Metric/CNC conversion of lathes/mills

[Oscar Lazo]

I would think that the decreased friction would offset the increased torque required because of the lower mechanical advantage. Generally, manual operation of a CNC mill isn't recommended although there are those who have. A search on this forum will turn up a few.

I run my mill manually,more often than with CNC by using a jog/shuttle. It's not ideal as there is no biofeedback and motion is in steps but it works. The CNC conversion kit for the mill that you mentioned uses a 5mm pitch ball screw so one rotation of a hand crank is 5mm of travel as compared to a more typical 2.5mm for a lead screw on a small mill.

I see. I agree that the biofeedback is the nice thing about the cranks! From what I am gathering, it seems to me that the best way of doing the thing I want would be to get the DRO signal to direct the CNC mode and to keep the leadscews in order to prevent backlash issues and loose axes when the holding current is absent from the steppers. What would you say is the main issue preventing the DRO to drive the CNC? Perhaps common DRO's don't usually have the necessary electronic outputs since it is meant only for display?

 

[RJSakowski]

Let me state up front that I am not a programmer and there are others on this forum who are far more qualified than me.

I would envision a stepper driven CNC would establish a reference datum and would issue commands to the stepper motors as dictated by the g-code. On the sideline, the DRO would be tracking the position as determined by the g-code and compare it with its actual position and generate any necessary error signal. to the CNC controller. It would have to make corrections consistent with the commanded velocity as well. The Path Pilot controller in my CNC does a look ahead to anticipate changes in direction and maintain a uniform feed rate. The Mach 3 controller that I had used previously did the same.

 

[Oscar Lazo]

Let me state up front that I am not a programmer and there are others on this forum who are far more qualified than me.

I would envision a stepper driven CNC would establish a reference datum and would issue commands to the stepper motors as dictated by the g-code. On the sideline, the DRO would be tracking the position as determined by the g-code and compare it with its actual position and generate any necessary error signal. to the CNC controller. It would have to make corrections consistent with the commanded velocity as well. The Path Pilot controller in my CNC does a look ahead to anticipate changes in direction and maintain a uniform feed rate. The Mach 3 controller that I had used previously did the same.

Myself, I am a programmer, and I have a fair amount of experience with electronics. I am far less familiar with machining though. but I also expect generating a real-time error signal from the DRO measurement would be the gist of it. Perhaps there isn't an out-of-the-box implementation of that at the hobby level. Even expensive 3d printers still rely on step-counting, so perhaps that is not surprising. I guess they just rely on the ballscrews to get rid of the backlash, but that would be the issue with my leadscrew idea without the closed loop control...

 

[Huub Buis]

I have converted both my lathes to CNC without replacing the lead screws to ball screws. Yes, I can't do exotic forms (chess piece) with high accuracy, but I can do all things (and more) that a manual lathe can do with increased precision and ease.
I placed the Z-axis stepper in the gear box and the X-axis stepper at the back of the spindle. So I still can (and often do) use the lathes manual.

I don't want a ball screw on the z-axis because than I can't move the carriage by hand any more.
I though about replacing the X-axis lead screw for a ball screw but the X-axis lead screw has a left handed thread and left handed ball screws are very expensive.
I am also afraid that once a ball screw is fitted on the x-axis, I have to hold (lock) then handle when I run the power feed because a ball screw has not enough friction to keep the cross slide in place.
For accuracy, I don't think adding a top quality super accurate ball screw will make my parts more accurate. There are so many other things that influence the turned dimensions that the improved spindle accuracy probably won't be noticed at all.

 

[Oscar Lazo]

I have converted both my lathes to CNC without replacing the lead screws to ball screws. Yes, I can't do exotic forms (chess piece) with high accuracy, but I can do all things (and more) that a manual lathe can do with increased precision and ease.
I placed the Z-axis stepper in the gear box and the X-axis stepper at the back of the spindle. So I still can (and often do) use the lathes manual.

I don't want a ball screw on the z-axis because than I can't move the carriage by hand any more.
I though about replacing the X-axis lead screw for a ball screw but the X-axis lead screw has a left handed thread and left handed ball screws are very expensive.
I am also afraid that once a ball screw is fitted on the x-axis, I have to hold (lock) then handle when I run the power feed because a ball screw has not enough friction to keep the cross slide in place.
For accuracy, I don't think adding a top quality super accurate ball screw will make my parts more accurate. There are so many other things that influence the turned dimensions that the improved spindle accuracy probably won't be noticed at all.

That's encouraging to hear! How do you deal with the backlash though? Is it enough to just keep track of the last moving direction and substract a constant backlash when the direction must change? Is backlash the reason for limited precision? Or perhaps do you know of some DRO-based positioning system of the kind previously mentioned?

 

[Huub Buis]

How do you deal with the backlash though? Is it enough to just keep track of the last moving direction and substract a constant backlash when the direction must change?

To get the tool at the right position, you have to consider backlash and move the spindle (tool) a bit (the amount of backlash) further. But that is not all.
Suppose you want to turn, chamfer and face in one pass. You start at the chuck and move the tool (Z axis move) towards the tail stock. At the chamfer point, you also move the tool deeper in to the work piece (Z & X axis move). At the facing point you only move the tool deeper into the stock (X axis move). The z-axis spindle is still pushing the carriage towards the tail stock. And if you pull at the carriage, it will, due to back lash, move towards the tail stock. So if the facing forces are large enough, they will push to tool towards the head stock. To eliminate this you have to reverse the spindle the amount of the play before facing.
It is the CAM software that knows from what side the cutting forces will come and only the CAM software can deal with backlash. But that is not all.
Suppose you want to turn an outer taper and start at the smallest side (tail stock side). The carriage moves towards the chuck and the tool is slightly retracted to turn the taper. But during retraction of the tool, the X-axis spindle is pulling the tool and the stock is pushing the tool in the same direction. So the tool can be pushed by the stock the amount of play in the X-axis. So the CAM software must calculate the tool path so that the pushing direction of the spindle is opposite of the pushing direction of the stock.
The CAM software has to compensate for backlash and plan the tool path considering the cutting forces.

Is backlash the reason for limited precision?

A precision ground ball screw will be more precise than a lead screw. But it will only be significant when all other errors are reduced. That is not realistic for most hobby users.
CNC control can (re)position the tool very reproducible (same speed, acceleration, deceleration, only axial forces on the spindle). That can get the highest precision out of the setup. But if the setup is crappy, the results will still be crappy.
I can manual turn to tight dimensions but I can't beat the CNC. On my lathes, 0.01 mm is the tolerance I generally can get. Sometimes I am off without knowing why.

Or perhaps do you know of some DRO-based positioning system of the kind previously mentioned?

You could use a controller that uses a DRO to position the tool in a closed loop system. I think, Linux CNC can do this when using a Mesa controller. But even then, It only measures and positions the carriage and cross slide. Head stock misalignment, spindle runout, homing errors, tool wear, measurement errors, stock temperature rise, etc are still undetected.
I think buying a precision ground ball screw will be an easier and cheaper solution and if the ball screw is the largest source for error, precision could benefit.