1 Attachment(s)
Mach 3 going out of the toolpath, no step loss (i think)
Hello everyone,
Here is the situation,
I have a cnc router from china for years and it was working properly. But as everything can be improved, I upgraded my machine with 2 new ballscrews, closed loop steppers and a new BOB (NVEM2 from novosun) during past month. The computer is till the same, old but reliable, not connected to internet, no antivirus or anything, never had a problem with it.
After all the modification were done i'm now testing the behavior of my machine. I'm very pleased with the closed loop stepper (so quiet), and everything was looking good.
After the tuning of the motor etc, I started my first project. Simply making a hole into a wooden ball.
And weird stuff happened... The cnc kind of loose it's path, randomly. After some time, the cnc kind of drift to a side (random too) and continue the job, but offset in x and/or y but not Z. I don't think it's a step loss from the motor (as they are closed loop, it should not happens) nor a interference in the cable of the encoder, as the path display by mach3 is correct. It draws the drifted path correctly on the toolpath screen. You can see the behavior on the link here :
https://youtu.be/mt-XLlwWptg
In this example, the behavior of the motor is normal, but in some case, the feedrate decrease a lot and sometimes the feed start to be "bumpy", you can hear it a bit in this example. (whith normal moves, the sound of the closed loop stepper is very smooth
second example here:
https://youtu.be/wXuBna64f4o
The gcode file is attached to this post.
Any idea of the problem?
Thanks
Re: Mach 3 going out of the toolpath, no step loss (i think)
After it's done all that, and you tell it to return to XYZ zero, does it go to the same place where you set those values? If so, it's probably a problem with your tool paths. If not, it's either lost steps or a mechanical problem. "Closed loop" steppers can still lose steps; don't rule that out just because you've got them.
Re: Mach 3 going out of the toolpath, no step loss (i think)
Hello.
The test given by awerby can give you a very good idea of your machine status. Since you replaced the ballscrews I suspect your couplings might not be tightend enough. Taking into consideration that usually no flats are machined to the shafts, slippage is very common.
Remember to do the test under load as slippage conceals itself very easily.
Another idea that might help. Using a felt marker draw a line across the shaft of the motor then over the couplings and finally over the transmission shaft.
Run the machine doing actual work to load the axes. Then check the lines. If they appear misaligned you surely have slippage.
If you are still in doubt, removing the couplings and checking the shafts for scratches is another means to confirm slippage.
Let us know what you find.
Regards.
Re: Mach 3 going out of the toolpath, no step loss (i think)
Hello, tanks for your advices.
I don't think it's a slipage or missing step, here is why, the toolpath should be contain within +-4.5 on X and Y for the operation. As you can see on the video, the toolpath shift and the coordinates change with the effective change of position of the spindel position. Mach3 doesn't lose track of the toolpath, and it display the real position of the tool. There is no feedback between the closed loop driver and mach3. So if it's missing step mach3 doesn't know. Here mach3 knows where the tool is but does not follow correctly the gcode. As it's random, I have no clue on what it could be.
I don't know if it is a computer problem, mach3 problem, BOB problem? Could the noise of the VFD or the stepper put the computer in trouble?
Re: Mach 3 going out of the toolpath, no step loss (i think)
Hello.
Rethinking your problem, may I suggest you test your machine without the tool compensation? I think it´s worth a try.
Keep me posted on the result.
Regards.
Re: Mach 3 going out of the toolpath, no step loss (i think)
Quote:
Originally Posted by
bob_dit
Hello everyone,
Here is the situation,
I have a cnc router from china for years and it was working properly. But as everything can be improved, I upgraded my machine with 2 new ballscrews, closed loop steppers and a new BOB (NVEM2 from novosun) during past month. The computer is till the same, old but reliable, not connected to internet, no antivirus or anything, never had a problem with it.
After all the modification were done i'm now testing the behavior of my machine. I'm very pleased with the closed loop stepper (so quiet), and everything was looking good.
After the tuning of the motor etc, I started my first project. Simply making a hole into a wooden ball.
And weird stuff happened... The cnc kind of loose it's path, randomly. After some time, the cnc kind of drift to a side (random too) and continue the job, but offset in x and/or y but not Z. I don't think it's a step loss from the motor (as they are closed loop, it should not happens) nor a interference in the cable of the encoder, as the path display by mach3 is correct. It draws the drifted path correctly on the toolpath screen. You can see the behavior on the link here :
The gcode file is attached to this post.
Thanks
Okay. I've ran the code in my Mach and I had no problem.
But can I make a suggestion?
I woudn't just plunge straight in and then fly round I would ramp to the inner circle depths instead.
There is a chance that you plunge in to your depth and then.
As you go to fly around the circle a stepper stalls from the force causing the position change.
Plunging that deep with a 6-32 (4.76mm) seems a little much to me.
I'd only be stepping down 2.4mm at a time (1/2*tool Dia) if plunging and then a woc of about 0.5mm a time to bore out to size (1/10th*tool Dia).
Or.
Ramp to 1.5*Dia (7.17mm) then woc 0.5mm a time to bore out(1/10th Dia).
Well anyway. Mach showed up as fine so I'd do a few fresh air runs and check your 0 point is still in the exact same place after a few runs.
Then you'll know if it's your mach or your machine or even more simply the way you are trying to cut it.
Routers are never great at deepish cuts.
This program would be iffy on my mill. Even once I've adjusted the feed speeed accordingly to match your chipload and that has 4nm of torque at the spindle.
Re: Mach 3 going out of the toolpath, no step loss (i think)
Hello, Thanks for your advise dazp1976.
I should have precised that the two videos that are in my post have been done in the air, not during the project. So the load does not affect the behavior of the run. I made a few cuts (in wooden ball) without problem with that load.
I m more and more thinking to an EMI interference causing random issue in the computer. Does anyone now a bit about that subject?
Re: Mach 3 going out of the toolpath, no step loss (i think)
One step after another...
If I don't run the spindel, the program doesn't behave weirdly... At least for ten times. On average, it was doing weird stuff every two runs...
Probably a noise problem.
Did someone experience such a problem in the past?
Re: Mach 3 going out of the toolpath, no step loss (i think)
Hello.
I remembered an old, and I mean old system, well over 25 years ago where a similar problem got us crazy. The drives were of the COMPUMOTOR series.
The problem as it showed up whas that steps were actually gained reducing the travel length of any axis and in variable amounts and in different parts of the workpiece.
We usually wire our machines using twisted pairs but in that machine adding line filters, changing to shielded wire, grounding it with a second copperwell bar, fully rewiring it and a number of other things and just nothing worked.
It occurred to me to use some ceramic capacitors connected to each STEP & DIRECTION terminals. That made the trick. For current drives I estimate their value to be 0.001uF/50V. They should be connected between each input pair, that is between the STEP- and STEP+ terminals and between the DIRECTION- and DIRECTION+ terminals.
They are inexpensive and I really think they are worth a try.
If they affect your movements by loosing steps smaller capacitors can be used. I´d try 100nF/50V caps.
I hope this helps.
Regards,
Re: Mach 3 going out of the toolpath, no step loss (i think)
Quote:
Originally Posted by
bob_dit
One step after another...
If I don't run the spindel, the program doesn't behave weirdly... At least for ten times. On average, it was doing weird stuff every two runs...
Probably a noise problem.
Did someone experience such a problem in the past?
That's an important clue. Yes, noise from the spindle can cause step loss (or gain). Often it happens because the wires between the spindle and VFD are unshielded, which allows them to act as a broadcasting antenna, sending erroneous signals to the comparatively low-level logic inputs that run your motors. The usual fix is to slide a braided uninsulated wire shield over that whole wire bundle, and ground it at one end only (most often at the VFD end). It's not recommended to ground both ends, because that creates a loop that induces the noise back into the wire. Here's some additional info about EMI and shielding from the LinuxCNC manual: Best Wiring Practices
Re: Mach 3 going out of the toolpath, no step loss (i think)
Hello everyone,
update on the subject after vacations and orders received. I ordered and installed a noise filter (CW4L2-20A-T ) on the power line off the VFD, changed the cable from the VFD to the spindle for a shielded one (the original that came with the machine was unshielded and never had any issues with it but I changed the BOB that might be more sensitive), and add a bunch of ferrite.
The problem was "definitively" noise. I had issues after 2 runs on average and now i made like 20 runs in the air and got no issue :). So I'm happy. I'll keep you update if the problem come back.
Now I'm really happy with the changes I made on my machine, the closed loop stepper give me entire satisfaction, they are so quiet !
Thanks again for your yelp and advises
Regards
