Testing alignment on a PM1440HD

[Larry$]

I have never seriously checked the alignment of my lathe. I ordered an MT3 test bar. India made. Claimed .0002 precision.
First trick was to see how good the test bar was. I have a Mitutoyo dial test indicator with marks @ .0005". I cleaned everything, put the bar in the tailstock, mounted the indicator on a Noga mag base on the cross slide. I traversed the bar @ 90 degree increments. Top, both sides, & bottom using the carriage. About 9" of travel. Best I can tell the bar is with in about .00025." That requires some estimating and good eyes. The tailstock & quill were locked. Squinting I get about .0003" of variation with the biggest part near the tail stock. I tested going off both sides of the bar. Running a long the top & bottom show a droop of about .0005" Firm finger pressure will move it about .001" Slack check of the cross slide gives about .001 when exerting considerable hand pressure. When I extend the quill 100mm I get less than .0005" deviation on the sides & top, quill locked after moving. Unlocking the quill I can get about .002" side to side. Considering the skill of the operator and all the sources of variation, it passes.

Head stock alignment: Insert 5 to 3 MT sleeve & test bar. Following similar procedures the test bar angles about .0035" toward the operator at 9" from the spindle. Rotational runout is .0004 near the spindle and .0007 at the far end of the bar, about 10" out. The .0004 can be accounted for by the assumed .0002 of the bar, plus having a MT sleeve in the spindle, plus any variation in the spindle and bearings. The only thing that sort of bothers me is the .0035 angle between the bed carriage travel and the spindle. Over a 9" travel that isn't a lot of an angle and wouldn't affect the work if it was left in the chuck for all operations. If anyone has managed to get this far I'd like to hear what you think.

 

[benmychree]

A deviation of .0035 over that distance would indicate about .007 taper in a cut of that length, it sounds like you need to do some alignment, taking the twist out of the bed; this is best done by the two collar method, although an accurate sensitive level with .0005 PF per graduation will get you there also, the level should read exactly the same at both ends of the bed ways.

 

[Larry$]

I've done the thing with a very accurate level. Bed is as true as it can be by that method. What this looks like to me is the head stock is at an angle to the bed. Turning between centers I don't get a taper.

 

[benmychree]

I've done the thing with a very accurate level. Bed is as true as it can be by that method. What this looks like to me is the head stock is at an angle to the bed. Turning between centers I don't get a taper.

How accurate is "very" accurate? A .005 grad level is not good enough for lathe leveling. I suggest using the two collar method, turning between centers is not a measure of alignment, taper can be adjusted out by offsetting the tailstock.

 

[ddickey]

Rotate the test bar by 90° and try again. Check all 4 90's and see how it repeats. I bet that 3.5 thou does not repeat.

 

[davidpbest]

I agree with the above comments/suggestions. Here is a photo sequence that illustrates the method I used to align the headstock on my PM-1340GT (each photo has a description below the image with details). Mark Jacobs discusses lathe bed leveling on this post, and recommends affordable high precision levels suitable for the job on this post.

 

[xr650rRider]

I've got one of those MT3 test bars. Using a .0001" DTI, I checked the spindle taper and moving up the taper with lathe turning, I detect < .0001" of movement of needle. Clean the spindle taper, clean MT5 to MT3 adapter, insert adapter and test run-out of adapter, again around .0001" of detectable run-out. Clean inside of adapter, clean MT3 test bar. Insert test bar and on far end, have as much as .004" run-out, down to .0003" run-out, depending on where the test bar is inserted into adapter. When I got down to .0003" and got tired of knocking it loose, rotating slightly and trying again, I marked the adapter and test bar with a scribe and aligned those marks with existing match mark on spindle. So while the bar may be true, doesn't mean the tapers are exact to each other.

For tailstock alignment, I installed collet chuck and used a piece of 1" OD stock and turned a 60 degree dead center using compound set on 30 degrees. Inserted a dead center in tailstock (in spindle it has about .0003" runout), then put MT3 test bar between the centers and you can traverse adjusting tailstock until you get down to the run-out you can live with. Not sure I would like the alignment with tailstock so close to spindle using the collar as in post above?

 

[xr650rRider]

To add to my post above, I also have a MT5 test bar. It exhibited as much as .004" run-out when installed in spindle and unsupported on tailstock end. I could use it in one position and traverse and get headstock alignment pretty close, but when I turned it, it had .004" run-out on unsupported end. I didn't do an exhaustive knock it loose and rotate as I had done with MT3 test bar (learned this after what is about to occur). What I did do, was install 3 jaw chuck and using a 12" piece of 1.5" OD aluminum stock, prepared to make some test cuts. I took off just enough that I had full cut around the piece, then took a skim pass of .0005" on diameter. It was cutting same amount over 10", so I measured and it was exact same diameter along it's length using Mitutoyo micrometer out to 5 decimal places. I thought that was close enough. Then got to thinking about test bar and since spindle was running that true, thought a skim pass would have it matching my taper and could get it down to perfection with my spindle. That's when I discovered how hard it is to turn a hardened bar without chatter no matter what carbide or HSS steel cutters you own. Within 10 minutes had pretty much ruined a test bar. So when the new one gets here, I'll check it for run-out and try different positions in spindle if it does.

Should also mention that all the above was at least 4th time checking and adjusting alignment and level. Each time loosening head stock bolts turns into several hour iterative process. I've got one of the Shar's 12" precision levels (zeroed on granite slab) and have my PM-1340GT nearly perfectly level side to side and lengthwise on both ends.

 

[Larry$]

More fooling around: conclusion, head stock is off relative to the bed. For those that think my level isn't up to their standards, this is what I have: 12" VIS brand, made in Poland. Repeatable at 10" to 0.0005" (claimed) that's about 10 arch seconds. It has plastic finger locations to prevent hand heat from causing issues. Appears to have been hand scraped after grinding. The bubble takes awhile to settle each time it is moved. I checked the front to back of the bed by using a set of 123 blocks that are as close as I can measure. I maintained relative positions of everything for each try. The lathe had moved slightly since last winter when I last checked it. It sits over a seam in the concrete, not ideal but that's the choice. So I now have it set as close as I can in the front to back direction. It isn't perfectly level side to side, irrelevant!

I checked the headstock sleeve MT5-3 by clocking it 90 degrees 4 times leaving the test bar in it. Then did same but only clocking the test bar, not the taper sleeve. Very similar results were obtained for all 8 tries. About .0035+ over 9" of travel with the indicator mounted on the cross slide. Only conclusion I come up with is the headstock is out of parallel with the bed.

Not looking forward to trying to align the headstock. Instructions are in the Operation Manual. Need to get an 8" long x 2" bar of steel.
Headstock is held by 4 M12 bolts, pivots around a 16 x 40 pin, has an alignment block with opposing screws.

 

[xr650rRider]

Just received my new MT5 test bar and knocking it out of MT5 spindle and rotating it every 10 degrees and reseating it. There is definitely a sweet spot on runout. Had as much as .012" with .0006" being about average until I got to 1 spot just a few degrees wide that measured .00015", same on both ends over 12" span. Knocked it loose, removed, cleaned and reseated and had about .0003". Reseated 2 more times and back to .00015" runout. Marked it with a scribe to line up with spindle match mark. Definitely worth testing if you have a sweet spot.