a dual use CNC and manual milling machine project - something learned

[aninventor]

I first posted this on another site and was told it would be more appropriate to post it here - so I am.
This is the original post and some responses to questions asked.
Only my posts are reproduced here.

My Wells Index 520 CEN mill has now been retrofitted with a 3 axis step motor based CNC system and is fully operational as a 3 axis CNC mill. It has also had manual handles and indexed dials and table locks and etc installed as required for manual machining.

The step motor drivers are half step, the NEMA 42 step motors are 200 full steps per rev, the ball screws are 5 TPI or 0.200 advance per rev, and the cog belt drive ratio is 2:1. This yields a step increment of 0.00025 which is quite satisfactory for my needs.

A big problem now arises when attempting to use the machine in manual mode. These NEMA 42 step motors have considerable residual magnetism in their magnetic poles and generate considerable cogging torque when rotated manually. As a result, the axes can not be set to any points other than ones where the physical magnetic poles of the motor are in alignment. A 200 full step per rev motor has 50 physical poles/rev. With a 2:1 cog belt ratio and a 0.200 inch advance per turn of the lead screw, those 50 poles align every 0.002 inch. You can not reasonably set an axis to anywhere other than those points where the motor poles align, and that only occurs every 0.002 inch of travel. That is clearly not acceptable given that I expect to be able to set an axis to 0.0005 inch as I currently do with a DRO on an all manual machine.

So, my conclusion - you can not expect to build a dual use manual and CNC machine using reasonably large step motors for the axis drives!

As a result, I have removed the step motors and will start over again with DC servo amps, DC servo motors, tachs, and encoders.

Before you get too carried away with any criticisms - this is a retirement hobby project and while I have a need for such a machine I am doing this for the challenge of doing it as much as anything - not because I need a mill to make customer parts with. I enjoy manual machining and I do not want to give that up, but I sometimes need to make parts that require a CNC mill. Next year I hit 70 and I am seeing benefits to downsizing and I do not want to have two machines if I can get by just as well with just one. So this approach fits my particular needs. As an engineer I am also pretty embarrassed that I did not anticipate this problem much earlier.

I am still convinced I can make a no compromises dual use machine, but I will have to do it using DC servos.




The Wells Index 520 CEN is a CNC machine and has ball screws as manufactured. This one was bought without any CNC motors or electronics as they had all been removed by the former owner who used them to retrofit a much larger mill.

Setting the table locks lightly gives enough friction to simulate a normal manual mill with the friction of normal screws and nuts, so that has not been a problem. I could have provided a screw with a brass tip pressing against the lead screw shaft itself to provide friction but that has not been necessary. If I see any indication of increased wear on the ways/gibs, that solution would eliminate it and would be easy to implement.

So far as disengaging the cog belts - not much room in the motor mount castings for that. The periphery of the cog belt pulleys is up against the sides of the castings. Doable? Perhaps, but not easy at all. I considered it but rejected it. Besides, I think there are some benefits to DC servos and I have most of the parts required.

So far as why would anyone want a dual use machine: I actually enjoy turning the cranks and doing manual machining the way I was taught to do it 50 years ago. It is not because I can not use a computer or a CNC machine - I was designing computers and computer controlled machines before most of you were working - it is because I am retired and I want to do what I enjoy the most and do it in the manner that is most pleasurable to me. Manual machining is a very different experience than programming and running a CNC. It is a bit like hunting with a bare recurved bow and broad head arrows instead of a scoped semi-auto rifle and mag ammo. I have done both. Both have their time and place. They are very different experiences.

Also, retirement and living on a fixed income has certain realities. A manual machine is much less expensive to own and to operate and that can be important in retirement. There is also a certain security in knowing that my manual machine will probably still be fully functional longer than I will be. CNCs are much more failure prone. Getting one fixed can set you back quite a bit. So, my CNC must be something I can troubleshoot and maintain myself and something I can expect to get cheap parts for when parts are required. Commercial CNCs with proprietary electronics and software do not fit that description very well. So, it is do it yourself time - and I am learning a great deal in the process.



Some stepper motor theory:

There are three cases to consider when forcing a step motor to rotate by mecahnically turning it. They are:

Case 1) the stepper motor leads are not connected to anything ---- in this case the winding circuits are open and there is no path for current to flow. Since there can be no current flow,ther can be no power generated and no electrical power dissipated and no net work is done in turning the shaft to generate electrical power. The windings do cut through the residual magnetic fields of the step motor physical poles as it is turned and a high voltage is generated across the open leads. Since there is no current flow in the windings there is no magnetic field generated in teh windings and no torque produced by the windings. There are magnetized pieces of iron in the poles of the rotor and stator and these do attract each other and this attraction does produce a cogging torque which you can feel as you rotate the motor shaft. Each pole produces an attractive torque pulse and a cogging action as it attempts to align. This is the source of my problem. The unconnected step motor prefers to stop in certain positions and stay there. I will turn away from intermediate positions and return to a position where the internal poses are in alignment. You can not position it anywhere you want it and let go - it will snap back to where it wants to be. That is a problem.

Case 2) The stepper motor leads are connected to electronics but the electronics is not powered ---- in this case the motor windings are cutting through the residual magnetic fields and a voltage is being generated across the open leads. The magnitude of this voltage depends upon the strength of the residual magnetic fields, the speed of rotation, and the impedance of the electronics it is attached to. The higher the impedance the higher the voltage and this voltage can be many thousands of volts, enough to damage electronics. It is relatively easy to design the electronics to limit these generated voltages to safe levels and avoid this problem by giving the induced current a safe path to flow through. The generated power is safely dissipated in the electronics itself. It takes torque to generate this power and this torque is added to that required to over come the cogging action discussed in #1 above. When the motor is stopped, there is no torque generated by the windings as they are no longer moving through a magnetic field and we are back to case 1.

Case 3) The stepper motor leads are connected to electronics and the electronics is powered ----- What happens in this case depends on how the circuits are designed but generally the driver reduces the step motor drive voltage and current to holding levels when the axis is not being moved. The holding voltage/current values are chosen to produce a certain design break over torque that the motor can resist while holding position and still not over heat. If this torque is exceeded, the motor will turn through a full set of steps to the next aligning position, typically 4 full steps. The torque required to make the motor rotate is much greater than in case 1 above because the magnetic fields generated by the motor drive current are much higher than the residual fields of an unpowered motor.

So yes, removing the power or the electrical connection from the motor to the driver will allow a step motor to be rotated with much less effort. Doing so does not eliminate or even reduce the cogging torque discussed in 1 above, and that is the source of my problem. The bigger the step motor is the greater the amount of iron in it and the greater the residual magnetic fields in all that iron and the greater the cogging torque generated when trying to force it to turn when it is unconnected. There is no solution other than to use another type of motor.

I have both Baldor and Siemens DC servo motors. DC servo motors are designed very differently as compared to step motors and they rotate very smoothly. They are the solution to my problem.

Because DC servo motors do not have any preferred position and will not hold any position on their own, the control system must have feedback loops that measure actual position and control the drive voltage/current to the motor to actively maintain the desired position. Most of these DC servo motion control systems also include a speed sensor (tachometer) and close a second feedback loop on speed as this gives smooth acceleration and more accurate position control.

So, if you want a dual use manual and CNC machine you had best use DC servo motors to build it.




Finally, I have no further use for the New NEMA 42 step motors themselves or Wells Index CNC controller - so they are for sale

Similar Threads:

• Milltronics Partner II - CamSoft CNC milling Machine project
• CNC Milling Machine Retrofit Project - SOLD
• Sieg X3 Milling Machine CNC Conversion Project
• New milling machine project
• My Milling Machine Project

[awerby]

Have you ever tried using the MDI (Manual Data Input) function for machining? It's a lot like manual machining, but instead of cranking, you simply enter the endpoint of each move, push a button, and it goes there. It's good for jobs that don't really call for a program, and it's more accurate than hand-cranking, since you don't overshoot. Plus, you can tell it to move diagonally, or in an arc, and it will do that much better than even an etch-a-sketch master.

I would advise against tearing off the stepper system you've put together, and to learn to use it instead. Or sell the whole thing, mill and all, and buy yourself a manual mill, if that's what you really like to use.

[Jim Dawson]

Your conclusions are absolutely correct. Brushed DC servos are the way to go for a CNC/manual machine. They operate nice & smooth. This is exactly the system I have on my machine. I have not had to lock the table for manual use, seems to work fine. Now for extra credit, make the Z axis (quill) CNC/manual with a flip of a lever. Mine does that.

I recommend glass or magnetic scales on the table for encoder feedback, with a preference for the mag scales, rather than encoders on the leadscrew/motor.

Now you just have to choose the motion controller and CNC software that you want to use that will operate as a closed loop system.

Looking forward to seeing your progress.

[aninventor]

Your conclusions are absolutely correct. Brushed DC servos are the way to go for a CNC/manual machine. They operate nice & smooth. This is exactly the system I have on my machine. I have not had to lock the table for manual use, seems to work fine. Now for extra credit, make the Z axis (quill) CNC/manual with a flip of a lever. Mine does that.

I recommend glass or magnetic scales on the table for encoder feedback, with a preference for the mag scales, rather than encoders on the leadscrew/motor.

Now you just have to choose the motion controller and CNC software that you want to use that will operate as a closed loop system.

Looking forward to seeing your progress.

__________________________________________________ __________________________________________________ __________________________________________________ _________________
If this is not the best place for this thread - let me know and I will move it.


I have a set of Newall sealed magnetic DRO scales and read heads with 0.0002 resolution and a 2 axis display, but the RS232 data output rate is not high enough to be used in a CNC feed back loop.
I also have a set of Accurite glass scales and read heads that are quadrature output that I could use with direct TTL inputs to the CNC computer, but their resolution is only 0.0005
I have a set of 3 Baldor DC servomotors with encoders and tachometers internal to the motors that are usable.
I also have a set of 4 Siemens DC servomotors with internal tachs and external encoders and one with an electric brake that I could use.
I am not yet sure which to use.
The Baldor motors are imperial so the mounts and pulleys are bolt on. The Siemens motors are metric so new motor mounts and reworked or new pulleys are required. The Siemens motors seem to be much more substantial - they are certainly much heavier. Any comments on the two different brands quality and reliability wise?

I have a set of Glentek servo amps from a Centronics CNC and a set of Servodynamics amps from an old Dynapath 10. The Glentek amps are much newer and smaller and would seem to be the better bet.

I am very opposed to Windows as there is no effective real time operating system there - or at least so I understand - and I am opposed to MS practices regarding upgrades and updates and op system obsolescence etc. So, I intend to go with a Dell Optiplex tower PC running Linux Ubuntu and LinuxCNC. I am a Linux newbie but an old machine language and DOS guy so it should not be too terrible of a learning curve. I am unsure of what interface board to go with for the DC servo amp analog speed signal generation etc. and for all the position and speed and digital input and output control signals. I have been looking at MESA and sent an inquiry to them but got absolutely no response from them.

Some local guys are big into the Arduino type controllers and highly advocate them, but I think the learning curve and programming work load of the Arduino is not justified for this project. I think it is a better idea to purchase a known good board from a reliable vendor that works with available and tested and proven software.

So far as removing the step motor CNC system - it is already a done deed. The step motors are off and for sale as is the old 1980 vintage Index (Bendix - Summit - bandit) CNC controller. That controller had limited memory requiring drip feed via RS-232 for contouring. It was an open loop system. I can do better. It did have the ability to enter and edit a single line of G code and then execute it conveniently, and it had manual axis jog and travel controls too - so I could do somewhat manual like machining with it and get something done. To be blunt - I would very much prefer to just turn the cranks and watch the DRO and get on with it. And yes - I have used a MPG and I do not like trying to machine with that as the input control device. I have and will include a MPG in my final system but it is not my preferred means of control for machining. If LinuxCNC includes a good enough conversational interface, I may find that handy.

I also have a tower PC with legal Mach 3 and a BOB and Gecko drive installed. The Gecko will not drive the NEMA 42 steppers and I do not like Windows. I had considered building an interface between the older direct winding control inputs of the older Bandit system and the newer step and direction control signal outputs of Mach 3 and a BOB to get the software and memory upgrades, but the problem of step motor cogging makes that a bad option for me. Also, while Mach 3 seems fine requires Windows as the operating system so I am rejecting that option. This PC and Mach 3 and the Gecko drives etc will be sold. So will all the rails and carriages and steppers that came with it as part of a big foam cutter.

I am now committed to a dual loop DC servo system.

Now I need to get busy and get clever and figure out how to integrate full manual spindle down feed with the DC servo CNC system. First step is to get the spindle under DC servo control. I have a servo motor with an integral electric brake. Current thought is to use that motor as a power feed motor with a spring centered manual pot to control feed speed and direction up/down and use the electric brake to hold the spindle in position. Then design and build a fully manual spindle feed and install that at my leisure. When I have the other axes DC motors and sensors etc. installed and the computer and software sorted out I go for a full 3 axis system.

And just to put the Manual vs CNC debate to rest - IT IS MY CHOICE AND MY JOY TO DO MANUAL MACHINING. It is an entirely different experience than loading a program and watching a macine run. It is definately not an aversion to new technology. I am a Physicist and Mechanical engineer by educatin, a licensed professional engineer (PE), and was designing computers before most of you were born. I was on the design team at HP that did the first ever basic language desk top (HP 9830), the laser printer, and the first ever work station (HP9845) in the early and mid 70's - all well before the Imsai Altair, Sinclair, Commodore PET, TRS 80, Apple, Byte Mag, Bill Gates, Steve Jobs, or the IBMPC. In the late 60's I dd the dynamic modeling and control algorithms for 4 leg walking vehicles for ARPA - the fore runner of DARPA. When it comes to manual machining, I like turning the cranks and feeling the cutter bite into the work and I like the smell of cutting fluid curling up from blue chips. It is joyous work.

But - I need some complicated parts that I can not do manually - so I am also going to CNC. I am looking forward to the technical challenge of it all.

All help and advice is appreciated.

[aninventor]

Now for extra credit, make the Z axis (quill) CNC/manual with a flip of a lever. Mine does that.

_________________________________________

Exactly how did you do that?

[Jim Dawson]

_________________________________________

Exactly how did you do that?

Thought you'd never ask http://www.cnczone.com/forums/knee-v...-software.html

I have had very good results with the Baldor/SD setup. In fact I'm running mine on a 75V power supply with no loss of performance so they should last for ever.. They will rapid at >200 IPM, but I limit to 100 IPM, I'm not in that big of a hurry. I assume your SD drives are 1525s. I'm running the DC servos on X and Y, and a NEMA34 stepper on the Z in closed loop velocity mode, looks like a servo to the motion controller.

As a machine control OS, Windows sucks. But it does make for a nice operator interface. I just offload all of the heavy lifting to the motion controller. I can surf the net while running a job (not recommended ), but it behaves as it should.

I'm using Renishaw 1 micron mag scales on all axis, but am switching over to Ditron for my next projects since I can't source the Renishaw LM10 takeouts any longer. I have done a side by side comparison of the Renishaw and Ditron read heads, and found the Ditron to be just as accurate.

It sounds like you and I have somewhat similar backgrounds, the difference being you went the academic route and I went to the school of hard knocks. About the same age also, I turned 68 a few months ago.

[aninventor]

Well, that is an interesting way to do the Z axis spindle down feed drive. I would not have thought that you could hold 0.0001 with non-precision gears - ie not ground or honed or shaved - even with the gears preloaded together.

The idea of using the step motor in velocity mode as a servo is interesting as well - something I will consider.

My spindle axis problem is very different from yours. My machine was CNC and already has a ball screw in place. There is no easy way to get a hold on the ball screw itself, so the position encoder will have to go on the motor shaft or inside the motor itself and will be operating through the cog belt and cog belt pulleys. I am concerned a bit about the accuracy of all that.

This mill does not have a pinion engaging a rack on the spindle ram for manual down feed and it does not have a manual spindle travel lock either, so providing the fine and coarse manual down feed functions and holding the spindle in position while milling is going to be interesting from a design standpoint. I am thinking that the manual feed needs to be easily disengaged when operating in CNC, especially the fine feed when in rapids. Some suggest using the CNC's Z axis for manual spindle down feed and position control by providing a manual dial to control it, but that eliminates any manual force feed back for operations like drilling and tapping and I find that unacceptable. So, I have some serious design work to do. I have managed to get an indexed dial onto the Z axis ball screw end so I now know where the spindle is without the CNC 's DRO function operating. This dial can also be manually turned for fine positioning of the spindle but it is not convenient. I want fairly normal fine and coarse down feed handles with normal force feed back, and some precision down feed stops as well, and those too will have to be easily disengaged for CNC operation.

Where do you get your micron scales and how much do they cost?

[James Newton]

Really good post with lots of detail. Question: How would you feel about a jog function in the form of hand cranks? E.g. replicate the actual "hands on" control with cranks mounted to encoders which send keyboard commands to the control software to make the CNC machine move just as if you had directly moved it. The limitation would be 1. possibility reduced resolution (or actually, possibly /enhanced/ resolution) 2. no force feedback. e.g. you would not be able to feel the machine moving from the cranks. I'm asking this to see how many people would be interested, because it would be relatively easy to make as a kit.

[Jim Dawson]

Well, that is an interesting way to do the Z axis spindle down feed drive. I would not have thought that you could hold 0.0001 with non-precision gears - ie not ground or honed or shaved - even with the gears preloaded together.

I was surprised and quite pleased with the accuracy and repeatability, my target was +/- 0.0005. The gears have very little, if any, effect on the accuracy, they are only preloaded to remove backlash so that there are no tracking inaccuracies. The positioning is done with the mag scale.

My spindle axis problem is very different from yours. My machine was CNC and already has a ball screw in place. There is no easy way to get a hold on the ball screw itself, so the position encoder will have to go on the motor shaft or inside the motor itself and will be operating through the cog belt and cog belt pulleys. I am concerned a bit about the accuracy of all that.

Why not attach the reader head to the ball nut, and affix the scale to the housing? (far left in the picture) That is how we did it on the Shizuoka AN-S upgrade, everything fits neatly inside the housing.

http://www.cnczone.com/forums/attach...d=358404&stc=1

This mill does not have a pinion engaging a rack on the spindle ram for manual down feed and it does not have a manual spindle travel lock either, so providing the fine and coarse manual down feed functions and holding the spindle in position while milling is going to be interesting from a design standpoint. I am thinking that the manual feed needs to be easily disengaged when operating in CNC, especially the fine feed when in rapids. Some suggest using the CNC's Z axis for manual spindle down feed and position control by providing a manual dial to control it, but that eliminates any manual force feed back for operations like drilling and tapping and I find that unacceptable. So, I have some serious design work to do. I have managed to get an indexed dial onto the Z axis ball screw end so I now know where the spindle is without the CNC 's DRO function operating. This dial can also be manually turned for fine positioning of the spindle but it is not convenient. I want fairly normal fine and coarse down feed handles with normal force feed back, and some precision down feed stops as well, and those too will have to be easily disengaged for CNC operation.

That is going to present some interesting design challenges. Like you, for myself, force feedback when manually drilling and other operations is mandatory.

Where do you get your micron scales and how much do they cost?

Currently buying from Ditron. I'm using their DMR-200 read heads. You won't find those anywhere on the internet yet, it's a new product for them. Last quote was $90 for the read head, and $28/m for the 2mm mag tape. This combination gives you 1 micron resolution. I have attached a product brochure, but the published specs make no sense. For specifications, look at the Renishaw LM10-001 series mag readers. They are identical.

[aninventor]

James

I probably would not like that as opposed to real hand cranks because there
is no feel - no force feedback, but for step motor CNC machines that cog -
and there are a lot of them - it would be better than MPGs or the keyboard
or etc. - so maybe you have a winner.

Now - I will give you an idea - a crazy one - and I want credit for this
original Idea but feel free to use it or sell product based on it if you
simply add a note that it came from me.

As a pilot on an ILS approach to landing I can fly an aircraft at 120 mph
through a slot about 15 ft high and 50 ft wide with nothing to look at but a
pair of crossed needles telling me that I am high or low or wide to either
side. I can usually keep it within a few feet of centerline. I bet I could
turn milling machine and lathe handles to very fine tolerances with similar
guidance. If I had a screen display of a tool path and a cursor moving along
the path and deviating up or down depending on the error as measured by the
DRO, I could engage power feed and watch the moving cursor and see if I
needed to crank in or out on the cross axis. CNC results without all the
motors and amps and power supplies and etc. No need to modify your manual
mill or lathe either. Just plug a new box into your DRO cables. I wanted to
try it long ago but never found the time.

You could then do anything graphic that you can digitize without trying to
get to G code. Just keep the cross hair on the edge while the cursor moves
along. It is a video game that cuts metal


Lee

[Jim Dawson]

James

I probably would not like that as opposed to real hand cranks because there
is no feel - no force feedback, but for step motor CNC machines that cog -
and there are a lot of them - it would be better than MPGs or the keyboard
or etc. - so maybe you have a winner.

Now - I will give you an idea - a crazy one - and I want credit for this
original Idea but feel free to use it or sell product based on it if you
simply add a note that it came from me.

As a pilot on an ILS approach to landing I can fly an aircraft at 120 mph
through a slot about 15 ft high and 50 ft wide with nothing to look at but a
pair of crossed needles telling me that I am high or low or wide to either
side. I can usually keep it within a few feet of centerline. I bet I could
turn milling machine and lathe handles to very fine tolerances with similar
guidance. If I had a screen display of a tool path and a cursor moving along
the path and deviating up or down depending on the error as measured by the
DRO, I could engage power feed and watch the moving cursor and see if I
needed to crank in or out on the cross axis. CNC results without all the
motors and amps and power supplies and etc. No need to modify your manual
mill or lathe either. Just plug a new box into your DRO cables. I wanted to
try it long ago but never found the time.

You could then do anything graphic that you can digitize without trying to
get to G code. Just keep the cross hair on the edge while the cursor moves
along. It is a video game that cuts metal


Lee

You mean something like this? https://shapertools.com/ more or less a flight director for a router

[aninventor]

Jim Dawson

Thing is all my salvaged CNC encoders are rotary - so I had a rotary mind set trying to keep costs down by using what I have on hand.

Earlier in this project I had looked into using the linear Newall DRO scales and read heads I have but the DRO's RS-232 output was way to slow to be of any use in a CNC setup closing loops etc. As of right now I have no idea what type signal the Newall read head puts out so I do not know if I can use just the scales and read heads and send their output directly to the CNC computer. I have some research to do there.

Later, I discovered by accident that the Accurite glass scales and read heads I have output digital quadrature signals just like rotary encoders do and are pretty much interchangeable with rotary encoders. Thing is I don't have any short ones. Can you just cut down a long glass scale when you need a shorter one?

If I start thinking linear scales the quill position sensing problem pretty much solves itself, so it is time to gather info on linear scales, their outputs, and direct inputs to LinuxCNC. Thanks for that head change. I will look at the scales you are using.



Lee

[Jim Dawson]

Jim Dawson

Later, I discovered by accident that the Accurite glass scales and read heads I have output digital quadrature signals just like rotary encoders do and are pretty much interchangeable with rotary encoders. Thing is I don't have any short ones. Can you just cut down a long glass scale when you need a shorter one?

It has been done successfully, but it is glass so there is always some risk.

If I start thinking linear scales the quill position sensing problem pretty much solves itself, so it is time to gather info on linear scales, their outputs, and direct inputs to LinuxCNC. Thanks for that head change. I will look at the scales you are using.

Lee

I'm a real fan of linear scales on CNC machines. We did a test cut on the Shizuoka AN-S project on a 4 inch diameter interpolated pocket, with a 4.5 inch OD. When checked with a 1 micron resolution CMM and checked at 12 points, there was zero derivation on the OD, and zero deviation on the ID except in one spot that was +0.0047'' where there was a tool mark. This machine is running Baldor DC servos and SD 1525 drives and Renishaw LM10-001 mag readers.