Emco Unimat lathe CNC retrofit

[ZeroBacklash]

I tried mach3 and Linux CNC but i found it to complicated for simple jobs. Combined with CAM (Fusion360) it took more time to prepare than to turn. I looked at making an extension for Mach but the documentation was to confusing. Commercial controllers like USBCNC (Eding CNC) where to expensive to start with. So i start building my own CNC controller based on a CNC controller i made for the electronic clutch for my motorcycle. Then i stumbled over GRBL, it had all i needed and i must say I can't do it better. So i decided to use GRBL for controller. Combined with my own software i can do 'Conventional" turning the CNC way without CAD or CAM. For free style parts, witch i haven't made yet, you can still use CAM and one of the available GCode senders.
You can take a look at the software CNCL in the Microsoft store, download, try it.

Its amazing to learn from your experiences, so if I got you right you have a pc with parallel port and that goes to a grbl controller then a breakout board and then the drivers and the steppers. And you use custom software without going into the hassles of CAM.

What breakout board did you choose to go in for or did you choose to go in for? Also modern PC's dont have a parallel port so do you have a PCI card that you have inserted in you PC? Also are you running on linux or Windows?
Its fascinating that you could get a CNC up and running in as low of USD100.

[ZeroBacklash]

You can take a look at the software CNCL in the Microsoft store, download, try it..

CNCL seems to be good, so all I need to use it is a Arduino Uno connected to a PC (Windows 10). I already have the breakout board that I ordered from CNC drive.Drivers I can possible use the leadshine ones.
I have a few questions regarding this setup.
1. Do you have noise issues, I mean the arduino getting effected by noise from the motors or via USB and it behaving erratically?
2. How did you connect the spindle motor for threading, I believe you left the original spindle motor as is for normal lathe operations and you added another stepper for threading etc. So you connected both via belt drive to the final chuck shaft?
3. For spindle is your existing motor an ac induction motor?

[hfjbuis]

Yes, all you need is an arduino, some Dupont wires for connection and your leadshine drivers. The genuine arduino board is rated for 20V but the ebay boards are most times rated lower.
I don't have noise issues. Keep the arduino control signals and power supply away from the stepper A+A-B+B- and power cables.
Below is a picture of my control board with a ESP8266 for WiFi connection to my tablet, a small adjustable powersupply for the fan on the Y (C) axis, the arduino with a self made shield for connecting the arduino to the drivers, 3 TB6600 drivers and a power supply. The steppers are connected using 3, 4 pin , 16 mm aviation plugs.
On the picture below you don't see a genuine arduino but a self made arduino having a power regulator ( LM7805) that can handle the output of the power supply (24 V). On top there a self made shield for easy connection of the stepper drivers.

The original spindle motor, is used for all common turning tasks. For threading, i activate a 2 (4) Nm stepper by a lever (the stepper also turns de original spindle motor). This stepper turns the gear on the spindle. See the pictures below
Stepper mounted on top of the gearbox, with on top a leaver to engage the stepper


Stepper not engaged


The stepper is engaged, the gear is driving the spindle

You could also mount the stepper in the gearbox and use the tumbler that switches the spindle direction to engage the stepper.

The original spindle motor is a DC motor with variable speed control.
Any step/dir driver can be used for the GRBL controller, so yes, you can use your leadshine drivers.

GRBL has a good Wiki. Look at: https://github.com/gnea/grbl/wiki
The CNC conversion of my lathe is documented. All the info, drawings, pictures and documents are available at: Welkom op www.mwt.messageboard.nl

[skrubol]

CNCL seems to be good, so all I need to use it is a Arduino Uno connected to a PC (Windows 10). I already have the breakout board that I ordered from CNC drive.Drivers I can possible use the leadshine ones.

If you got a parallel port breakout board, it probably won't interface directly to the arduino. I'd look for a GRBL parallel port shield. Should be cheap.

[hfjbuis]

The problem is that parallel port interfaces are designed for Mach or Linux CNC where you can select de function for each pin. The pins in GRBL "can not be changed". A GRBL paralel shield will probably not work. However there are parallel breakout boards you can use to connect to GRBL using dupont wires. I found a YouTube video where it is explained.

From the previous post it looks like @ZeroBacklash has not yet purchased a BOB.

[hfjbuis]

If you consider using GRBL and need to buy an Arduino, consider buying this Arduino Uno version. It has a build in ESP8266 WiFi interface that looks ok. I have ordered one but not received yet. If i get it working, i am going to make a manual on how its done and will post it here.I have a working standalone ESP8266 module so i think it is no problem.
https://www.ebay.nl/itm/263178315394...m=263178315394

[ZeroBacklash]

Thanks @hfjbuis and @skrubol for your continued updates.
GRBL and arduino seems to have opened new avenues to CNC conversion. I already have some Arduino UNO boards and probably will have a ESP module as well.

I would need a breakout board, probably the one from CNC Drive will work(HDBB2 model).

For spindle is there a 1 motor solution that can handle both the threading / turning / facing operations?

I am planning to buy a bigger lathe probably a 4feet one, which apart from x, z, spindle I want to do some kind of auto tool changing, something like rotating the tool post. Will CNCL / GRBL handle this kind of operation?

[hfjbuis]

I would need a breakout board, probably the one from CNC Drive will work(HDBB2 model).

You do not need a breakout board. You can connect the driver to the arduino using some cheap Dupont wires.

Thanks @hfjbuis and @skrubol for your continued updates.
For spindle is there a 1 motor solution that can handle both the threading / turning / facing operations?

I haven't found yet and as long as GRBL doesn't implement threading you can't use the "standard motors" for threading. But GRBL is evolving. Currently development is target to easy implementation of other hardware. After that, threading is a highly requested feature.

I am planning to buy a bigger lathe probably a 4feet one, which apart from x, z, spindle I want to do some kind of auto tool changing, something like rotating the tool post. Will CNCL / GRBL handle this kind of operation?

Currently CNCL has implemented a tool library and toolchange. If GRBL supports a toolchanger, CNCL will support that to.
Toolchangee communication is not time critical. So toolchanger support could be implemented another way and is a feature that is on the CNCL todo list. As soon as you know what brand/type of tool changer you would like to have, let me know and i will look at it. I think today the solution will be in connecting a second GRBL controller connected to the motor of the toolchanger. In future this feature will probably supported by GRBL or one of the many GRBL compatible boards.

[ZeroBacklash]

I haven't found yet and as long as GRBL doesn't implement threading you can't use the "standard motors" for threading. But GRBL is evolving. Currently development is target to easy implementation of other hardware. After that, threading is a highly requested feature.

You got threading working in your lathe with GRBL right? So basically you have the x, z axis and the spindle all working in tandem during threading. How did you manage that with GRBL?

[skrubol]

You got threading working in your lathe with GRBL right? So basically you have the x, z axis and the spindle all working in tandem during threading. How did you manage that with GRBL?

I think that's why he has a stepper on his spindle, to allow for threading. To GRBL it's just another step/dir axis. I have no idea if/how good GRBL's rotary axis support is. Without some sort of support for rotary, the controller doesn't know that 0 deg = 360 deg, so there may be a lot of wasted motion in a threading op.

For lathes "tool changers" are usually implemented as gang tooling or turrets. Gang tooling is just having a bunch of toolposts on one cross slide. It's generally incompatible with using the tail stock, so it doesn't work well for long, skinny parts or turning between centers. Turrets are usually vertical, but there are a few small horizontal turrets that are similar to an old style rotating tool post. Usually they're 4 position, whereas vertical turrets can have many more positions.
GRBL might be a bit limited for getting into a lathe of that size. Better not to skimp and go with something like LinuxCNC or MachineKit (or something commercial like Mach or Centroid.)

[hfjbuis]

You got threading working in your lathe with GRBL right? So basically you have the x, z axis and the spindle all working in tandem during threading. How did you manage that with GRBL?

Starting a thread is done by a goto
Position command on Z and C (Y). GRBL then keeps both axis (stepper on Z, stepper on C) in sync.
No hightech involved, any stepper controller can do tis.
The size of your lathe does not matter. The controller only sens step and direction information.
Mach and linux CNC HAVE BUILD IN SUPPORT FOR TOOL CHANGERS. Works as long your tool changer is supported.
There is nothing wrong with starting small, learning and then do it your way and the right way.

[hfjbuis]

I think that's why he has a stepper on his spindle, to allow for threading. To GRBL it's just another step/dir axis. I have no idea if/how good GRBL's rotary axis support is. Without some sort of support for rotary, the controller doesn't know that 0 deg = 360 deg, so there may be a lot of wasted motion in a threading op.

GRBL does not know what it is controlling and does not need to. Waisted motions are generated by the CAM software. I can judge CNCL, it doesn't waist motions.

[skrubol]

GRBL does not know what it is controlling and does not need to. Waisted motions are generated by the CAM software. I can judge CNCL, it doesn't waist motions.

So how do you deal with say multi-pass threads. For simple example, 1" long, 10tpi thread with the end of the bar at Z=0 (forgetting clearance issues, etc.) you'd command a G1 Z-1 C3600, then an X retract and a G0 Z0 to get back to the beginning. For the second pass (after feeding the X,) what would be next, G1 Z-1 C7200? Or do you use incremental, or reset the C0?

[skrubol]

Mach and linux CNC HAVE BUILD IN SUPPORT FOR TOOL CHANGERS. Works as long your tool changer is supported.
There is nothing wrong with starting small, learning and then do it your way and the right way.

I believe both Mach and LinuxCNC use macros for tool changers. You would have to modify the macro to suit how your tool changer is set up. Gang tooling needs nothing special in the control, as it's using the existing axes, it can all be taken care of in the coding of the G-code or CAM post processor. It's certainly easier if the controller is aware of the tool offsets though.

[hfjbuis]

So how do you deal with say multi-pass threads. For simple example, 1" long, 10tpi thread with the end of the bar at Z=0 (forgetting clearance issues, etc.) you'd command a G1 Z-1 C3600, then an X retract and a G0 Z0 to get back to the beginning. For the second pass (after feeding the X,) what would be next, G1 Z-1 C7200? Or do you use incremental, or reset the C0?

@skrubol, Sorry for my late response, didn't see your question until now.

I haven't programmed multi pass threading yet, but i would do the passes one by one, by rotating the c-axis a bit more at the start and then do the same cut at the same depth until all passes at that cutting depth are done. Then increase the cutting depth and start over again. To do this, it would require a change in my software. Until now, nobody has asked for this feature so i didn't make it. But i will put it on my to do list.
I wonder if the threading tool needs a (lot) more clearance because the helix angle will be quite large!

Here the recorded gcodes for the first threading pass
;V01;UNC 3/4 10 TPI first pass
;infeed angle 29.5 ° full rotation infeed (chamfer)
;Ending with 400 ° rotation for easy tool extraction
;With back lash compensation
;Cutting mode constant volume cut
; Spindle = C-axis = controlled by GRBL y-axis 1 mm = 1 °
;
G92X24.150; Goto to start position at max rate
G92Z2.310
G92Y-361.600
G92X24.150
G92Z2.310
M8; Coolant on
G90G0X10.150;
G90G0Y-361.600Z0.810
G90G0Y-212.598Z2.310
G90G0Y-212.598Z1.500
G90G0Y-0.142Z1.084
G90G0X9.390
G90G1Z1.083F60.000;
G90G0X9.390
G90G1X9.390Y-0.142F10800.000
G90G1X9.380Y360.000F10800.000
G90G1Y360.000Z1.083F10800.000
G90G1Y1068.661Z-3.917F10800.000
G90G1Y1068.661Z-3.917F10800.000
G90G1Y1458.661Z-3.917F10800.000
G90G0X9.540
G90G0Y1066.871
G90G0Y1066.871Z-3.107
G90G0Y358.210Z1.893
G90G0Y358.210Z1.893
G90G0Y-1.932Z1.894
G90G0X10.150
G90G0Y-1.932Z1.894
G90G0Y-214.388Z2.310
G90G0X10.150
G90G0Y-214.388Z2.310
G90G0Y-214.388Z2.310
M9; Coolant off

[hfjbuis]

Multi-pass threading will be implemented in the august release of the software.

[ZeroBacklash]

@hfjbuis

I have seen you build at the link you earlier posted.
1. Did you do any backlash elimination for leadscrews of the z-axis (carriage) and x-axis (cross slide). Are there any general guidelines to eliminate backlash via hardware / software?
2. What kind of accuracies do you get for this build?
3. What feed rates do you generally run at for say mild steel?
4. Are geared stepper motors preffered to non-geared ones?

[hfjbuis]

1. Did you do any backlash elimination for leadscrews of the z-axis (carriage) and x-axis (cross slide). Are there any general guidelines to eliminate backlash via hardware / software?

I implemented backlash compensation (elimination) in (my) software (X,Z and C-axis). It could be done in hardware, but that would mean a change in the controller software (GRBL). The GRBL developers decided that this feature could and should be resolved in the CAM or GUI software. I agree to this.
Basically you have to compensate for backlash at every change of direction. This could be done in hardware. But when i chamfer during threading and move from the Z- to the Z+ direction (to the start of the thread) and then would chamfer by moving the threading tool in the X- direction, the carriage would / could move in the Z+ direction by the force of the tool. So before the chamfer X move, i move the carriage 0.001 mm in the Z- direction. There is no way, the hardware could do this so it can only be done in the CAM or GUI software.

2. What kind of accuracies do you get for this build?

Accuracy depends on the play / setup of your machine and of course the wear. I modified the sliders for the carriage that made my mini lathe more rigid with less than 0.005 mm play. You should do this anyway. At the moment I am redesigning these sliders and will replace them with a ball bearing.
After the setup, I tighten the top slide gib screws.

In general i can turn the diameter accurate to 0.04 mm without measuring or adjusting anything independent of witch of the 10 turning tools I select.
When I measure the diameter before the last 0.1 mm to turn, i can turn withing 0.01 mm. If i don't change the tool, i do not need to measure again.
I did a test on how thin i could turn aluminum using a CCGT060204 insert. Even when turning 0.001 mm down each time, I still could see it was turning, not rubbing and it was turning each of the 10 passes i made. Needed a magnifying glass to see this! This was impressing and means that the positioning of the X-axis is repeatably very good.
The accuracy of positioning the Z-axis is also repeatably very good. I can't measure this but when i repeatedly face 0.01 mm off, it is facing each pass. So the positioning of the Z-axis must be within 0.01 mm.
I checked the positioning of the Z-axis using a iGaging DRO. Over a total length of 200 mm, the values from the iGaging DRO differs no more than 0.02 mm from the CNC DRO.
Beware

• These measurements are done with very little stress on the machine so there is very low influence of the rigidity of the lathe.
• The measurements are done, moving the axis in the same direction.

Backlash compensation is a bit complex on my lathe. There is a difference of 0.04 mm between the first and the second pass in the same direction on the Z-axis. I think this is caused by the modified sliders under the carriage. That is why I am going to change them for a ball bearing type. Then I expect more accuracy but less rigidity.

3. What feed rates do you generally run at for say mild steel?

I usually turn small parts and run slow between 30 and 60 mm/min and 100/500 rpm. I could run faster, especially when using insert tools, but it would give more stress on the machine (more wear) and less accurate turning because of the deflection of the part. I only make prototypes and time is never an issue.

4. Are geared stepper motors preferred to non-geared ones?

My X-axis is driven directly by a Nema17 0.5 Nm stepper.
The Z-axis is driven by a Nema17 0.5 Nm stepper, geared down 20/60. I needed more torque and the gearing covered the distance between the stepper and the Z-axis.
The C-axis is driven by an Nema24 4Nm stepper, geared down 20/45. If I would do it again, i would gear it down 20/100 just for threading large diameter (60 mm) Cr42Mo4 steel. I never run this C-axis at it's max speed!

I used the original plastic (POM) lathe gears. I made some of aluminum but they made more noise than the plastic once so i switched back to the plastic once. Until now, i didn't wear out a gear nor broke one. This is one of the advantages of using low torque steppers and not running at high speeds.
Geared steppers a very compact and expensive but also longer. Once you buy them, you "can't" change the gears. So for the first machine (prototype), I would go for external gears.

[ZeroBacklash]

@hfjbuis, Thats a thought provoking extended reply. Many Thanks!

ok so by sliders you mean the long screws that run either of the axis right?
i have posed this question for a bigger lathe machine, I will post pics soon. The bed of this lathe as around 4 feet long, Z axis lead screw could be 6 feet long and x-axis screw may be around a foot long.
I have not done any tests for play/ backlash on this machine. But considering its a used machine I have a feeling that there would be considerable play.

So do you do backlash compensation in your software, CNCL? is this just a number to be fed into the software, and once set will the software do the necessary corrections like retracting the axis a particular amount before any direction change? I would feel that you would need to move the axis a few microns ahead due to the play in the thread interactions instead of retracting them, this way the threads will mate during the initial baclash compensating movement and then would start the actual movement for machining.

I am considering using 7Nm geared stepper for the Z axis and 4Nm for the X axis, right now these 2 axis are all I need, I will add the spindle (C-axis later).

[hfjbuis]

So do you do backlash compensation in your software, CNCL? is this just a number to be fed into the software

Measure the backlash, the error in position when you move 1 mm back and forth using a dial and fill in this value in the setup. The software takes care of the rest.

I would feel that you would need to move the axis a few microns ahead due to the play in the thread interactions

I start threading by a lead in (x-axis) to the cutting depth for a full turn and then do a full turn to chamfer the start. So even when you start in the mid of a workpiece, the lead in will be OK.
When you move back to the start after a threading pass, the Z-axis spindle backlash compensation is pointing in the wrong direction. To get it right, the spindle is moved 0.001 mm beyond the start position, moved back to the start position, the lead in turn is done and the chamfer turn is done. Doing it this way, the spindle will hold the carriage in place during the lead in and chamfer turn.

Turning is done in the direction from Home Position to Start Position. So when you set the Start Position (by jogging or by dials), there is no backlash involved. As soon as you move a axis in the opposite of this direction, backlash is in involved and should be compensated before the actual move. Basically for all lathe operations the backlash compensation value is not critical as long as it is larger or equal to the actual value. Except for turning a Gear Cutter or Timing belt Pulley Cutter, the backlash compensation value for the Z axis should be right.

Example:
After a turning pass, the tool is retracted back to Start Position X and then moved back to Start Position Z. If the backlash compensation X is set to high, the tool is just retracted a bit to far, no harm done. If the backlash compensation X is set to low, and the cut was small (0.01 mm) the tool could scratch the surface during the travel to the start position.

I am considering using 7Nm geared stepper for the Z axis and 4Nm for the X axis, right now these 2 axis are all I need, I will add the spindle (C-axis later)

4 and 7 Nm is quite strong. Your lathe must be able (rigid enough) to handle this kind of force!

ok so by sliders you mean the long screws that run either of the axis right?

Here some pictures of the "carriage press boards" .For all details, pictures and drawings look at: http://www.mwt.messageboard.nl/forum/viewtopic.php?f=6&t=15



The old "sliders" using press and pull screws, not rigid and difficult to setup



The new press boards, very rigid and easy to setup, just tighten very well




The carriage and the new press boards mounted.
The play is achieved by milling down the press board at the sliding side.
To much of play (wear) is reduced by milling down (0.002 mm) or grinding the slider at the side where it rest on the carriage.