Stepper Motors

[timo_gross]

Hi,
I have a question in mind. Since this is a CNC related topic I will ask it here.
Why do CNC setup requires stepper motors and ball screws with minimal backlash?
My thinking is that if we can get position feedback from DRO to motor driver the setup will be backlash free.
And you could use normal acme feed screws and DC motors. Maybe DC motors with electromagnetic brakes for holding position.

My question is, Is this possible or I my idea is totally absurd?

Regards
Nikhil

Hello,

Your idea is "semi-absurd" nothing new and nothing old.
For static machines, like a cutoff saw, a drilling machine etc. the "position, clamp, machine, reposition, clamp .... " works and is used.
Then again low precision ball screws, timing belts, pneumatic or hydraulic actuators, rack and pinion are usually seen on automatic machines.

An electromagnetic brake does not help you much to "hold a position" if your goal is to move not to stay were you are.
The acme screw is still used in manual machines, because it holds its position. A ball screw will move, when not held in place by motor or brake.
A manual machine usually moves one axis at a time and not ( all five of them ) at the same time.

Stiffness against machining forces and cutting vibrations is not the same as measurement of a position. RM-MN gave a good explanation.
You have to drive the workpiece into the cutter and the cutter is "chiseling" steel off the workpiece. (great deal of brute force)

The backlash compensation on my machine is pretty much useless a) backlash seems to be not constant b) the stepper system is too slow to take the backlash fast enough out when direction is changed. There is only so much you can do to fix a poor mechanical system with clever software (not necessary the softwares fault)
Do not underestimate the effort it takes to make clever software.

 

[ninefinger]

This may be a dumb question, but how do I determine a stepper motor size for my bench top milling machine? What are the determining factors & why?

Leveraging my expertise from converting one machine and updating another I can offer this advice:
Simple method - copy someone else's "working" example of a configuration. 98% chance if it works for them it'll work for you and save lots of time figuring things out.
Complicated method:
1. make lots of assumptions (or even more work, take measurements of) for mass and friction forces on the axes, and what cutting forces to expect.
2. determine the max velocities, accelerations and movement resolution you THINK you want (you don't know them cause you don't have a cnc yet...)
3. Do math for the required forces required to act on each axis to achieve the desired performance
4. make assumption on stiffness needed for ball screw (forces and whipping)
5. figure out combination of ball screw pitch and motor step resolution (including micro steeping) that results in desired movement resolution
6. Validate that chosen motor can output appropriate torque at the desired speeds
7. have fun building it and see what it ACTUALY does.
(simplified example)
I made spreadsheet that took in all the assumptions and let me change variables to determine the motor size and step resolution, screw pitch etc.
Reality is that most benchtop systems end up way overpowered on their axes compared to the cutters ability to cut (Hp and/or RPM limited) or machine rigidity (deflection), it may travel at 200in/min (5m/min) but there's no way its actually cutting anything properly at that speed except air. There are exceptions, but I'm generalizing.

As mentioned look at the CNC Zone for benchtop builds, lots of examples there including one of mine. Having done my first build the mostly hard way I advocate the simple method for a machine that someone has done before. Its not cheating, its smart to copy someone else's work and learn from their experience.

also, that is just the stepper (servo) size and resolution selection and screw definition - next is power supply, control system, e-stop, limit switches, etc.

Get ready - that rabbit hole is a hidden cavern!

 

[ninefinger]

I also forgot to mention you need to iterate a bunch of times either based on what is available or what you can afford, or what can physically fit or all of those. Just cause your calcs say you want it doesn't mean its readily available, or affordable or makes sense hanging off your machine (think of a cartoon race car where the engine is bigger than the car....)

 

[TSutrina]

To determine the size of a stepper motor you need to measure the torque needed to turn the part with the same force pushing against the bed. You can do this with a torque wrench and someone with a video camera that you can go frame by frame. Put weight on the bed for the downward force. If your doing a lathe make sure to take into account that the cutting force is in front.
An alternate method is to put a pulley to replace the handle or if it is round wrap some flat plastic strapping around it and then hang weight on the end of the strapping Torque is radius time weight when moving steady. The weight may put a load on the bed bearing. I would be conservative and multiply if possible by 2 to get the needed torque. Note: you have not taken into account the acceleration and your loading will not be perfect.
The other question is why use stepper motors instead of PM DC motors and position measurements. Actually that is done on commercial machines. And hydraulic rams have replaced the motors. The problem is that you will likely not find as easily to maybe very hard to find an off the shelf controller and supporting program. These approach is considered when the motor get large. For example commercial milling CNC machines cut thread single point by position sensor of the spindle motor that are seldom stepper. Say the same thing for a lathe.
I worked for Sundstrand that built big machine tools that use a infinitely variable drive (and applied the same gear box with hydraulic motors to constant 400Hz generators connected to aircraft jet engine spindle.) made up of gears and a hydraulic motor. Load down the squirrel cage motor and the drive adjusts to follow the CNC program.

 

[ddmckee54]

The primary reason to use steppers in these small systems is their low cost. A stepper allows the average hobbyist to use relatively simple software to do CNC control in an open loop - with no position feedback. The control software counts the number of steps so, under normal conditions, it doesn't need position feedback and the complexity/expense that goes with it. As long as you don't lose steps the control software knows where the machine is. You tune the stepper drives so they DON'T lose steps during acceleration. That's also why it's best to use ball-screws and hardware with minimal backlash. The software has backlash compensation built in but that's more work for the software and another place for error to creep in.

40+ years ago I worked on maintaining some of those Sundstrand machines, the were big, powerful, impressive, complicated, and expensive. We're not talking about commercial machines here, we're talking about hobby machines. Often these machines are a mill or lathe that is either entirely scratch-built, or a CNC conversion. These machines have some pretty impressive capabilities at a cost that is orders of magnitude less than their commercial counterparts. Are steppers used on commercial machines, usually not. Servos are faster, more powerful, more accurate, and more expensive. Can steppers be used on a lathe to do single-point threading, or on a mill to do thread-milling - you betcha! Depending on how fancy you want to get, you can even add a tool-changer to a hobby machine. Can they do work as fast as their commercial counterparts? No way, but then they don't have to - a hobby machine doesn't have to earn its' keep by putting parts out the door.

Don

 

[TSutrina]

The primary reason to use steppers in these small systems is their low cost. A stepper allows the average hobbyist to use relatively simple software to do CNC control in an open loop - with no position feedback. The control software counts the number of steps so, under normal conditions, it doesn't need position feedback and the complexity/expense that goes with it. As long as you don't lose steps the control software knows where the machine is. You tune the stepper drives so they DON'T lose steps during acceleration. That's also why it's best to use ball-screws and hardware with minimal backlash. The software has backlash compensation built in but that's more work for the software and another place for error to creep in.

40+ years ago I worked on maintaining some of those Sundstrand machines, the were big, powerful, impressive, complicated, and expensive. We're not talking about commercial machines here, we're talking about hobby machines. Often these machines are a mill or lathe that is either entirely scratch-built, or a CNC conversion. These machines have some pretty impressive capabilities at a cost that is orders of magnitude less than their commercial counterparts. Are steppers used on commercial machines, usually not. Servos are faster, more powerful, more accurate, and more expensive. Can steppers be used on a lathe to do single-point threading, or on a mill to do thread-milling - you betcha! Depending on how fancy you want to get, you can even add a tool-changer to a hobby machine. Can they do work as fast as their commercial counterparts? No way, but then they don't have to - a hobby machine doesn't have to earn its' keep by putting parts out the door.

Don

I said commercial machines which you expanded and added the reason in hobby machines. I agree but as most thing it is a little more complicated.
Ball screws do not eliminate back lash so do not save on software to compensate.
Understand that a ball screw can be back driven which creates it's own problems. All axis stepper motors are under control and supplied current even when no motion is required for a motor attached to a ball screw.
Ball screws and smaller stepper motors, stepper drivers, and power supply need to be balanced against the cost in money, time, and ability to change over plus smaller drive system. Against bigger drive system. Sometime throwing more amperes for an individual maybe a better solution. And a person may not need an axis CNC controlled.

 

[Toymaker]

<snip>
Ball screws do not eliminate back lash so do not save on software to compensate.
<snip>

Some ball screws do have backlash, however, "pre-loaded" ball screws have essentially zero backlash. On my CNC Mill shown in posts 15, 16, & 20 of this thread, which I built 12 years ago, I use back-to-back ball nuts which I can tighten together via a loading screw, allows me to eliminate all backlash.

I used newer, better, pre-loaded ball nuts on my CNC Lathe; I cannot measure any backlash in either the X or Z axis.
If there is any backlash at all, it's less than 0.0001", which is the limit of my dial indicators.

 

[awake]

My conclusion, based on the discussion thus far, is that when it comes to this sort of thing ...

... one needs to take it ...

... step-by-step.

 

[TSutrina]

Some ball screws do have backlash, however, "pre-loaded" ball screws have essentially zero backlash. On my CNC Mill shown in posts 15, 16, & 20 of this thread, which I built 12 years ago, I use back-to-back ball nuts which I can tighten together via a loading screw, allows me to eliminate all backlash.

I used newer, better, pre-loaded ball nuts on my CNC Lathe; I cannot measure any backlash in either the X or Z axis.
If there is any backlash at all, it's less than 0.0001", which is the limit of my dial indicators.

OK, so you trading off more drag and wear since the balls has to be sliding with preset force against each other and on the ball tracks of the nut. I have never seen a ball screw with a cage, but I haven't purposely looked. I do have limited exposure to ball screws.
Did they apply the solution that Rockwell used to reduce to eliminate back lash? (Rockwell B-1 Lancer bomber) sweep wing ball actuator place smaller balls of plastic balls between the steel balls. I was not part of the design team but engineers shared information. This is what I remember since this was decades ago. Note it sat on a cart with the wheel base of a VW Bug. I walked past one of the test stands for months every day.

 

[ninefinger]

Hobby grade rolled ball screws with preloaded ball nuts are very common today. Readily available (China origin) and relatively cheap. Look for DFU1605 for example (D for double nut, 16 for screw major diameter, 05 for 5mm screw pitch). The added friction is not important especially compared to that from a lead screw. The benefit of eliminating backlash is huge. Being able to mill in either direction is a major victory, and also the lack of lost motion or sudden changes in amount of motion enable smaller cutters that just can't tolerate an increase of even 1thou in chip load. I currently have this issue (backlash) as my machine I built it ~13 years ago before it was "easy" (read also cheap) to get preloaded ball screw. I can't use cutters under 3/16" without very careful toolpath planning even with backlash compensation (where the stepper or servo motor moves more to accommodate the backlash). One day I'll update the ball screws (or ball nuts), but for now I just stick to big cutters that are more tolerant of a few thou backlash.