Ballscrews and CNC conversions

It seems to be a given that a mill needs ballscrews for a CNC conversion. I'm wondering why. Is it to eliminate backlash or to make the machine crank easier? I have a knee mill and I was thinking of the following modifications in this order: fit DRO scales and readout, Fit DC motors to X,Y, and quill, close the loop with a microcontroller.

Why would I need to use ballscrews in this application? The system would be closed-loop and the backlash would be compensated.

Reply to
Jim Stewart
Loading thread data ...

I think speed/efficiency is an important to think about. Ballscrews are so efficient that if you hold one vertical, the nut should (under its own power) spin down the rod, no problem. If you've got a powerful spindle motor, and you want to use your machine to its highest capacity, you want to use it all - which means efficient axis drives.

Also one thing about ballscrews is that I don't think they wear/break very easily, especially compared to a system using soft nuts. If you crash your machine with bronze nuts, you may strip/damage them (a fair amount of time and money to get things running again). With a ballscrew, your cutter would probably break or the motors would stall (or the head/ram knuckle joint would give) before the screw would fail.

!These thoughts are worth as much as you paid for them.!

Regards,

Robin

Reply to
Robin S.

Think about when the direction of cut changes and rigidity is suspended until it hits the other side of the threads. Cutters do strange things if you let them wander.

Reply to
Tom Gardner

Agreed. But it can be handled the same way a human handles it with leadscrews can't it?

Reply to
Jim Stewart

Jim Stewart wrote in news: snipped-for-privacy@omsoft.com:

Just imagine you are pocket milling what is supposed to be a nice round hole. What happens when the controller starts changing directions in one axis while still feeding in the other? (Chatter, broken tool, or bad part). Most controllers will only compensate for 0.25mm (0.010") of backlash, after that, all bets are off. If the backlash is too bad, with linear scales, the drive will fault because the following error is too high. (the controller tells the motor to move, but doesn't see any movement in the slide, after X amount of motor movement, it faults thinking there is a problem.) I would put ballscrews in it.

Reply to
Anthony

I'd say no, after watching machines run - esspecially at the speeds CNC's run. People normally creap up when stoping the cutter within the materail (as opposed to going beyond the edge of the material). This means the work probably isn't being ramed into the cutter. On a CNC, the motors/screws are capable of accelerating and decelerating MUCH faster than the table which could (and probably would) cause the table to float into the work. At this point, there is no resistance from the nut so your cutter would be at the mercy of the materail it's cutting.

This condition could be overcome by spring-loaded antibacklash nuts, but this causes other problems. These nuts are of course spring-loaded. This means their antibacklash capability can be defeated if there is enough pressure from the drive motors and/or there is enough resistance from the pressure of the cut. At this point, perhaps the preload can be increased.

Unfortunately this causes wear on both the screw and the nut. Ballscrews experience virtually no sliding action, just rolling, so this pressure is not as big an issue. In fact, this is how it's done on real CNC (AFAIK), not only to prevent backlash, but also to prevent the loss of support should one nut lose its balls (seriously) which would be a very significant issue for the z-axis drive.

Regards,

Robin

Reply to
Robin S.

Without the frictional drag, you might start to take a cut, pushing the table in a direction the lead screw has slack in. The computer sees this, starts the motor, the motor takes a bit of time before it slams the table in the other direction. OOPS too far. The motor starts to go the other way, except, it now has to take the slack out in the other direction. OOps too far again the other direction.

It might make less than perfect circles, at the four corners of axis reversal.

I have several cnc machines, one has an EZ Trak conversion on it. The EZ Trak is speced for no more than .005" hysteresis (backlash) in the HiWin ball screws, (including thrust bearings). It has encoders on it just as you are contemplating,and ball screws, although not real tight ones.They have maybe .002" backlash total, and are compensated by the encoder scales.

Watching it work is, interesting.... If you push on the table, you can watch one axis handle wobbling as it goes back and forth against the backlash, trying to keep the table where it belongs. The scales are 5 micron (approx .002") scales. Al scales that I have seen are really metric, the readouts convert to inches, but the native units are in microns. Surely there is an exception to this, I ahve not seen one.

Enough rambling... Pete

Reply to
Pete Logghe

It would work, but require bigger DC motors. Or if the motors are big enough, using ball screws will let it move more rapidly.

Dan

Reply to
Dan Caster

When you're cutting something like a circle, at some point the backlash will let the work move into the tool suddenly to take it up. You can end up with a not quite perfect cut, or a broken tool, depending on how bad your screws are.

Not cutting circles? how about pockets? Those are so common most machines have those as a canned program.

Even if your screws are good to start, they can get a lot of wear once you start using the machine as a CNC. While it's pretty tiring to crank those handles all day, it's a lot of fun to watch the computer twiddle them.

Unless you're doing it for fun, you can generally find an old CNC machine with a dead controller for less than you'd pay for the ballscrews. I found a Bridgeport knee mill with servos, ballscrews, toolchanger and a VFD drive on the spindle in great shape for $1100. I think that was out of a training facility that got shut down. Most of the machines got moved to California; they didn't have room on the truck for this one.

Check out the Cad_cam_edm_dro group on yahoo. They have lots of info on building and converting machines.

Reply to
Paul Amaranth

It's not an absolute need.

But the ballscrews are 90 percent efficient versus 50 percent for acme, so you need roughly twice as much motor and twice as much power supply for the acme screws, enough that the expense of the ballscrews.may be less, given the same performance.

"Performance" here means not just table slew speed, but cutting force. A slow table is just a nuisance, but a drive system that can't keep up with the cutting tool resistance will stall and ruin the part.

You can compensate for backlash in software, but you still get several

0.001" of "noise" when you do so. This "noise" is much worse than the overall machine accuracy, not just because of the slop (loss of rigidity) during backlash reversal, but the fact that the backlash itself varies many thousandths from place to place on the screws. It is a lack of tolerance that is necessarily much worse than the rest of the machine's accuracy. Screw pitch accuracy does not imply repeatable backlash. Machine wear won't spoil screw pitch accuracy, but it will increase and randomize backlash.

More on the economics and engineering of a CNC drive at the end of my page:

formatting link
Picture of a part showing the "noise" from cutting a curve with backlash compensation on acme leadscrews on my page:

formatting link

Reply to
Richard J Kinch

One important consideration that hasn't been mentioned is that a system using servos will need to be detuned if the connection between motor and load has any backlash or is too springy. The optimum tuning of the servo loop is a function of the load characteristics, including mass, stiffness, friction, damping, etc. The backlash introduces two load conditions; first, where the lash is taken up and the motor 'sees' the load as long as it continues to apply torque in the same direction, and second, where due to the backlash, the motor can rotate a small amount without applying any force to the load.

Tuning the servo loop for optimal stiffness and response for the first case will usually result in violent vibration when the lash isn't taken up. I've seen couplings and even motor shafts break as a result of this.

None of this applies to steppers.

Ned Simmons

Reply to
Ned Simmons

The PID loop in the controller should manage the first, and the software the second, then there is no instability.

Reply to
Richard J Kinch

The PID loop *is* the servo loop (though there are other feedback schemes besides PID). The tuning of the servo loop can accomodate backlash and insufficient mechanical stiffness, but at the expense of reduced performance. In other words, the stability is paid for by sacrificing stiffness and response.

This is the reason that preloaded ball screws are used in positioning systems where high performance is required. Increased mechanical efficiency is usually a secondary concern.

Ned Simmons

Reply to
Ned Simmons

A good example of such is on the OmniTurn CNC lathes. The ballscrews are considered badly worn when there is .0006 backlash.

gunner

The two highest achievements of the human mind are the twin concepts of "loyalty" and "duty." Whenever these twin concepts fall into disrepute -- get out of there fast! You may possibly save yourself, but it is too late to save that society. It is doomed. " Lazarus Long

Reply to
Gunner

I have a Shopmaster machine that we added our own home grown CNC stepper system. The Shopmaster has spring loaded acme nuts standard and we first used them with good results, but speed was not fast and in circular interpolation we got small " stairstep" cuts. It seems that the backlash compensation works best in linear moves, but in circular its not so efficient. We changed to ball screws and saw a big leap in speed and much less "stairstepping". We also found that the stepper systems are far mor efficient if you have micro stepping in your software. Ball screws work fine by hand, but have a funny springy sort of feel when you put cutting loads on the handle.

Reply to
DonPowell

Got any "badly worn" ones you'd like to sell?

Reply to
Jim Stewart

Hummm I have a few. They only have about 14 inchs of travel however. They do include the nut.

Gunner

The two highest achievements of the human mind are the twin concepts of "loyalty" and "duty." Whenever these twin concepts fall into disrepute -- get out of there fast! You may possibly save yourself, but it is too late to save that society. It is doomed. " Lazarus Long

Reply to
Gunner

Like I said, design the system to separate the backlash compensation from the servo loop and give it instead to the software, then you remove the instability in the servo loop instroduced by the backlash. The design should not burden the servo loop with backlash compensation.

Reply to
Richard J Kinch

A linear move by definition does not involve backlash, so it is not surprising that you didn't see a detrimental effect from backlash.

The stairstepping could be from non-linearity in the acme threads, not necessarily from backlash. Backlash problems should only appear at curves passing through 90 degree tangent angles. Ballscrews may have solved both problems for you, but for different reasons.

Reply to
Richard J Kinch

Backlash compensation in the positioning software won't do anything for servo loop stability. The problems with stability result from the fact that the motor is effectively uncoupled from the load when the nut and screw are in a position where the motor can rotate slightly without moving the load. This causes two loading conditions for the motor, coupled to the load, and uncoupled. Tuning can only be optimized for one condition.

You can simulate this by tuning a servo motor on the bench for max stiffness and response, with no load attached to the shaft. Then pick up the motor by the shaft; this has the effect of changing the load inertia from the inertia of the rotor to the inertia of the frame. If you disturb the shaft position a small amount the motor will start yawing about the zero position, violently if enough torque is available, as it tries to stabilize at the commanded position.

Ned Simmons

Reply to
Ned Simmons

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.