[replicapro]

On my Mill the lead is 10 turns per inch, which is a lead screw. It is 5/8 so I figured Id get 5/8 ball screws. The only difference is the ball screws are 5 turns per inch. Will this make a huge problem?

[bb99]

You’ll loose in accuracy but gain in speed. Which axis will they be for? There is a slight issue with the z-axis. The current setup you have may have closely paired the motors with the screw and thus they may have a problem with lifting/holding (due to more distance traveled per turn).

--Alan

[dcd121]

Here is a related question. I have some 5/8" dia ball screws for all axis that travel 20 mm per turn. I plan on using a servo setup with them. I know I will have enough power and speed but would someone like to speak about the accuracy issue?
Thank you

[replicapro]

So other than the inherent backlash lead screws tend to have the accuracy of the ball screws at that feed would make more sense to keep the lead screws.
Or find ball screws that have equal feed as the lead screws.
I looked at Mcmaster carr and the longest feed they carry is 5 tpi

[replicapro]

After thumbing through the NOOK catalog I read the first few pages regarding lash in ACME lead screws. They recomend using anti lash flanges with periodic adjustments for wear.

Basically the say to mount your anti lash flanges on either side of the block and create compression similar to the way preloaded ball nuts work.

This got me thinking about providing a constant light flow of lube from the main flood pump to allow the anti lash flanges to be constantly lubricated during use to not only reduce wear, but to keep the flange clean and running smooth.

A small access hole drilled near the block on all axis's could allow the introduction of a fluid flow that would stay pointed at the block yet move with the ways.

This would reduce the lash, keep the threads clean and reduce wear.

[ESjaavik]

Originally posted by dcd121
Here is a related question. I have some 5/8" dia ball screws for all axis that travel 20 mm per turn. I plan on using a servo setup with them. I know I will have enough power and speed but would someone like to speak about the accuracy issue?
Thank you

If connected directly to the servo it will be lower than one with 5mm/turn. But if you gear down 1:4 and do it without backlash in the gear (planetary or cyclic) , it will be the same. Your moment of inertia will be lower because you have geared down. So your motor will be able to accelerate it quicker. That's a pro.

But a gear without backlash is expensive and bulky. That's against. Unless you get a ballscrew integrated with a gearbox it's also more components to make to adapt the gear to the machine and the motor to the gear.

[dcd121]

ESjaavik,
Thanks for you input. The question is, how much less accurate. I think I will go ahead with the direct drive method. I'm fairly certain it will be good enough. If not, then I could go the gear reduction route.

[metlmunchr]

I think accuracy and resolution are being confused here. Accuracy is solely dependent on the lead accuracy of a particular screw, and not on its pitch or how many revolutions the motor may make per turn of the screw. Speed reduction via gearing or belts, or the use of a fine pitch screw doesn't increase accuracy. It increases the apparent resolution, but this can be misleading in that you're showing resolution to .0001 yet the position may not be within .005 of the indicated position. General statements cannot be made about the lead accuracy of ball screws versus lead screws either. For example, I have a manual jig borer with ground Acme screws that will equal or exceed the lead accuracy of the ball screws in any CNC machine I own, and these are CNC machines that come with 6 figure pricetags. I've made a lot of crossfeed screws for large engine lathes one of my customers owns. Although these machines all have DRO's on them, we still use .003/ft acme rod to make the screws instead of the cheaper .009 material. That way, if something happens to the DRO, the operator can simply continue by reading the dials and know he's getting reliable readings. You can usually make an Acme lead nut with less backlash that the ones readily available for purchase. The typical commercial acme nut is tapped rather than having the thread chased. By grinding a threading tool to the proper Acme form but leaving it a bit narrow, you can chase the thread to depth, and then by advancing the start point, either via the compound or via tool offsets if you're using a turning center, and gradually widen the thread until the matching screw will thread thru the nut with firm hand force. Running the two back and forth a few times will burnish the thread in the nut and the resulting combination will have very little backlash. Works great for a good tight manual machine, but still not adequate for a cnc mill or lathe IMO. Regarding accuracy and resolution, always remember, in the vast majority of CNC setups, the position you read on the screen is, in actuality, simply the position the control commanded the motor to go to.

[ESjaavik]

Resolution depends on your screw pitch, gearing and stepper/resolver/encoder resolution.
Accuracy is dependent on these, but also on the stiffness of your stepper / servo.

With a servo, the stiffness is something you have to tune in. And it comes at the cost of instability. That is you cannot increase the stiffness beyond a limit, then it will become unstable, make noises and wear down the mechanics.

With a stepper you increase the stiffness by applying more current. But you cannot go beyond the point where the drive or motor overheats.

With a 20mm lead, direct drive and a 200 step/rev stepper motor your accuracy will be around 0,1mm. I'm fairly sure the motor will be the limiting factor here. But any inaccuracy in your screw has to be added. Microstepping will increase your resolution, but it will not make the motor stiffer.

Acme screws and ball screws can be made to the same accuracy except that Acme screws usually must have backlash or there will be no place for lubricant.

To see how stiffness works, try to take a step motor. Apply current to it and try to turn it. It works like a spring up to the point where it cogs over. Increase the current and try again. The "spring" becomes stiffer. But important: the shaft does *not* become locked even if you increase current to maximum smoke.

[ger21]

Microstepping will increase your resolution, but it will not make the motor stiffer.

I've read a lot of posts on this topic. Don't count on microstepping increasing resolution. The main benefit of microstepping is smoother running steppers. Due to the fact that when microstepping, the more microsteps, the less torque each one gets (something like that), so you won't necessarily see all the microsteps. But you won't be losing any steps, you just can't always count on the microsteps to be accurate (precise?). There was a large discussion about this on either the CCED list or the DIY CNC list at Yahoo a few months back.

[ger21]

With a 20mm lead, direct drive and a 200 step/rev stepper motor your accuracy will be around 0,1mm

I think you meant to say precision here.

[ESjaavik]

ger21:
Sorry, not my native language. I don't know the difference between accuracy and precision.
I thought both to be the deviation between the intended and the actual position. And the resolution to be the minimum difference between 2 points that can be represented. Please help improve my english vocabulary.

I don't fully agree with your view on microstepping. If the current levels of all the microsteps are well matched with the motor and it's detent forces, each microstep will have a resting point proportionally spaced within a step to it's microstep number. And the compliance will be largely the same as more than one winding will carry current except for the steps corresponding to full steps. The reason i say "fully" is that in real life microstep currents are not well chosen, and almost never consider the detent forces. With my drives quarter stepping is so close to perfect I cannot measure the error with a dial indicator, while 1:20 stepping is no better except for more quiet running. The last 4 steps before going to zero current does not overcome the stick slip then the 5'th goes to where it should be.

I think there was a thread on using a laser pointer secured to the shaft to measure the movements of microsteps. That's a bright idea.

[ger21]

With my drives quarter stepping is so close to perfect I cannot measure the error with a dial indicator, while 1:20 stepping is no better except for more quiet running.

That's basically what I said, isn't it? Or am I missunderstanding you?

If you're drives can be setup for 1/20 stepping, can you move, say 3 microsteps, and measure to see if it's in the right position?

If I can find the time this weekend, I'll try to dig up some of the info I'm talking about.

[ger21]

Sorry, not my native language. I don't know the difference between accuracy and precision.

Somewhere around here, I think in HomeCNC's building thread, is a huge debate about accuracy and precision. I'll try to post the link later.

[ger21]

Read this:
http://www.machinedesign.com/ASP/vie...strSite=MDSite

[ESjaavik]

That page is useful in one respect: It helps kill the common misconception that a stepper goes exactly where it's commanded and stay there. But it avoids any arguments for a better precision using microsteps, thus loosing credibility.

Yes, I can measure each microstep even in 1:20 stepping. If my drive electronics did not show a problem it would better my precision compared to 1:4 stepping and certainly to full stepping.

Let me see if I can include a picture here:

If that worked, you can see the current through one winding while stepping. It is not a sinewave! If you superimpose two of those on each other with 90 degrees phase shift you will see 4 discontinuities for each full wave. And that's exactly what I canobserve using a dial indicator. But between them the indicator shows each microstep and it is as close as a 1/100mm dial indicator is able to show. And this is including the stick slip not only of the motor but also of the >50Kg axis. I trust my dial indicator more than a webpage.

I write this not to flame you, but I believe you have been taken by the black/white debate on microstepping. I did myself before I actually tried it.

One thing I do realize though is that when cutter forces come into play, the axis will not be where it is without it. The same when there is a 50Kg lump of iron on the table. Because the motor has a "magnetic spring" in it, it will not go to the commanded position. But it will go to the commanded position +- errors. And most of those errors are not dependent upon where it is commanded to go even if it's a microposition. So assuming I can get rid of the uneven current shown I *will* get a better precision because I can command it to 0,00125mm plus/minus errors which is better than 0,0025mm plus/minus the same errors.

Now what about replicapro and the original question? Well so far my setup seems to be fine direct driving a 25mm x 5mm screw which is very close to his 16mm x 5,08mm screw. If I go in one direction and stop. Then go the other direction one step, the dial indicator will also change direction for that one (micro)step. So the backlash is as it should be: less than a good indicator can measure. Going from 10 to 5 pitch is no problem for the program, it's just a setting to change. The handwheel indication will be wrong by a factor of 2. But I assume the handwheels will not go back, so that is no problem either. So I'd say go for it. But use a ground and preloaded screw.

[arvidb]

Originally posted by ger21
Somewhere around here, I think in HomeCNC's building thread, is a huge debate about accuracy and precision. I'll try to post the link later.

Here it is: http://www.cnczone.com/forums/showth...&pagenumber=16

I cannot refrain from adding how I interpret these terms (*oops* :bat: ):

Accuracy: if your machine has no systematic errors, then it is accurate. That is, if, disregarding backlash and flex, your machine always moves X mm when commanded to move X mm, then it is accurate.

Resolution: if the smallest movement you can command is 1 micrometer (um), then you have a resolution of 1 um. This is *not* the same as accuracy, since your machine will (most likely!) not be able to move and keep position within 1 um!

Precision: the ability to return to the same position many times (= repeatability). A machine with lots of backlash or flex is imprecise.

You cannot make an accurate part (= a part that lives up to spec) unless you have enough of all of the above.

Arvid

[High Seas]

Arvid - I'll add my OOPS at the risk of doing a "restart" on the previous discussion:
System Resolution also includes effects from the drive mechanism (ie, rod in this case).

TO divert from the CNC for a moment, maybe use the CRT/digital camera or similar analogy - maybe that helps. A great video card can have excellent resolution - but you can't see the results on an 80 column green screen CRT.
Or another example; my aging digital camera won't take advantage of the video card or a HiRes screen, 'cause the system resolution is limited by the "weakest link," in this case the camera.
The controller may command and produce accurate positioning and micro step - but the tool can only reach the Resolution of the combined components, controller, motor, drive system, mechanical rigidity of the table, etc.
Sorry for so many mixed metaphors, just trying to keep it simple. In most hobby uses the results are most likey better than "good enough!" Striving for Six Sigma is important - just not always necessary.
Cheers - Jim