[blueboy13523]

Hello, I am new to the site and to the world of small cnc machines. My boss bought a Chinese 6040 cnc with a parallel control box. HE wants to use the machine to make plastic name tags. I am using Mach3 and ArtCam. I can zero the machine and it randomly misses steps on the Z axis. I have swapped the driver boards from the Y axis and still no luck. The mechanical side of things looks solid. No binding or slipping on the screw shaft. I have also reduced the acceleration and travel speed on the Z axis. This only delayed the bit from digging through the material. I have measured the difference in height when I send the head to home and some times I am 1/8" off and some times 3/8" off. Any help would be greatly appreciated.

[fretman_2]

Well...about the only think I have to offer is make sure you are using right units...metric to metric, imperial to imperial.

[blueboy13523]

Mach 3 is setup to metric as well as ArtCam. I am exporting the G Code as G-Code mm *.tap if that helps. I am thinking that the motor for the z axis might be too small. It is about half the length of the x and y motors.

[komatias]

Its the parallel port that is probably the issue. Get yourself a motion controller like the Ethernet Smoothstepper or similar. it is the only way

[blueboy13523]

Thanks. I think I have the boss persuaded to get a Geko 540. He wanted to up the voltage on the steppers, but the drivers in the control box are not marked with anything I can use to find a spec on. Does anyone here run their machines on anything other than 24vdc?

[komatias]

The inbuilt power supply is 24V I think. The machine runs ok at that voltage but could do with more if running nema23s. The issue is the pulse train first, dirvers and voltage second. You cannot use the parallel port in this day and age and expect the machine to function as you want unless you are using the Linuxcnc

[coherent]

I'd triple check all the Z mechanical again. Not just the coupler but the bearing mounts for the screw and the ball nut mounting etc. 70% of the time mechanical looseness or binding (if things aren't completely square and true) issues are the cause of lost steps. From there I'd slow all my motor speed and acceleration settings much slower and make sure you're not exceeding the speed the motor is capable of raising and lowering your spindle without loosing steps. Last I'd look for cable interference. Power or other stepper cables emit interference that can cause such issues especially with lower quality cables.

[blueboy13523]

I got the Gecko 540 installed and there is a night and day difference in the smoothness of the stepper motors. I was able to cut out a few parts on a 1" MDF sacrificial board. I changed to a 1/2" board and noticed that at random times the z axis will skip when raising from the lower height. Thinking I might be pushing the z motor to its limits. I slowed the motor to 800 mm/min and the acceleration to 200mm/sec. The shaft coupler is snug, and it doesn't skip if it starts from the 1" board.

[blueboy13523]

Here is a picture of the last part I ran. After speaking with one of our engineers, it looks like a step is being added after every letter. The G-Code is consistent through. Our engineer was suspecting it might be a setting in Mach 3. Would there be anything I might have missed in my initial setup?

[coherent]

If everything is tight, my guess would be an axis is binding or missing steps. It's kinda hard to tell if it's binding or tight unless you remove the motor or coupler so it can move freely. In the photo it looks like the Z to me but hard to tell. The Z often looses steps when lifting which in turn adds to the depth of the cut. It's common with Z's because of the extra weight from the spindle. Most CNC routers and mills use a larger stepper/servo and/or slower speeds on the Z for that reason. (don't assume that the settings on the Z should be the same as the X & Y simply because it uses the same size screw or motor). Test each axis it's complete travel for any tight spots or binding. It may seem like you are adding a step, but more likely you're loosing a step or steps not adding. Your steps per and most of the other settings in Mach 3 would create a repeatable consistent error. I think you're loosing steps or have motor interference.
Simply slow things down... really slow cut, jog speeds and acceleration and do some test cuts.

Yes the G540 is huge improvement over the import drivers. If you haven't already, get rid of the parallel driver & cable and get a UC100 and connect the G540 using a USB port and eliminate the parallel port driver issues that Mach has. It will run even smoother. It's a cheap option. You won't regret it.

[CNCMAN172]

This actually sounds like you might have noise impacting the stepping motion on Z axis. You have eliminated the fact it is not the driver you swapped cables, keep in mind the Z axis is very close to the spindle that is cutting the material. The noise from the spindle can interfere with the pulses being sent to the motor. You might want to check that the cables going to the z axis motor is shielded. Since you have already reduced acceleration and velocity this leads me to believe you have interference at play. You can run a test running the same gcode and leave the spindle off and measure the Z axis, you say it does not return to the same height. Well if that is a homing routine normally the spindle motor is not running but this depends on the machine, some machines do not control the motor on/off. Anyway something to look into for you.

Russ

[blueboy13523]

Finally found something. The bottom bearing for the Z axis is loose in the plate. When lifting up on the spindle it would rise up about 1/16". Put some green locktite on it and will see if this helps. Thank you very much for your help. I will definitely look into the UC100.

[blueboy13523]

It did help a little. now there are no drastic changes in depth. Still getting a repeatable increase in depth over the run. I can slow the motor setting and the problem slows down, but is still there.

[A_Camera]

The inbuilt power supply is 24V I think. The machine runs ok at that voltage but could do with more if running nema23s. The issue is the pulse train first, dirvers and voltage second. You cannot use the parallel port in this day and age and expect the machine to function as you want unless you are using the Linuxcnc

You can use a PP just as well today as you could 10 years ago, and in fact even better if you manage to install XP or W7 32bits on a new and fast PC. The problem comes first when you try using W7 64bits OS or newer... forget about that.

OK, using parallel port is a bit of a dinosaur and using USB or Ethernet is much better, but with the right OS it isn't really a huge problem.

[A_Camera]

Here is a picture of the last part I ran. After speaking with one of our engineers, it looks like a step is being added after every letter. The G-Code is consistent through. Our engineer was suspecting it might be a setting in Mach 3. Would there be anything I might have missed in my initial setup?

If extra steps are added consistently after every letter then it is a configuration, or G code problem. Extra steps due to mechanical or electrical issues are random.

Read the Mach3 manual and check every parameter and the configuration of Mach3. There are plenty parameters, but Mach3 has no clue about what you are milling/engraving, it is just doing what it's been told to do, so it can't really add extra steps after each letter, since it has no idea about where a letter starts and ends or that it is a letter it is working on. What do you mean by "The G-Code is consistent through"? If there is an error in the G code it may still look consistent. Is the table flat and runs parallel with the tool? That picture shows a very uneven work, which might come from table (or the plastic sheet) not being flat. Work piece holding is also very important. How is it fixed to the table?

I don't think your problems are caused by electrical noise, more like mechanical, configuration, setup or G code problems.

[CNCMAN172]

Take your G-code file and run it across a CNC emulator program that plots the Gcode, that will tell you if the Gcode is actually correct. If those plots look perfect then you are back to a machine issue.

Russ