System Kill Switch

[Tom Smith]

I've been thinking about the best function for a kill switch, and exactly how much of the system should be "killed"

Would it be best to shut down everything, spindle, motor and controller, or is it better to just kill the moving parts? I assume that if a job has to be stopped, restarting it might be a challenge. UNLESS, of course, some facility to pause is available/can be added to grbl so that when the switch is hit and movement stops, the commands are paused and can be resumed once the blockage/cat/sleeve/etc is removed from the work area.

What is the conventional wisdom for this kind of thing? Personally, I think it should all stop - spindle, steppers, processing - as whatever triggers the emergency stop is (probably) significant enough that the job needs to be restarted anyhow.

Thoughts?

Tom

[RedOko]

When I was working as a machinist, the "BIG RED BUTTON" killed all movement except the power on the computer.

Now this is different from the feed hold function, where the actual spindle still spins but there is no movement in any of the axis.
IIRC, this can be found in CNCinfusion (pointed to me by edward) but I have not heard anything else since it came out a few months ago.

On a safety note, we have shields/guards around the machine and no jewelry, no long sleeves, no long hair and no gloves around rotating equipment.

[Tom Smith]

Thanks Luis - I am pretty clear on the "no loose ANYTHING around moving parts" rule from working on wood lathes, but as they say, s**t happens

In this case, it would be easier to knock out power to the whole control center, arduino and all, assuming that the power supply is powering everything. The PC, of course, will be on a separate circuit, and would be unaffected. In fact, if the arduino power comes from USB rather than from a regulated source, it need not be shut down at all.

On that note, is there positional feedback from the steppers? ie. if the motors are off, will the arduino/grbl continue to process commands as if they were on, or would it pause waiting for stepper feedback?

Tom

[RedOko]

Its been a while since I worked on my hobby I actually forgot

[edwardrford]

Hi Tom,

These are good questions.

First, there's a thread over on the grbl 'issues' list on github regarding this exact issue: https://github.com/grbl/grbl/issues/51 Although lengthy, It's worth the read from top to bottom.

As for an 'estop': On my rig I use a 6 plug adapter to supply power to the 24vDC power supply and the spindle. In a true e-stop situation, it's my thought that you want to kill the power/stop the machine as fast as possible. Period.

There is no positional feedback from the stepper, if you computer were to remain on after you powered down he steppers then no, it will not stop streaming the gcode to the arduino and your position will essentially be lost.

Also, there is some capacitance in the power supplies, so even after throwing the reset switch on the 6 plug adapter, the steppers will still move some distance before the power is completely gone. I've been meaning to put an actual estop switch between the DC power supply and the shield, which would essentially kill the power to the steppers 'right now', without a delay.

I go with the 'big red button' theory for shutting everything down and think that if all else fails I'd rather scrap a part than risk either injuring myself or breaking the machine or some of it's components.

-Edward

[MLange]

There is no positional feedback from the stepper, if you computer were to remain on after you powered down he steppers then no, it will not stop streaming the gcode to the arduino and your position will essentially be lost.

Unless you added rotary encoders to the steppers, or maybe a 3-axis accelerometer on the spindle? (if for no other purpose than to verify that the axes are moving) I'll have to ponder on that one.

[orcinus]

Processing input from accelerometers and encoders and doing any kind of sensor fusion is a CPU intensive job, as many quadcopter / arduino enthusiasts will atest. grbl does a lot of processing to begin with. I don't believe there's much "headroom" left for something like that (although you could devote a secondary MCU to that job).

[ForestRoyse]

I'm not sure if this is the right place to post, but this conversation inspired me to source a physical e-stop button switch. I found this one on eBay: http://www.ebay.com/itm/170520711348 for $3.89 shipped. It took only 3 days to arrive, and I love it! It's not really industrial quality but close enough for my purposes. It's got two sets of connections, one normally opened, one normally closed, and if you buy multiple switches, the bottom comes off, and additional sets of terminals can be attached to one switch as shown in the listing. For under $4 to my door, I think it's awesome. I took it apart and applied some 3-in-1 oil to some of the internal contact surfaces, and now it operates more smoothly. Love it!

[edwardrford]

Great find! I just went in for 2. I've got another project that could use a big red button!

-Edward