Spindle registration faces, and mounting threaded accessories

[LanceC]

I have a question about how chucks, faceplates etc. lock/register against the spindle. I've searched and searched and not found this addressed anywhere explicitly. Perhaps it's obvious to everyone else and I'm the only one struggling to get it.

After a bunch of reading and extrapolation, I'm going to offer my understanding below, and would welcome corrections from others who are "in the know".

My understanding is that the threads of the spindle and faceplate (could be a chuck too) are not precise/fine enough to to align the axes (centre of rotation) of the parts to a sufficient degree. To overcome this the spindle must provide a registration face perpendicular to its axis. The faceplate too, must provide a such a registration face perpendicular to its own axis. If when the faceplate is mounted on the spindle, their respective registration faces are seated against one another, the two axes will be aligned.

The above being correct, wood lathes appear to adopt one of two systems. Some spindles provide a shoulder at the back of the thread, whilst others use the face (end) of the of the spindle itself. Please see photos below. At left, a spindle with a shoulder, and at right, my Woodfast without a shoulder.

476363 476364

It would seem to me then that not all faceplates and chucks would be compatible with both types of lathe spindles.

Now what led me down the murky path to get to this topic. I noticed that both my faceplate and chuck visibly wobbled when mounted, yet my spindle runs true. Over the course of many evenings of reading/trying something new iterations, I think I finally diagnosed the issue (against my developed assumptions above), they both want to register on their backs (against a shoulder). The faceplate because the "nut" is too deep and hits the back before the front engages, and the chuck because it has no face with which to register against the front of the spindle. Now I can machine down the faceplate nut simply enough, but the chuck is a little more complicated.

In conclusion, is my understanding correct, and if so, how would I use my rear registering accessories on my non shouldered spindle?

Any thoughts would be most welcome.

Oh, I should point out that I can still use them as is, but would prefer to get my alignment issues sorted out.

[Skew ChiDAMN!!]

That's pretty much my take on the subject. FWIW, to my mind a shoulder is a more reliable registration than the end of the spindle.

One possible fix would be to thread on a narrow lock-nut first. In use it wouldn't be an instant fix; you'd need to mount the chuck, spin up the lathe and check for wobble. Stop the lathe, loosen the nut, move the chuck slightly, tighten the nut.. rinse, repeat. With no indicator of how far to adjust, etc. beyond blind luck and the chuck running true on spinning up.

If you were doing mainly small, delicate chuck mounted finials, etc. then it may be worth the effort. But probably not.

I guess you could also try to set up two lock-nuts and do the same process with them, twiddling and fiddling until, together, they gave you a reliable shoulder for the chuck to register against... but I'm betting that the odds are good that sooner or later one'll come loose when removing the chuck and you'd need to do the whole laborious process all over again.

I hate to say it, but if I was doing any job that needed accurate chuck mounting (eg. multiple re-chucking - lidded boxes maybe? - or batching in stages for a production run) then I'd be looking for a lathe replacement.

On t'other hand, for jobs where the piece is mounted once or twice and finished in basically one session then it's really not such a big problem and I wouldn't bother trying to fix it.

[malb]

From a mechanical aspect, the shoulder behind the thread has a lot going for it in that it is larger in surface area (more contact area), further back from the tool zone (less likely to be damaged and then induce runout), typically closer to the headstock bearing (less overhang and hence less spindle runout) and there is more metal in the area than there would be at the open end of the thread assuming that there is a Morse taper in the spindle, so the spindle would be more rigid at a shoulder than at the tip.

I can understand the concept of using the face of the spindle to register against, provided that the spindle and accessories have well machined registration faces etc. However there will always be less contact area for the same sized spindle thread, and verifying the state of the registration face recessed into the accessory is harder due to difficulty in getting common measuring tools down to the face. Continuous heavy duty work beyond that envisaged by the designers can create wear and deformation on the contact faces and reduce the contact area further.

[powderpost]

Lance, There should be a fibre dust shield, spacer and a nut on the spindle nose. The thread on the spindle nose extends about 50mm measured from the front of the casting. There is about 32mm of thread in front of the nut. The dust cover is held in place with a circlip, but does not rotate with the shaft. The back of the face plate or chuck rides firmly against the nut. Here is a photo of the set up on my Woodfast lathe (1981 model).

Jim

476390

[LanceC]

Jim,

Thanks for the photo and your explanation.

On mine, I don't have a dust shield, just the sealed bearings then the circlip. I also have the nut, which was removed at the time of that photo. The only surface my nut has to register against is the bearing inner race. If your dust shield does not rotate, what does the nut register against on yours? On Artful Bodger's lathe, an earlier model which uses unsealed bearings and has the grease nipples, he has a metal plate screwed across the front. I wonder what his nut registers against.

In tracking down the issue, I measured the radial runout of just the bearing, then bearing plus nut. My understanding of the readings show 0.17 mm on the inner race and 0.29 mm on the nut. For reference, the inner wall of the spindle shows 0.09 mm, but given how pitted it is from past rust, I expect it averages better. I only mention that as I don't think that axial runout is contributing to the radial measurements.

I think my first port of call will be to machine the nut to ensure that the front and back faces are parallel so that it's not compounding issues.

476399 476400 476401

[powderpost]

Lance,
Had anther look at the nut on the spindle nose, without taking the nut off, it appears that either the nut has been mac0hined,or the spacer rides against the inner part of the bearing and the nut is only about 12mm thick. I am reluctant to pull the nut off, (if it aint broke, don't fix it)
The other Woodfast lathe has a chrome plate behind the nut, so I don't know the set up there.

Jim

[Richard Hodsdon]

On the Nova DVR , 3000 and 1624 lathes the manual states that the register is the bearing in the head stock and that
"Correct fitting of chuck to lathe spindle is important to ensure accuracy. The chuck body must contact an accurate shoulder on the lathe spindle or bearings (as on Teknatool DVR & 3000) to ensure chuck will run true "

Extract from the Teknatool manual for the Supernova chuck

[LanceC]

Thanks Jim and Richard for your input.

I took ten minutes and seated my bearings hard against the circlip, which I assumed would be purpendicular to the spindle axis, then snugged the nut up to the bearing inner race. And now the bearing race shows about 0.0015 ” lateral deviation, and the nut face 0.002”, so a significant improvement.

Now to turn some more logs into bowl-esque shapes.