[PaulRowntree]

This morning, 2 minutes after finishing a simple job with my Mach3/G540 system, the machine came back to life and started to rerun the code, apparently on its own. It seemed to be triggered by my switching off the dust vacuum shopvac (at the shopvac), but it started the router, waited for the delay, then moved the tool over to the starting point. It was about to go down when I killed it from the keyboard.

The shopvac was plugged into a separate circuit in the house, and the G540 controls the router power. I don't understand how this is even possible, since (I think) only keyboard controls can start a cycle. Am I wrong?

I would be very appreciative of any insight that you may have. It scared me silly, to the point that I will start closing the gcode files as soon as the cuts are finished.
Cheers!

[DonFrambach]

Is it possible that you combined two cut files into one?

[PaulRowntree]

No, the file is as expected, and it ended normally. In fact I had even changed the bit for the next operation (yikes!) on the next file before it did the resurrection thing.

[Al_The_Man]

This is one reason that I would never rely on software disable for a spindle when changing tool.
Sounds like spurious noise getting in to the PC, try taking your P.P. commons to an external Earth ground point.
Al.

[PaulRowntree]

The grounding idea sounds like a good precaution in any event ... I have a faucet on the other side of the garage. Is there any danger of creating ground loops with the nominal ground of the 120VAC plug?

Still, I don't understand how noise could trigger such a specific software event.

[JerryBurks]

Huh, scary. Maybe another reason why I like the VFD/spindle better than a router. Just does not start that violently. But I am not familiar with Mach to have an opinion why (other that it relies on a Microsoft OS)

[Lee DeRaud]

Still, I don't understand how noise could trigger such a specific software event.

Had to appear to the OS (and/or Mach) as a keystroke or mouse event. From your description, it wouldn't surprise me if the cursor was still over the 'Cycle Start' button after the job ended.

[neilw20]

Do you have an external cycle start button?
Make sure it has LONG debounce time so transients can't emulate a button push.
In General Logic Configuration, top right hand corner, make it at least 1000, ie. 40mS

[PaulRowntree]

Do you have an external cycle start button?
Make sure it has LONG debounce time so transients can't emulate a button push.
In General Logic Configuration, top right hand corner, make it at least 1000, ie. 40mS

I don't even see 'external cycle start' as a possible input, and the only input active is the estop. All of the trigger inputs are set to -1. Is there anywhere else I can look?

There are two plugins that I don't think I should need, yet are checked : one is a video and the other is SWF
The debounce was set to 0, so I will change that immediately. Thanks for the tip!

Lee : that is a good point, and I don't recall now. A spurious mouse click would set it in motion if the mouse was in the wrong spot. Thanks!

[neilw20]

Oh, another thing. ENSURE sticky keys are turned off.
If you hold down the shift key for 5 seconds and anything changes, then sticky keys are on.

[dodger889]

Even if you have two different outlets in the same room you could still to be on the same leg of power. You should have 220v coming into your house. And that splits to two legs of 110v. So with that you could get an spike through the power line. Could have trigger a run. Just like a funny glitch that happens that stops programs from running.

[S3MIH3MI]

I had a similar situation at the beginning. Every time my furnace or air compressor kicked in the spindle would slow down just a bit and the programed position while running would change. I broke a few bits before I decided to run a separate circuit for the control box and the PC, then I put an industrial surge protector on the line. Have not happened again since. It has been over about a year.

[ger21]

You said you had changed the tool. Did you run an auto zero or anything else?

[PaulRowntree]

Tool was changed, but no zeroing. I do it manually with a PCB plate, but this had not been done yet. The second toolpath file had not yet been loaded.

I am leaning towards the mouse picking up a false left-button click while the pointer was still over the Start Cycle control.

Would there be any risk of losing position if the Mach3 front panel Reset was activated during tool changes? I am 1/2 way to finishing the Hall proximity home switches, so I will soon be able to power down much of the machine when my hands are on it. The router is the obvious issue, but this chain drive is scary fast and I don't want to be too close if it wanted to launch towards the wall.

Cheers!

[ger21]

Would there be any risk of losing position if the Mach3 front panel Reset was activated during tool changes?

Yes, there is that possibility.

If the cursor was over the start button, I'd agree that that may be the most likely scenario.

[neilw20]

I've had a problem, that caught me out till I figured what it was.
I have a touch screen, and fine splashes of coolant were giving random hits on the screen.
I now have a clear perspex cover i put over the screen when the drips are flying.

edit:
RESET will probably lose position if you are using a charge pump to kill power to the steppers.
It is the microstepping in the driver that will lose sync.
I have never really checked for spurious output when in RESET state, and this will depend on whether active high/low state of stepper clock drive pulses occur.
I just never use RESET or ESTOP.
ESTOP is exactly that, and loss of position is very likely.

[RomanLini]

I'm not sure it was a fault with the mouse. The micro inside the mouse usually does a good job of debouncing button input so it should not detect a button press on local electrical noise. And the comms from mouse->PC are USB coded so that is unlikely to give a mouse button event on noise.

I think it was likely the PC at fault, a power spike through the mains may have made it through the PC's PSU and caused the CPU to glitch, often that will cause the PC to hang, or to resume some previous state or task.

That symptom can also mean the PSU or motherboard capacitors in your PC are starting to fail.

If you don't already have one, I would get a plug-in powerboard with surge protect/filter built in and use just the PC plugged into that so it gets cleaner power.

To be extra safe you could add another mains filter for the shopvac alone, just be sure it is rated for a high powered machine. And it's always a good idea to run the PC from a different circuit or power point to high powered motors like shopvacs.

[PaulRowntree]

I'm not sure it was a fault with the mouse. The micro inside the mouse usually does a good job of debouncing button input so it should not detect a button press on local electrical noise. And the comms from mouse->PC are USB coded so that is unlikely to give a mouse button event on noise.

I think it was likely the PC at fault, a power spike through the mains may have made it through the PC's PSU and caused the CPU to glitch, often that will cause the PC to hang, or to resume some previous state or task.

That symptom can also mean the PSU or motherboard capacitors in your PC are starting to fail.

If you don't already have one, I would get a plug-in powerboard with surge protect/filter built in and use just the PC plugged into that so it gets cleaner power.

To be extra safe you could add another mains filter for the shopvac alone, just be sure it is rated for a high powered machine. And it's always a good idea to run the PC from a different circuit or power point to high powered motors like shopvacs.

These are all good suggestions, and I will take a look for 'bulging caps' on the mother board.

Thanks!