Carriage Discussion

Carriages

David G. LeVine

Sunday, April 27, 2008 11:46 AM

Traditionally the carriages were driven by screws, ball screws are just
another kind of screw, albeit more efficient and with better control of
friction.

Often, on lathes with half nuts, the carriage was driven by a rack and pinion system when the half nut was not engaged. When discussing a lathe built for an electronic lead screw like the ELS, is there a good reason not to consider rack and pinion for both axes? Also, is there any reason for a compound on an ELS type system? On manual systems, it is critical for such tasks as small tapers and threading, but the computer can do that with only two axes or can adjust the tool height on the fly.

Finally, is it desirable to plan for a DRO even if none is used most of the time? Having the mounting pads (which is all we are discussing here) may be incrementally cheap, especially if they are not machined unless the DRO is planned for installation, and a purpose-built grinder would make short work of a surface parallel to the way. Servomotors running with a DRO, rather than a rotary encoder, get VERY accurate, in the 1970s we were getting 0.00025" (yes, a quarter of a thousandth) over a 24" x 24" X-Y table driven by steppers which corrected based on linear encoders.


David G. LeVine
Nashua, NH 03060

Michael Fagan

Sunday, April 27, 2008 12:50 PM

—- In moc.spuorgoohay|ehtaLnepO#moc.spuorgoohay|ehtaLnepO, "David G. LeVine"
wrote:
”Traditionally, the carriages were driven by screws, ball screws are just another kind of screw, albeit more efficient and with better control of friction.”

Good idea, particularly for use with the ELS/CNC conversion. The only problem I see is that you can't have halfnuts on a ballscrew, so you can't disengage it. However, many ballscrews used for CNC work are very coarse, compared to a leadscrew, to allow for fast rapids and large balls. They are also lower friction than a traditional Acme screw. Perhaps we'd just have a leadscrew handwheel connected to the ballscrew and do without the rack and pinion entirely.

Now I've gotten an even better idea. Traditionally, leadscrew handwheels have been mounted on the end of the bed, which is fine on a little Unimat or whatever, but a not very elegant solution for a 36" between centers lathe. However, if you stuck the ballnut (the part in the apron) on bearings and made it free to turn, then locked the screw in place, and hooked a pair of bevel gears to the carriage handwheel and the ballnut assembly. Now you simply spin the ballnut, and the carriage moves. For ELS/CNC use, you disengage the carriage handwheel, lock the ballnut in a fixed orientation, and spin the ballscrew with your stepper/servomotor like normal. You get all the benefits of a carriage handwheel without the awkwardness of reaching down to the end of the bed to get to it. And, we can eliminate the need for a rack and pinion. Finally, the bevel gears need not be 1:1, so if the pitch of the ballscrew did not appear suitable, it could be geared up or down to the carriage handwheel, to approximate the speed of travel given by a traditional rack and pinion carriage drive. Often, on lathes with half nuts, the carriage was driven by a rack and pinion system when the half nut was not engaged. When discussing a lathe built for an electronic lead screw like the ELS, is there a good reason not to consider rack and pinion for both axes?
Or, possibly, a lever cross slide (like on a turret lathe), or at least being able to disengage the leadscrew and attach the lever.
Also, is there any reason for a compound on an ELS type system? On manual systems, it is critical for such tasks as small tapers and threading, but the computer can do that with only two axes or can
adjust the tool height on the fly.

What about infeeding at 29.5 degrees when threading? If you had synched motion, it would be possible, and I suppose even the ELS could do it if it moved back by the correct amount (equivalent to the distance the tool tip moves longitudinally when infeed at 29.5 degrees.

I would definitely make the compound removable so that a more rigid cross slide could be used. The bearing surface would bolt onto the tee slotted cross slide (like a 9x20 import), rather than being cast into the cross slide (like an Atlas or South Bend)
Finally, is it desirable to plan for a DRO even if none is used most of the time. Having the mounting pads (which is all we are discussing here) may be incrementally cheap, especially if they are not machined unless the DRO is planned for installation, and a purpose-built grinder would make short work of a surface parallel to the way.

Yes, I totally agree. Is this a place where people would support something like the Shumatech, or would most people want an Accurite or something similar. I personally am attached to Newall system, but they cost roughly half again what I am hoping to build the entire lathe for.

Servo motors running with a DRO, rather than a rotary encoder, get VERY accurate, in the 1970s we were getting 0.00025" (yes, a quarter of a thousandth)
over a 24" x 24" X-Y table driven by steppers which corrected based on
linear encoders.

I've always wanted to set something like that up.

David G. LeVine

Hank S.

Sunday, April 27, 2008 2:08 PM

Hi All,
Sorry for the mix up, as said before, this is Just a Rough Draft.
Anything and or everything can be changed. Looks as though, the more you add pieces to, you are creating a damping system for the rest of the components. If the carriage is built as one piece or a few pieces, you tend to isolate the vibration to somewhere else in the unit, besides the carriage itself. There are more pieces to be added to this, as for holding the top piece to the brass angle inserts, and many more as time goes by. This is Just a Draft, and can be Hacked, or edited. It will most likely, that most of us, meaning me, have small machines to

try to build something of these sorts. Which leaves us with trying to do the machining in pieces, to work around in the limits of our machine capabilities. I tried this once, and have made one much like it before, but have gotten with other jobs to do, and has been put aside, and parts where used for something else. Feel Free to post an opinion, as that’s all where it starts. Trying to find the right Flavors of Material is going to be the task, as for the machining goes, well, that’s a Different Page.

Thanks,

Hank S.

David G. LeVine

Sunday, April 27, 2008 3:24 PM

Good idea, particularly for use with the ELS/CNC conversion. The only problem I see is that you can't have halfnuts on a ballscrew, so you can't disengage it. However, many ballscrews used for CNC work are very coarse, compared to a leadscrew, to allow for fast rapids and large balls. They are also lower friction than a traditional Acme screw. Perhaps we'd just have a leadscrew handwheel connected to the ballscrew and do without the rack and pinion entirely.

That was why I was suggesting that we eliminate the screws entirely and just use rack and pinion (with preloaded anti-backlash pinions.)

What about infeeding at 29.5 degrees when threading? If you had synched motion, it would be possible, and I suppose even the ELS could do it if it moved back by the correct amount (equivalent to the distance the tool tip moves longitudinally when infeed at 29.5 degrees.

The ELS already does that, no change needed from the box stock
configuration.

David G. LeVine
Nashua, NH 03060

Michael Fagan

Got it, good idea. What does a set of antibacklash pinions cost?
Also, do you need a high-precision rack or is normal commercial
quality good enough?

Ian Newman

Sunday, April 27, 2008 4:43 PM

Hi,
Ballscrews do not have the friction inherent in a conventional threaded lead screw. They will not hold a component in position and have to be held locked by their drive motors to prevented from moving under load. If not, you can push the table and the ballscrews just spin and let the table move. This means you cannot simply put handles on the ballscrews and work them manually.

Ian.

David G. LeVine

Sunday, April 27, 2008 8:12 PM

Michael Fagan wrote:

Got it, good idea. What does a set of antibacklash pinions cost?

About twice what two normal pinions cost. All they are is a pair of pinions back-to-back with a little spring, which tries to move them out of mesh. This can be two on a common shaft or two mounted on parallel shafts. Just as long as both are in contact with the rack and there is a load between them, backlash can't exist. The spring(s) MUST generate forces, which exceed the forces on the pinions, or backlash shows up.

From Pic Design:

Anti-Backlash Gears
Two independent gears mounted to same hub with a spring between the two providing a constant full-tooth engagement with the mating spur gear, thereby eliminating backlash in the mesh. Available in Spur, Worm, and Miter Gears.

<http://www.pic-design.com/PICSmartCAT/SectionSubFam.asp?section=sec12_data&section_file=sec_12/Sec_12.htm>

Again, from Pic design:

PIC
Class Rack Max Spacing Error Parallelism
To Base
Tooth-to-Tooth Accum in 11"
PREC-1(Standard) .005 .002 .001
PREC-2 .004 .001 .0007
PREC-3 .002 .0006 .0005

<http://www.pic-design.com/PICSmartCAT/SectionSubFam.asp?section=sec12_data&section_file=sec_12/Sec_12.htm>

Also, do you need a high-precision rack or is normal commercial quality good enough?

Now for the weird questions and answers. Precision costs money, how precise do you want to go.

For most operations, standard precision is probably fine, for highest precision, use of a DRO/linear scale/etc. as feedback for a servo system (note, not a servo motor, but a system) will give you precision and accuracy which the cut metal will not be able to duplicate due to thermal stresses. The accuracy of the rack determines the *open loop* accuracy of the machine. The DRO/scales/etc. determine the *closed loop* accuracy of the machine. For many purposes, +/- 0.003" absolute will be fine, especially if in a short segment the differential accuracy is +/- 0.0005". There are techniques for averaging the error of a rack over a large interval, but I am not smart enough to discuss all of them.

Absolute vs. differential accuracy: If you are turning a 3.502" diameter piece to 3.500", as long as you try to move the tool 0.001", do you really care if it is 5% off? Now ask the same question about a piece 10" long, will 5% make a difference? Being able to determine position for small moves with lower percentage accuracy is generally okay, large distances generally do not like 5% errors.

Of course, if you want to be really anal, you can lap one rack against the other at many positions, which will average out all tooth-to-tooth errors. Bigger pinions also tend to average out tooth-to-tooth errors (they are in contact with several teeth.)

David G. LeVine

Michael Fagan

Sunday, April 27, 2008 7:59 PM

Right. I knew they were low friction. In this system, the motor would remain connected at all times, so you have that friction as well. I have used a CNC'd Bridgeport (Accurite MillPWR factory conversion) which had ballscrews, with large servos as well as handwheels for manual operation. They are definitely a different feel from normal Acme screws (if anything, the motor plus ballscrew was slightly harder, but smoother, than an ordinary machine), but they seemed to work fine.

I'm also comparing this to a normal manual lathe, where the friction is provided by the weight of the carriage on the ways. If you think about it, the rack and pinion drive of the carriage handwheel on a normal lathe provides minimal friction.

Michael Fagan

Sunday, April 27, 2008 8:56 PM

Yes, that's the feed screw (a misnomer, because it's usually just a plain rod, but everybody calls it a feed screw). Also, you will notice that the Atlas machines with power cross feed have a keyway cut down the length of the leadscrew. This is what drives the power cross feed through a bevel gear. On many machines such as the South Bend lathes, this keyway also drives the power longitudinal feed through a clutch and shifting lever. On the lathes with two shafts, these functions are simply separated. One is a plain shaft with a keyway, the other a precision threaded screw.
You are correct that the reason is to preserve the precision components used for thread cutting, although usually it's the half-nuts that wear down (because they're usually bronze) rather than the hardened steel screw.

Michael

moc.loa|51taobtac#moc.loa|51taobtac> wrote:

On Sun, Apr 27, 2008 at 5:40 PM,

On our Atlas machines we use the lead screw for two purposes, one to move the carriage to cut threads at a high rate compared to plain turning. On most "tool room quality" lathes (whatever that is) use different drives to move the carriage. Often a lead screw that is used only when cutting threads and for just plain turning there is a rod or some other kind of method of moving the carriage. (I have seen pics of lathes with apparently two lead screws. I think the reason is to preserve that expensive precision lead screw and use it only for cutting threads while plain turning is fed by some other means.

David G. LeVine ten.nozirev|441eniveld#ten.nozirev|441eniveld

Sunday, April 27, 2008 8:21 PM

Ian Newman wrote:
Hi,
Ballscrews do not have the friction inherent in a conventional threaded lead screw.
Actually, check out a thing called "preload" in ballscrews. They just have a lot less friction. They will not hold a component in position and have to be held locked by their drive motors to prevented from moving under load. If not, you can push the table and the ballscrews just spin and let the table move. This means you cannot simply put handles on the ballscrews and work them manually.
Ian.
Pretty much an accurate evaluation of the matter. Of course, one could easily put friction devices on the screws and eliminate the problem. How about a gear and a ball pin (Oh look, it goes 'Click' every half thousandth!)

David G. LeVine

Michael Fagan moc.liamg|88rekrowdoow#moc.liamg|88rekrowdoow

Sunday, April 27, 2008 10:22 PM

Yes, that was exactly what I had in mind. The worm/pinion sounds even
better than bevel gears.

On Sun, Apr 27, 2008 at 7:12 PM, Rexarino <moc.liamg|oniraxer#moc.liamg|oniraxer> wrote:

Michael suggested, "If you stuck the ball nut on bearings and made it free to turn" which makes me think of; Put a worm and pinion on one end of the ball nut, with the worm mounted on an eccentric so it can be disengaged from the pinion. When disengaged, the ball nut can turn freely, or by gearing or toothed belt can be turned by a stepper/servo motor. When engaged, the pinion shaft (which protrudes out through the apron) can be turned by the convenient hand wheel mounted to it, thus rotating the ball nut at a ratio determined by the worm and pinion, say 10 to 1, which allows for small movements with a coarse threaded ball screw.
Rex

Now I've gotten an even better idea. Traditionally, leadscrew handwheels have been mounted on the end of the bed, which is fine on a little Unimat or whatever, but a not very elegant solution for a 36" between centers lathe. However, if you stuck the ballnut (the part in the apron) on bearings and made it free to turn, then locked the screw in place, and hooked a pair of bevel gears to the carriage handwheel and the ballnut assembly. Now you simply spin the ballnut, and the carriage moves. For ELS/CNC use, you disengage the carriage handwheel, lock the ballnut in a fixed orientation, and spin the ballscrew with your stepper/servomotor like normal. You get all the benefits of a carriage handwheel without the awkwardness of reaching down to the end of the bed to get to it. And, we can eliminate the need for a rack and pinion. Finally, the bevel gears need not be 1:1, so if the pitch of the ballscrew did not appear suitable, it could be geared up or down to the carriage handwheel, to approximate the speed of travel given by a traditional rack and pinion carriage drive.

Rex

TAPR Open Hardware License (OHL) and the GNU GPL