How Stable/repeatable?

[kizmit99]

I'm thinking about building tooling fixture that will basically have a 2" diameter threaded steel rod (let's say 28tpi) that is threaded into a corresponding hole in a 3/4" steel plate, then 'locked' in place with a 1/4" lock-ring (finger tight)... If X-Y is the plane of the surface of the plate and Z is through the center line of the rod - if the point at the center of the face of the rod is what we're concerned with and we arbitrarily call that (0.000, 0.000, 0.000), assume the plate is fixed and cannot move, the lockring is loosened, the rod is rotated 14 turns, then the lockring re-snugged; what should be my expectation be for the end position of the original point? In a perfect world I would expect (0.000, 0.000, 0.500) - In the real-world, obviously it will be something different...

Obviously there are a ton of variables - what I'm really wondering about is how close to a line will the path of that point (the center of the face of the rod) follow? (I'm actually not even concerned with the ending Z position, just the X-Y errors introduced) I suspect the real world path would be something like a spiral on the surface of a cone. What I'm having difficulty imagining is just how far off from the perfect line I should expect it to move assuming reasonably fitted threads. Could I reasonably expect less than .001" 'drift' in X-Y? Or would .010" be more likely?

I will be turning both the ID and OD threads - I can make them whatever TPI makes sense, 28 just seemed like a reasonable number. I was planning on "normal" threads (Unified National I guess, not Acme, etc). I should be able to thread the first part, then use it for test-fitting while cutting the second, so should be able to (hopefully) get a fairly slop-less fit...

Anyway -- if anyone has any thoughts on how accurate I should expect something like this to be, thanks in advance for your thoughts...

 

[JimDawson]

Good question! It can't move more than the clearance in the threads, but if you had no clearance you couldn't screw the parts together. Maybe lapping the threads together would help, but I really don't know how to predict the outcome.

Please let us know how it works out.
.
.

 

[derf]

I would think that the thicker the threaded plate is, the straighter the rod will rotate. Think of it as being square, if the thickness of the threaded plate was equal to the the diameter of the threaded rod it has more thread engagement and less chance to lean.

 

[4GSR]

Put a shoulder on the threaded rod so "shoulders" as a flange against the plate. 28 TPI is awful fine thread for a 2" rod. 16 pitch thread would be more manageable on a 2" OD rod.

 

[Tony Wells]

If the -Z- position will change, or need to change, thicken your base plate to allow threads plus a close tolerance hole concentric with the thread, and smooth portion of your ram above the thread. You would need move the knurled locknut to the underneath or come up with another locking means. The fit between the bore and ram could be very close, on the order of 0.0001's. If you were concerned about debris interfering in that close fit (I would be) you could either dimension it so that at it's uppermost position, there is still engagement between the ram and the bore. Or use an 'O' ring or seal as a wiper.

 

[derf]

Ok, I'm dyin' to know.....what ya buildin'?

 

[kizmit99]

I would think that the thicker the threaded plate is, the straighter the rod will rotate.

I think that would only hold up to a point. If the threads are engaged the total "slop" in the joint would seem to be driven by the clearance of the threads themselves (as Jim pointed out above). More threads engaged would, I think, tend to "smooth out" any variances in the thread clearance, but after some point I don't think it would help anymore. My gut says 20 threads at 3/4" would be "enough" - but I'm just spitballing at this point...

Put a shoulder on the threaded rod so "shoulders" as a flange against the plate. 28 TPI is awful fine thread for a 2" rod. 16 pitch thread would be more manageable on a 2" OD rod.

A shoulder won't work for this application, because I want to be able to adjust the Z position arbitrarily. I just pulled 28TPI out of my... err.. out of the air. I was thinking 20 turns of thread would be enough to "even out" any variations in the tolerances on my threads. 16TPI would give me 12 full threads (in the 3/4 plate) - maybe good enough? maybe...

If the -Z- position will change, or need to change, thicken your base plate to allow threads plus a close tolerance hole concentric with the thread, and smooth portion of your ram above the thread. You would need move the knurled locknut to the underneath or some up with another locking means. The fit between the bore and ram could be very close, on the order of 0.0001's. If you were concerned about debris interfering in that close fit (I would be) you could either dimension it so that at it's uppermost position, there is still engagement between the ram and the bore. Or use an 'O' ring or seal as a wiper.

I was thinking about getting more travel in the Z-axis than the thickness of the plate, but about a half-inch total travel is probably 'good enough' for this application. The load this would see would mostly be in the -Z direction (but could also involve some torque on the rod), so the lockring at the front was to try to stop the rod from unscrewing and "backing out" (moving 'forward' (+Z direction) would be unlikely to occur naturally when in use). I like your suggestion, it limits the X-Y motion to whatever the clearance between the ram and the smooth bore are (which should presumably be more easily controlled than the thread clearances). I will have to knudle on this for a bit... I really don't want the base-plate to be any thicker than 3/4" though (partly because I already ordered it, partly because I feel like anything thicker would be "overkill"). But I could see building this up with two pieces: the plate with a smooth bore, then a nut at the back which is simply tacked to the plate (the "nut" could even be thicker than needed with an oversized smooth bore partially through it to 'push' the threaded section even further to the rear). The threads are then just used for moving the ram forward and back, so the threads could actually be quite "sloppy" and could be a much lower TPI. My gut says (it just keeps throwing out it's opinion doesn't it) that a lockring on the back of that wouldn't add anything though, so I would still need someway to lock it once in position...

Ok, I'm dyin' to know.....what ya buildin'?

That would ruin the fun wouldn't it -- everyone could just say "oh that's been done", "no need for that", etc, etc... Seriously though, I haven't decided if I want the design for the actual assembly "out there" yet or not. I'm thinking this might be a useful tool (for some subset of the hobby-machinist crowd) and depending on how it works out, I may actually consider selling it. So I just want to keep my options open for right now...

Thanks for the suggestions! I'm definitely rethinking the threaded-ram and leaning toward a smooth-bore ram at this point...

 

[f350ca]

Sort of thinking out load or on the forum. If the threads were a loose fit, would the angle of the thread profile self centre the screw in the female thread when you tightened up the lock collar. If so x-y might be quite repeatable.

Greg

 

[kizmit99]

If the threads were a loose fit, would the angle of the thread profile self centre the screw in the female thread when you tightened up the lock collar. If so x-y might be quite repeatable.

I would think might be the case if the threads were perfectly formed (just loose), but any imperfections in the thread form would (I think) mean that their final position when loaded with the lockring would be highly dependant on the variation in each thread...

I'm definitely leaning toward the smooth ram approach suggested by Tony. Did a little mocking up yesterday and (as I suspected, and probably quite obvious to most) a lockring on the tail-end provided no locking function whatsoever. I then considered doing away with the threads altogether and using a split-collar to lock the ram in position. While this should work for locking, it provides no capability to use the threads for either (some level) of fine position adjustment, or to "push" the tool into the workpiece.

My latest thought is to return to the half-smooth, half-threaded ram, then split the threaded "nut" and use it as a split-collar lock (compressing against the threaded section). I could also make the nut out of aluminum which would (I would think) lend itself to making a better split-collar than steel. If I did that though I would have to use fasteners to attach it to the steel plate instead of just tack welding it in position... I just checked McMaster-Carr and they have a steel 2" ID threaded clamp-on shaft collar that would probably work, for $30... I'll probably give fabbing one up in Aluminum one a try first though...

 

[joshua43214]

Sorry to chime in late, a few thoughts.
The words smooth, precise, and tack weld do not work together. If you are looking for a precise, smooth fit up, then either machine after welding (why weld at all then?), or use fasteners.

It seems to me this has all gotten a bit overly complicated.
You will not get consistent positive stop and lock using a split nut. All re-loaders know this. Re-loading dies use split nuts (or grub screwed nuts), getting fine adjustment requires a fair bit of fiddling at times. Once set up, the die will thread in to the same spot consistently, but getting it set up is enough of a hassle that may re-loaders simply by another set of dies rather than disturb a current set up. You have probably had similar experience with things like leveling feet that move the machine consistently, then suddenly seem to move it randomly.

I get all the theory, if such and such is set up in so and so a manner, and if this and that is machined to some other standard, then everything will work perfectly if some particular state is started out from. Sadly, this set of circumstances only exist in a lab, and this is why laboratory grade machine tools make the best tool room grade machine tools look cheap. You can actually buy a lathe that will make a syringe small enough to draw the DNA out of a cell's nucleus, this precision instrument is about the size of a microwave oven and costs more than a house.

You have two different issues at work here, screw lash, and locking. Both have tried and true simple methods we use every day. Lash is reduced to minimum with a double nut or lead screw style nut. Clamping is done with a split collar or clamping pads on an unthreaded section.

It is the nature of threads to give when placed under tension. V-threads in particular give more, this is just the nature of the beast. This is why we use Acme or trapezoidal threads for lead screws - they less give along their axis less than v-threads because the thread face is closer to perpendicular. On the other hand, Acme threads tend to be looser in the bore, which produces inconsistent seating when you advance, then retract, then advance the screw again. Acme threads will also tend to lead off in a random direction unless supported at the far end.

If you want to be able to advance a thread in a very controlled manner, then lock it in place, look no further than the tail stock on your lathe. It is a tube that contains a ram that is threaded, it has a lash adjustment on the thread (mine does, not but it could be added easily), and locks in place using either a split collar around the ram, or threaded clamping pads that pinch the ram.
The thing to note about a lathe tail stock, is that the ram is guided by the tube, not the thread. You simply can not guide a threaded rod with another thread, it will always wander off axis.

Lastly, aluminum (and stainless), are poor choices for the threaded end. Aluminum threads will be more prone to wear, are very sticky compared to steel, cast iron, or bronze, and will wear at an accelerated rate once wear begins. Aluminum is just always a poor choice for anything that moves. Stainless wears better, but it is also sticky and does not thread well (usually).