[lumenire]

I am hoping someone can shed a bit more light on this issue. I have been struggling with this problem for quite some time, and now that many of us have a bit more free time I was hoping to get some additional advice.
So this is what I have and what I have done so far, but I still get false limit switch events. Limit switch trip will even happen when the mill is idle and the router is off.It is a very random trip, sometimes within 15 minutes, sometimes not for 2 hours. I know this is almost certainly some EMI problem but I have not been able to eliminate it.

K2CNC Model-KG3925

TC9 Servo Motors
AMT 103 CUI Encoders
Gecko G320X Drives
Dell desktop running XP and the latest version of Mach 3 (no network/no wireless card)


XYZ servos and limit switches use cat5 cables to connect to the controller.
The feed from the cat5 cable harness to the physical limit switch is a twisted pair.
The power cable to the Porter-Cable router is covered with a shield braid and grounded to the common ground on the controller enclosure.
Encoders power cables 5v (at least that is what CUI info states)
Servos and CNC frame go to the common ground on the controller.
All cables have clamp on ferrite cores on both ends.
Power strip has EMI/RFi suppression
Even set Debouce to 8000.

It is almost impossible to run any job outside of something that only takes a few minutes. Any suggestions as to what else to try or investigate would be greatly appreciated.

Thanks.

[Jim Dawson]

Wire the switches normally closed. That way they are almost noise immune.

[lumenire]

Thanks Jim, The switches are wired (NC) normally closed and the Mach 3 ports are set to Active Low.

[Jim Dawson]

Power supply problem maybe? Maybe add a large filter capacitor to the power supply output?

[lumenire]

Thank you for the suggestion. I am not exactly sure how to do that.

[Jim Dawson]

Are your switches wired in series or is each wired to a seperate input? Where does the LS power come from? It could also be a loose wire somewhere. Without knowing how your system is wired, it is really hard to try to come up with an answer. It really sounds like you have done all of the proper noise limiting steps, so may not be a noise problem.

[lumenire]

Each switch is separate. The ethernet cables that connect to the encoders also carry the power to the limit switches. They go directly into the K2 servo control box (This system came all setup, but K2 went out of business years ago). I would have to open up the servo box and see if I could tell where those leads go. I have changed the ethernet cables but that made no difference. I have wiggled all the connections and I can't get any of the switches to trip. I have also left the diagnostic window on Mach3 open and watched ports to see which it was, the trip will occur but none of the 3 port indicators flash. I know ground loops can be an issue and have looked at that too. In may tracing of the ground connections I know that the parallel port seems to carry a ground to the servo controller, the chassis of the Dell is also grounded through the power cable. I even put a ground lift on the power cable of the Dell, but that made no difference.

Seems that this issue got much worse when the machine was moved to a different location. So maybe it is coming through the AC power?
I can send pictures if that would shed any light on the issue.

I just have my own little shop as a luthier and I know a little bit about how this all works, but since I didn't build the CNC and I can't talk to the manufacturer it is been frustrating. I can't really put an expensive piece of wood on the machine knowing it is not reliable. Thanks again.

[Jim Dawson]

Some pictures might be helpful. It could even be a bad switch, maybe tap on each switch with a screwdriver handle and see if you can get it to fail. Or the problem could be inside the control box. It also be the parallel port connection. There are so many places that could be a failure point that everything is just a guess at this time. You need to narrow it down a bit and try to get a consistent failure in order really find the problem. Intermittent problems are the most difficult to find.

It seems unlikely that the problem is coming from the AC power.




I inadvertently posted the PDF here instead of my post. Cant delete it.
Al

[Al_The_Man]

In the past there have been many posts with the same subject, In my experience it is mainly down to insufficient grounding.
You will often see manuf. advising against grounding the common of the DC power supply to their equipment, I have usually disregarded this where possible and earth grounded the common at the source, and succeeded in eliminating many problems..
A while ago, Siemens published a paper on their site regarding equi-potential bonding of machine and equipment, also covered is the suggested practice of grounding shields at both ends where this type of bonding is used.
I will see if I can post a copy as it is no longer on their new site.
Many of the previous posts where I suggested this with success goes back so far that many have been deleted under legacy posts by the new owners.
Al.

OOP's Sorry Jim Inadvertantly posted the PDF on your post, can't seem to remove it.

[lumenire]

I replace all the limit switches. I have ordered a new parallel cable to try to eliminate that. If it is in the control box that will be a bit harder to figure out. I have not been able to get it to fail by doing something, it just does it on its own while idle. I know it doesn't fail if there is no power to the control box, but I don't think that really means anything. The only other anomaly that I have noticed when I have been chasing this problem is if I wiggle the ethernet cables and power cables to the servos at the control box I can trip the fault indicator on the servo box, but I have to move them a lot.

Can I also ask if you know if it is normal for servo motors to make a twitching sound when idle and if I put my finger on the belt I can feel it move ever so slightly.

I am going to try to run a small job and then I will take some pictures of the connections. I will open the servo box and take some of the inside.

Thanks for all your help. I have been at this on and off for over a year.

Bob

[lumenire]

Thanks Al. The machine ran fine when new so I would assume at that point in time it was grounded/shielded properly. Nothing had changed in terms of equipment when this showed up over a year ago. Since then I have done all the obvious, but no luck. If it is something going on in the servo control box I am afraid I am at a loss on how to trouble shoot that.

Bob

[Jim Dawson]

The twitching is normal, in a perfect world that would not happen, but the world is not perfect. Most likely encoder dither.

Given that your limit switches are wired normally closed and you are set for active low, it is unlikely that the problem is electrical noise, at least at the wiring/switch level. The false trips are an indication the controller thinks that there is an open connection somewhere.

Sorry I can't be of more help.

[lumenire]

Thanks. I will keep looking at possibilities.

[machinehop5]

....Power Supply problem? how old are they? AC ripple maybe getting noisy.
https://en.wikipedia.org/wiki/Ripple_(electrical)
wild guess,
DJ

[lumenire]

Original to the K2 CNC - 2010.

[lumenire]

Just another piece of information. I noticed even with my Servo control box powered off Mach 3 eventually reported that a limit switch was tripped. Could this be trigged by something in the Dell-PC to Mach 3?

[Jim Dawson]

Just another piece of information. I noticed even with my Servo control box powered off Mach 3 eventually reported that a limit switch was tripped. Could this be trigged by something in the Dell-PC to Mach 3?

Possibly. Parallel port issue? I wouldn't rule out anything at this point. What happens if you disconnect the parallel cable? I would expect this to trigger a fault but something to try.

[lumenire]

Very strange. I can pull the parallel cable out both with the Servo controller on and off and it doesn't trip the limit switch error. I going to leave the cable out and just let the computer idle with Mach3 running and see what happens.

[Jim Dawson]

Interesting

[datac]

Very strange. I can pull the parallel cable out both with the Servo controller on and off and it doesn't trip the limit switch error. I going to leave the cable out and just let the computer idle with Mach3 running and see what happens.

I thought Mach3 had a DEBOUNCE setting, no ? Other similar controls have the ability to apply a simple adjustment to the lines being monitored so that vibration or even dirty contacts have to be open or closed a definite amount of time before the control actually recognizes a change took place.

Debounce settings are in milliseconds, so there is no perceptible delay, just something that acknowledges an actual change of state.

I have primarily used Flashcut since 1997, and debounce settings have always existed. Once I played with LinuxCNC, and while not displayed or made blatantly obvious, it too allows debounce settings to be put into the machine config file.

In that episode, I had scoured LinuxCNC forums for such settings, and received much of the comments you have already... how you have something set wrong, have things wired wrong or have shield problems. You might not.

The fact is, practically every type of switch can have issues with confirmed continuity, and this is exactly why most modern OE equipment run dual channel safety circuits, ones that even send current thru contacts to keep them clean.

If Mach really does not have debounce.... move on. Pick something better. Or, if the machine is in a relatively safe environment, change over to normally OPEN limits and your problem probably disappears.