PDA

View Full Version : Ballscrews v. acme screws



russell
07-10-2011, 05:23 PM
While waiting for the SX2 mill that I want to convert to be back in stock at ArcEuro I've been giving some thought to the pro's and con's of converting to ballscrews for the x and y axes. Here are my thoughts (perhaps someone will correct any false assumptions):

1. Backlash:
Cheap (well, cheapish) chinese ballscrews (C7 spec) typically have about 0.003" backlash. I have seen reports of people acheiving 0.001" to 0.002" with an acme screw and a well made Delrin anti-backlash nut so, even if they are a bit optomistic, there doesn't seem to be much difference.

2. Driving torque requirement:
Ballscrews have about 90% mechanical efficiency while acme screws have about 50% so, on the surface, it would seem that a much smaller stepper motor could be used. However, the original leadscrew is 2 mm pitch while the ballscrews are 5 mm pitch so the ballscrews will require actually require about 40% more torque.
Of course we also have to consider the required resolution. I would be happy with the 0.0002" that the acme screw will give at 1/2 step. To get nearly the same resolution (0.00025") with the ballscrew will require 1/4 stepping. Now a stepper at 1/2 stepping produces about 70% of the full step torque. If 1/4 stepping it only produces about 40% of full step torque.
So, for the same resolution, I will require steppers with rated torque over twice as high for ballscrews as for acme!
I find this surprising! Am I right?

3. Fast traverse speed:
Clearly, for the same speed from the electronics, the ball screw is 2.5 times as fast. Or is it? Does the microstepping affect this?

4. Wear rate:
I can't find any figures but I guess the ballscrew wins here. The acme anti-backlash nut will require an easy adjustment.

For my hobby use 3. and 4. are not very important so, also considering cost, keeping the acme screw would seem to be the best alternative - or do you know different?

I look forward to comments.

Russell.

m_c
07-10-2011, 06:05 PM
1. Backlash nearly always comes down to price. Only you can decide if you want to spend the extra time and money to reduce it.

2. Are you basing your calculations on the effiency corrected torque?
For a given pitch, the ballscrew will transmit nearly twice as much torque compared with the acme, which means you could drive the same table at the same force, but at twice the speed using the same motor.

2/3. Speed depends on how much torque you're willing to sacrifice. If you go for gearing (usually done via belt drive), then you can increase torque and resolution, but loose speed, or vice versa.
Plus I've seen it mentioned that you shouldn't rely on microstepping for positioning and you should only expect accuracy to the nearest full step.

4. Everything wears, but ballscrews should be minimal provided that they're well lubricated.


If you're unsure, then go cnc with the standard acme screws, and see how you get on. You can always convert to ball screws at a later date.

Jonathan
07-10-2011, 06:40 PM
1) All the C7 screws I've got have been <0.001" backlash... even after a fair bit of use. Also, I'm not sure, but I reckon the delrin nuts will be a lot more 'springy'.
Either way for a mill your best off using two ballnuts with a spring in between to eliminate backlash (google it).
2) No, luckily that's not right. The stepper motor torque is vector sum of the motor winding currents, so 1.41 (root 2) for full step and 1.00 (as sin^2(x)+cos^2(x)=1) for a microstep drive. However that doesn't apply now as for over 10 years microstepping drivers have used morphing, distorting the waveform from sine/cos to closer to square to achieve the full torque.
3) Advantage for ballscrew as the higher pitch means screw angular velocity is much less, so the inertia and thus the energy the motor has to impart into the screw is a lot less. May not be that significant compared to the relatively high coefficient of friction of dovetail slides, but definitely is on a machine with rolling bearing slides.
4)Ballscrew definately wins ... hardened metal rolling versus hard steel *sliding* on plastic!

I still have ACME screws on my mill that I converted to CNC a few years ago ... it works fine, fast enough, but the backlash is really annoying so I'm going to convert it to ballscrews asap. When I've decided what diameter screws are best and will fit.

russell
07-10-2011, 07:53 PM
1. Backlash nearly always comes down to price. Only you can decide if you want to spend the extra time and money to reduce it.Agreed


2. Are you basing your calculations on the effiency corrected torque?I'm basing it on the torque required to transmit a given force to the table for the linear motion taking account of the efficiency of the screws.


2/3. Plus I've seen it mentioned that you shouldn't rely on microstepping for positioning and you should only expect accuracy to the nearest full step.
Good point and another argument against direct driving a ballscrew. Perhaps microstepping should only be considered to reduce vibration by cloer approaching a sine/cos waveform.


If you're unsure, then go cnc with the standard acme screws, and see how you get on. You can always convert to ball screws at a later date.

russell
07-10-2011, 08:15 PM
1) All the C7 screws I've got have been <0.001" backlash... even after a fair bit of use.
Interesting, they must be well within limits.

Also, I'm not sure, but I reckon the delrin nuts will be a lot more 'springy'.
Of course, delrin is about 50 times more "springy" than steel but the springs in antibacklash nuts should compensate.

Either way for a mill your best off using two ballnuts with a spring in between to eliminate backlash (google it).
A bit beyond my budget. It would also reduce table travel too much.

2) No, luckily that's not right. The stepper motor torque is vector sum of the motor winding currents, so 1.41 (root 2) for full step and 1.00 for a microstep drive. However that doesn't apply now as for over 10 years microstepping drivers have used morphing, distorting the waveform from sine/cos to closer to square to achieve the full torque.
Sorry, I don't understand this but my electrical engineering degree was obtained 44 years ago. I'll have to do more reading! Can you give me a technical reference?

3) Advantage for ballscrew as the higher pitch means screw angular velocity is much less, so the inertia and thus the energy the motor has to impart into the screw is a lot less. May not be that significant compared to the relatively high coefficient of friction of dovetail slides, but definitely is on a machine with rolling bearing slides.
Agreed, inertia is not a concern for a small mill with dovetail slides.

4)Ballscrew definately wins ... hardened metal rolling versus hard steel *sliding* on plastic!
Agreed, ball screws will wear much less but is it significant for a hobby machine?

Thanks for the replies chaps.

Russell.

Swarfing
07-10-2011, 08:27 PM
Russell i will vouch for TR and use 16x4 with no problems at all. I cut the thread on Acetal with a home made tap and when oiled works really well. The truth is i can replace these nuts 60 times over before catching up with the cost of a ballscrew. Would suggest you buy stainless if you go that route ;-)

Like whats been said you could alway change it later (don't forget to make allowances for it in your conversion). And yes i use ballscrews too.....

m_c
07-10-2011, 09:05 PM
Perhaps microstepping should only be considered to reduce vibration by cloer approaching a sine/cos waveform.

Primarily, the main function of microstepping is to help motors turn smoothly at low speeds.
If you want to read about various aspects of stepper drivers, and have several hours to spare, search out the many posts by Mariss Freimanis over on CNC Zone, particuarily in the Gecko Drive and Stepper Motor forums. Mariss is the brains behind Gecko Drives, so knows quite a bit about them!


Forgot to mention, keep us updated about the build/conversion!
I'm considering a SX2 once funds allow, as it's about the ideal size for some parts I'm thinking about making/selling.

Jonathan
08-10-2011, 12:11 AM
A bit beyond my budget. It would also reduce table travel too much.

Assuming you're getting the ballscrews from linearmotionbearings2008 on eBay the extra ballnut is pretty cheap. I can't remember exactly but I think it's less than $30 for an extra ballnut.


Sorry, I don't understand this but my electrical engineering degree was obtained 44 years ago.

Mine should be obtained in almost 3 years. Looks like I'll have to wait until the final year for the module containing stepper motors, sigh.


I'll have to do more reading! Can you give me a technical reference?

Err...not without using Google which I'm sure you can do. But I will explain a little more fully.

I said originally that the stepper motor torque is the vector sum of the motor winding currents (common sense really). More precisely it is proportional to the phase currents, up to magnetic saturation of course, but that is irrelevant to this discussion. For full step the rated current is applied to both phases, call it 1 amp, so the vector sum of those is:(1^2+1^2)^0.5=2^0.5=1.41
When microstepping a sine wave is applied to one phase, and cosine to the other phase. So the vector sum of those is:
(sin(x)^2+cos(x)^2)^0.5
Using the trig identity sin(x)^2+cos(x)^2=1,
(sin(x)^2+cos(x)^2)^0.5
=(1)^0.5=1

Hence the torque when microstepping would be 1/1.41=0.71 times what you get with a full step drive.

But, to avoid this (and due to other effects such as resonance damping / motor not ideal) the waveforms are not pure sinusoids. They are distorted to be closer to a square wave, and hence you get close to the full torque. Some drivers now revert to a full-step drive at higher speeds anyway, so clearly they must get the full torque.

Hope that helps, or maybe expanding my previous post was just stating the obvious...

russell
08-10-2011, 10:04 PM
Found a couple of useful references:
A tutorial from the University of Iowa http://www.divms.uiowa.edu/~jones/step/index.html#others (http://www.divms.uiowa.edu/%7Ejones/step/index.html#others)
and a book published by the IEE http://ebookee.org/Stepping-Motors-A-Guide-to-Theory-and-Practice-Control-Engineering-_598987.html
(Should have known about the last one being an MIEE myself!)
I've got more reading to do but discovered these points:
1. Step position is accurate for both full and half steps and is lower for higher degrees of microstepping.
2. Microstepping always involves driving the windings with an approximation to sin/cos waveforms. The "Morphing" referred to above involves switching from microstepping to full stepping at high speeds to increase the torque.
3. Torque depends on the angle between the present position and the desired new position according to a sine function. Thus for example if you are 1/8 microstepping and wish to move one microstep the torque will be reduced to 20% of holding torque. However if you wish to move say 4 microsteps at once the torque available will be 70% of holding torque and if you wish to move 8 microsteps the full holding torque will be available. So in a way both Jonathan and myself are right.

Now to study in a bit more depth!

Good luck with the degree course Jonathan.

Russell.

Jonathan
08-10-2011, 10:25 PM
However if you wish to move say 4 microsteps at once the torque available will be 70% of holding torque and if you wish to move 8 microsteps the full holding torque will be available. So in a way both Jonathan and myself are right.

True, but that cannot apply in out case as we use step/dir drivers which can only ever be commanded to move one micro-step at a time.

Swarfing
08-10-2011, 10:32 PM
Jonathan consider this telling off.....You are far too technical for your own good sometimes, this is a hobby site for a lot of people and "Captain i just cana get fix"

:-) :-) you uni boys will learn much more once you are in the real world...LOL!


PS

It is not a real telling off, just trying to slow you down a bit

sjzrifdsa
25-10-2011, 10:37 AM
really good!

Gary
25-10-2011, 06:48 PM
Just to add a few things about the ballscrews, most if not all of the low cost C7 ballscrews are actually made in Taiwan by the larger better known ballscrew manufactures, it is just the low cost ballnuts that are made in China.
Fitting a spring between the nuts while eliminating the backlash is not done commercially because it causes early wear.
There are a few ways to eliminate backlash and the best way is to have two nuts with a spacer between the nuts and the thickness of the spacer accurately place the balls on opposite sides of the thread.
As long as you look after the ballnut and screw by lubricating it, it will last for years even if you push it hard.
The argument of the spring takes up the slack when the nut beds in or wears is a bit daft because if you look after the nut, it shouldnt wear, especially on a hobby machine.
I have known 20 year old machines that are worked hard and still maintain good repeatable accuracy with a good ballscrew.


The low cost chinese ballnuts are ok for hobby use but i would not even consider them for a production machine.

russell
25-10-2011, 07:46 PM
Jonathan consider this telling off.....You are far too technical for your own good sometimes, this is a hobby site for a lot of people and "Captain i just cana get fix"

:-) :-) you uni boys will learn much more once you are in the real world...LOL!


PS

It is not a real telling off, just trying to slow you down a bit
Take no notice Jonathan. An engineering degree is a great start. Of course you will learn more later. It is much more difficult to start with the practical skills and learn the theory later!

If possible take a year out during the course to get industrial training. You will have much more to offer a potential employer.

Back to the topic:
Your quote "True, but that cannot apply in out (sic) case as we use step/dir drivers which can only ever be commanded to move one micro-step at a time" is not quite true.

Consider the case where the motor is stationary and you try to move it one microstep. If the torque when microstepping isn't sufficient to overcome the friction the motor stays where it is. You now try to move it another microstep. Now the torque produced by the motor is increased (almost doubled for fine microstepping). Each microstep asked for increases the torque (up to a maximum of 70% of the full step torque) until the motor is able to rotate. The positioning may thus lag behind the desired position by several microsteps.

Russell.

P.S.
The drive waveform is only sinusoidal during continuous motion.