be reduced by using a larger diameter test rod and/or a hollow test rod (pipe). I used the formula for a uniformly loaded cantilever beam found here to calculate sag.
To stack the deck: 1) Use an accurately ground test rod (drill rod, a shaft from a shock absorber, or a shaft from a line printer) to eliminate variation in rod diameter from the RDM equation, 2) Use a straight test rod and adjust chucking so the far end has minimal runout (the DTI is already set up) - this so the alignment error is within the DTI's range, and 3) Adjust the DTI at the chuck end to deflect equally above and below zero (making this average zero) to eliminate this term from the RDM equation.
If this is your first trial of RDM it may be helpful to print John Wasser's RDM page plus copy the simplified version above into WordPad and print it for reference.
Once you muddle through the above, it becomes clear that with your trusty test rod and DTI you can measure a lathe's vertical or horizontal axis error in under a minute -- something I puzzled over for a long time without getting anywhere prior to discovering the RDM article. If you don't have an accurately ground test bar and a DTI you can still make the RDM measurement by working from the original article, it just takes a few minutes longer.
Using the above approach I made measurements along the top and front of the test rod. The spindle was slightly off line, where the averaged rod position 10" from the chuck was 1.3 thou closer to the operator and 0.4 thou higher -- excellent for this inexpensive machine but clearly not in the Hardinge category. I heard from a site visitor whose errors were over 10 times as large as mine so not all mini-lathes have good accuracy out of the box.
After determining my spindle alignment error, I removed parts as needed until the headstock was off; the shafts and shift fork were lubricated but the gears were not -- my opinion is that plastic gears don't need lube. It took over an hour of trial and error work using aluminum foil as shims to improve the alignment; the headstock was removed and installed at least 6 times while selecting the best shim arrangement. The spindle now measures 0.3 thou high and 0.2 thou forward at the end of the rod 10" away from the chuck -- as good as I could do because the tightening torque on the headstock retention bolts causes changes larger than the observed errors plus there is some "noise" in the DTI reading from the test rod's surface finish.
A recurring misconception about RDM is that the accuracy of the chuck or the way the rod is held by the chuck (angle/offset) affects the result. In fact, RDM removes these errors by averaging them out as part of the calculation. The simplifications in the application of RDM suggested above may make it hard to see how this happens; refer to John Wasser's original document to gain a clearer understanding of this elegant technique. The original document suggests shimming the feet of the lathe to correct alignment -- this won't work for the 7x12 because it sits on rubber feet and isn't bolted to a stable surface so you must shim the headstock instead. RDM does assume that the prism which guides the carriage along the ways is absolutely straight...
Inspect the contact area between headstock and bed, removing any burrs or defects which might prevent good contact. Chamfer the holes for the headstock bolts slightly where they mate to the bed and also the holes in the bed where they mate to the headstock; burrs here can affect the alignment.
I found the mating part of the headstock was deliberately (albeit crudely) ground so the part which fits over the prism makes contact at areas near the left and right ends of the headstock, plus there is a third contact point on the rear flat part of the ways. There is a bolt near each of these contact points. This clever design makes it reasonably straight forward to adjust.
Apply machinist blue or Sharpie pen ink to the headstock where it mates to the prism. Carefully place the headstock in position and slide it back and forth on the prism about 1/8", then remove it. The blue on the prism and the clean points in the headstock V will indicate the contact points. Verify that the contact points are near the bolt holes, preferably toward the ends of the headstock from the respective bolt holes.
The shims to shift the horizontal or vertical alignment should be placed at the indicated points. To adjust the horizontal alignment, apply a shim to one 45 part of the prism near one hole and to the opposite 45 on the other hole (I extended each shim down the side of the prism to the horizontal part of the bed and held it in place with a dab of grease). I used folded aluminum foil for a shim so it may have compressed a bit; I used one 3 thou and the other 2 thou to shift my horizontal alignment by 0.0011 at 10" -- this may help in estimating the shims needed.
To adjust the vertical alignment, apply a shim to both sides of the 45 on either the chuck side contact point (to angle the chuck upwards) or apply a shim to both sides of the gear side contact point (to angle the chuck downwards).
As the 3 headstock bolts are tightened, torque can be used to "fine tune" the last half thousandth or so measured 10" from the chuck.
I used a 22 inch long piece of 1/2" drill rod I had on hand, passing it through the chuck until 10.5" protruded, then tightening the chuck. In retrospect, this was a poor way to proceed because the rod picked up some small scratches in the process of passing it through the chuck; these scratches affect the DTI reading by causing jumps of up to 1/2 thou as the rod rotates. I should have cut off a foot of drill rod, dedicated it to RDM, not passed it through the chuck, and preserved it in a velvet lined box when not in use. Rollie's dad owns a car repair shop and uses old shock absorber rods -- better than drill rod because they're hardened and tough to scratch (and they're free!). Enco has hardened and ground rod but keeps it hidden as 505-3290 -- this rod is VERY scratch resistant but more expensive than drill rod.
Prior to aligning the tailstock, check the tailstock "V" for "ridges" since these can affect repeatability of tailstock error measurements as well as tailstock alignment. Verify contact between the tailstock base and the lathe bed, adapting the ink approach outlined for the headstock in Practical Details above. Also see my tailstock notes for more on the difficulties involved in precise tailstock alignment.
The tailstock's angular alignment to the ways should be corrected prior to adjusting tailstock height and therefore prior to adjusting the headstock. Improving the angular alignment is quick but getting it near perfect takes time and fiddling because the results of shimming aren't completely repeatable, probably because things seat differently each time. To evaluate the angular alignment of the tailstock, extend the ram and lock it (to take up any slack). Mount the DTI on the carriage and indicate along the top of the ram to measure vertical alignment. Indicate along the front of the ram to measure horizontal alignment. My tailstock's angular alignment seemed fine initially but eventually I found the ram was pointing up and toward the front by 2 thou per 2 inches in each direction. Shims were installed as shown, a 3 thou shim across the rear of the tailstock to base joint and a 5 thou shim (about 1/2" long) on the front, forward of the center ridge in the base (I left the ends exposed for the picture). Shimming the tailstock is much easier than shimming the headstock: remove the tailstock from the lathe, loosen the bottom socket head screw and the rear screw a couple turns -- the base and top will separate enough so shims can be slid in. Re-tighten the rear screw and tap the base a couple times with a rubber mallet to ensure things are solidly seated, tighten the offset adjustment bar if fitted, tap again, tighten the socket head in the bottom, tap again. Install, check, and repeat as needed.
Once the angular alignment of the tailstock is correct then the angular alignment of the spindle should be completed prior to aligning the tailstock to the spindle horizontally and vertically as described below.
Mini-lathes are deliberately manufactured with the tailstock slightly higher than the headstock (see item 5 in this inspection report) to facilitate subsequent alignment. The implication of this didn't register with me initially so I thought I would need a mill to lower my tailstock by 2.5 thousandths. I reassembled my lathe after aligning the spindle and continued to ponder how to align the tailstock height.
Finally, the light came on: raising the headstock is equivalent to lowering the tailstock since the goal is to set them equal. I had to loosen the headstock again (not optimum) to correct the tailstock height error. Shimming the headstock is faster and easier than lapping the tailstock to match the headstock height; future wear may lower the tailstock which can be compensated by exchanging this shim for a thinner shim. My tailstock measured 0.0025 high so I added a 5 thou shim on the rear headstock support point (without disturbing the shims previously installed on the V), raising the chuck (in the center of the headstock) by half the thickness of this shim. This worked as anticipated: the height of the tailstock is now about 0.3 thou high while spindle angular alignment was unaffected. (Paper shims work but are not the best because they can be difficult to remove if re-shimming is needed later, best to use real shim stock or even pieces trimmed from a soda can.)
Tailstock offset was measured by holding the mounting stud of the DTI in the chuck, then indicating on a dead center in the tailstock (easier and more accurate than indicating the inside of the tailstock taper). The only difficulty with this technique is that a mirror is needed to read the DTI while the dial is facing down. The tailstock height error is half the difference between the indications taken at 12 o'clock and 6 o'clock; verify that the tailstock is higher than the headstock before proceeding to shim the headstock, of course. If the tailstock is lower than the spindle then the tailstock must be shimmed instead of the headstock - place equal thickness shims on both sides of the center section of the tailstock base to raise the tailstock.
While the DTI is set up, verify the horizontal alignment of the tailstock by taking readings at 9 o'clock and 3 o'clock. Any error here can be corrected using the adjustments provided. It can be tricky to adjust the tailstock because you must rely on the upper screw (under the crank) to hold the horizontal adjustment while you remove the tailstock to access the locking screw underneath. Then, the adjustment may shift slightly as you tighten the lock screw so it often takes a bit of trial and error -- I eventually added a bar and use shims to simplify accurately centering the tailstock horizontally. Theoretically the angular shims shouldn't interact with the horizontal shims -- but they do so if you change any tailstock shim check both angular and offset errors. This interaction may be due to seating errors since it doesn't seem completely repeatable or predictable -- I have about 0.0001 error per inch in residual angular error because this unpredictability prevents getting all settings to zero simultaneously.
My tailstock exhibited repeatability errors, e.g. the horizontal error would vary by up to 2 thou each time the camlock was released, the tailstock moved away and then back and the camlock re-locked. This bedeviled me for a long while before, while working on the carriage gibs, I found (and removed) a small ridge in the tailstock "V", probably caused by wear.
Some are surprised that DTI chucking errors (runout or anglular) don't affect aligning the tailstock to the headstock with this method. Note that the DTI's finger will rotate about the spindle's center of rotation regardless of how it is chucked; the measured error is the offset between this center of rotation and the tailstock's center. You can demonstrate that chucking errors don't matter by measuring the tailstock offset with this method, then offset the DTI's mounting stud by inserting a shim between the stud and one jaw and measure the tailstock offset again. With a thick shim you may have to adjust the angle of the DTI's finger but the offset will measure the same -- try it: the difference between the two readings (9 and 3 oclock) will be unchanged by addition of the shim.
Prior to alignment my lathe did not drill well from the tailstock -- the center drill would scribe a small circle when brought slowly into contact with work in the chuck, with increased pressure drilling would start but the drill would wobble; there was also a "scrapey" feel when drilling deep holes. Following alignment the drill starts nicely without wobble and is noticeably smoother in deep holes.
The tailstock is more difficult to align than the headstock (although access for shimming is easier). Measuring the headstock error and compensating for errors arising from chucking are handled nicely by RDM, assuming a good quality test bar. There are more sources of error/variability when aligning and/or using the tailstock including: clamping the tailstock to the ways, ram/taper concentricity, and chucking variability.
The method I used for tailstock angular alignment is flawed in that it assumes the exterior of the ram is accurately concentric with the interior taper. This alignment in my tailstock has an error of 1.5 thou per inch = .086 degree (which seems pretty precise when stated in degrees). I made a lever drilling attachment and while evaluating its accuracy ran into the error in the tailstock's taper alignment (made obvious by the increased distance from the taper to the end of the drill bit). Oddly, removing the rear shim inserted while aligning the ram to the ways aligned the taper correctly; hard to believe this was done deliberately during manufacture, but maybe.
It is instructive to check the tailstock chuck, indicating on a hardened and ground pin held in this chuck. This checks the overall error in the tailstock plus the arbor and the chuck itself -- my HF chuck isn't a precision gripping device although the basic alignment of chuck and arbor is good. Each time the hardened pin is gripped in this tailstock chuck the offset changes by up to 3 thou although the angular error isn't affected. My Jacobs Superchuck has less variation in offset, typically being under 1 thou. (An inexpensive source for a suitable hardened pin is the shaft from a discarded VCR's head motor.)
If the goal is to improve drilling performance from the tailstock then aligning the tailstock based on measurements on the ram leaves ram/taper concentricity and chuck idiosyncracies uncompensated. Using a hardened pin held in the chuck for alignment works but variability in how the pin is gripped by the chuck can add considerable error and is difficult to determine. Setting horizontal and vertical offset by measuring on the face of a dead center seems to be reasonably repeatable. One scheme would be to make h/v offset measurements with the ram retracted and extended, then take the difference to find the angular errors -- I haven't tried this (yet) but it suffers from being more complicated to explain and taking longer to make the measurements.
Reducing tailstock error is quick and easy and the improvement in operation is easily seen. Diminishing returns set in when attempting to achieve errors below a thou or so. The DTI is a double edged sword allowing measuring and correcting alignment errors but also showing residual and repeatability errors remaining after one's best effort. Perhaps it is best to be happy with the low hanging fruit :-)
"Carefully adjust the vertical position of the DTI for maximum deflection (center of the test rod's side)" seems easy enough to do -- and it is easy if you use a DTI mounted on a QCTP because then the tool height adjustment moves the DTI vertically. With the DTI on a mag base, ensuring the DTI moves vertically is more difficult -- the plane of movement is usually inclined slightly. Getting the DTI vertical position accurately on the center of the test bar based on needle deflection assumes the DTI moves vertically. Make this movement as near vertical as possible to minimize small errors that are difficult to analyze and quantify.
To check whether the cross slide is perpendicular to the ways, mount an indicator on the CS either on the toolpost or using a magnetic base. Remove the chuck jaws and touch the face of the chuck near the periphery of the chuck face and at center height, i.e. on the near side of the chuck and set the indicator to zero. Lock the carriage so it won't move inadvertently as you move the CS. Mark the point touched with a Sharpie, then rotate the chuck so the marked point is on the far side and at the same height(you'll have to help the indicator finger across the chuck slots). Advance the CS until the indicator again touches the marked spot on the chuck. Any difference in reading indicates (pun) that CS travel is not perpendicular to the spindle axis. If the CS moves toward the headstock slightly as it is advanced then a faced surface will be slightly concave; if the CS moves away from the headstock as it is advanced then a faced surface will be convex. Generally a very slight concavity is desirable, perhaps a tenth or two per inch. In practice, this measurement is helpful but not completely reliable: often tool pressure increases as the tool approaches the center while facing (because sfpm drops) and this tends to make the result convex, depending on the work material, RPM, slide loosness, tool sharpness, depth of cut, etc. Bottom line: if it's close it's probably OK.
Correcting the CS travel to make it perpendicular requires modifying the CS dovetail (or the saddle "V" way) slightly which is beyond the scope of this document.
If you have a comment on my site or its contents or would just like to chat, click here.
This page was by John Moran, tyro machinist and HTML tweaker.
