Tips for intermittent faults on a CNC?

[alanambrose]

Hi,

I have some hardware debugging to do on a 1/2 sheet-style CNC router. It has two intermittent faults on it:

1- sometimes it just stops halfway through a job and
2- sometimes it produces results that are not suggest it's losing steps (for instance, not making a square accurately square)

Any thoughts? Recognise these problems already?

I was going to check the power (~230V UK) to see if it has any dodginess on it. The scope I have around here is a TDS2024, which in theory has 300 V RMS CAT II input. Should I feel OK about that, or should I make some sort of attenuation network to put it under less stress?

Alan

[CatalinaWOW]

I have very little experience troubleshooting CNC, but the war stories of friends and relatives who do work on them always seem to revolve around the position sensing systems.  Dirt or wear reducing the SNR in the sensing system until their is a significant error rate.  That is where I would start looking.  Those sensing systems virtually all work on some reduced power from the lines, things like 24 V DC, so no problem looking with your scope. 

[Po6ept]

I have very little experience troubleshooting CNC, but the war stories of friends and relatives who do work on them always seem to revolve around the position sensing systems.  Dirt or wear reducing the SNR in the sensing system until their is a significant error rate.  That is where I would start looking.  Those sensing systems virtually all work on some reduced power from the lines, things like 24 V DC, so no problem looking with your scope.

Also related is wear and binding in the lead screws, table, anti-backlash nuts, etc.  A mechanical problem can look to the system like a logic or sensor problem.

[Fortran]

I'm not a hardcore CNC machinist, but I use one fairly regularly.
You don't mention what software you are using or your computer setup, which might be useful.

I have encountered similar problems and my two cents are:

1a. Computer problems.  A CNC is usually driven by the LTP port, and don't really like to share resources with other stuff. Turn off everything you don't need. USB to LTP is begging for problems.
1b. CAM problems.  Is your Gcode ok?  Are there more lines to be processed when it stops or does it simply "run out of" commands because of whatever reason?
Unlicensed CAMbam only gives you the first 500 lines of Gcode once the trial runs out.

2. Check your acceleration. Most of my problems lied there.   If you think it looks good, turn it down a bit anyways.

[0xfede]

I alanambrose,

I had the same issues on a CNC with stepping motors and mach3.
It turned out that the cause was ground loop between the pc and the driver board through both the parport and mains.
The solution was to put an optocupler board on the parport and everything runs smooth since then.

I'll discourage you from probing mains with a scope for various reason. It could be dangerous for you and your instruments. You will find plenty of informations about that scattered around the various thread of this forum.

Best,
0xfede

[smgvbest]

When it stops
Is it producing a estop or other message?  If so what message?

Ground loop is a common problem
Also faulty limit switches or configuration can be a problem
Are you using limit and/or home switches?

Loosing steps could be a cause of out of square but also miss alignment of the machine not just loosing steps.   Is the out of square consistent itself.  I.e.  If you run same gcode in the same offset location (your g54 offset) is it repeatable?

If you cut in a different offset (say at x10y10) does it get same result

I'm not sure I saw if you posted what software your using. mach3/4 , linuxcnc or other?  I'll look once I finished posting this but didn't think you did.

Sandra

[cncjerry]

try moving the plug for the power supply to the stepper drivers to a separate outlet from the router motor.  Generally noise or loading is the prime fault for missing or extra steps.  The other thing to try is turn on the spindle and see if the steppers are moving by themselves. You might have to enable the motors in software to do this test.  If the steppers move when the spindle is running then noise is feeding back into the drivers.  Excessive noise can halt the computer or trigger estop.

[alanambrose]

Thanks guys for all the suggestions and knowledge. Let me try some of those and report back.

Many thanks, Alan

[Kjelt]

Does your CNC have closed loop steppers? If not then also use material without much mechanical resistance for all the testing, so at least you also know it is not the resistance of the material or too large cutting depth steps which causes the loss of some steps.

[Emo]

Do not forget to check the environment. Is someone around the corner in the shop using an electric welding machine with RF assisted start function? It will definatly impact position sensors and may even halt/crash the stepper controller(my experience, took a long time to find out)