Limit switch grounding

[Richster]

This should be quite a simple one - I haven't got my head around it yet. When wiring a limit switch, am I looking to use single conductor shielded cable, so that my two connections to the microswitch are effectively the shield and the conductor, the shield being connected to 0v? Or am I to connect the shield to earth (from the mains input) and use twisted pair cable so the two conductors inside the shield are connected to the switch? In this situation I think the shield remains disconnected at the switch end, and therefore the extra bits of unshielded wire are kept as short as possible.

Same applies to the stepper power - I'm looking to use starquad 4-conductor shielded cable. In this case, does the shield again go to earth? Again at the motor end, there is no connection.

Machine is wooden, so nothing to connect to. Though is it a good idea to earth all the rails too? Wouldn't be too hard...

[chazmtu]

Hi Richster,

Ideally the shield should be earth grounded and the switch ground is separate. You don't want to be shorting power to earth ground so you were right to think twisted pair. As per grounding rails, I ran a wooden machine for some time now with ungrounded rails and never had any troubles.

Chuck

[Richster]

OK - thanks! So I'll make sure all my connections to both switches and stepper motors are via shielded cables, with my shields connected to mains earth at the the driver board end of the links only. Each switch will be via twisted pair shielded (there will be five of these), and stepper motors will probably be via Van Damme 'Starquad' audio cable which seems ideal for my purposes.

Let me know if anyone disagrees!

Cheers - Rich

[chazmtu]

Sounds like you got it down. Good luck

[Richster]

I completed all the wiring today - it hasn't gone well

Here's the electrics compartment. The IC on the LHS is an opto-isolator, which I am planning to use to drive the spindle relay. Other than that the only components on the board are the pullup resistors, everything else is just connections.


Here's a quick rundown...

Originally, I had just the x-axis home/low limit switch connected, and it ran fine. I had no debounce setting in Mach3, and I didn't need to connect the shield to earth on the cable. All switches are connected via 10K pullup resistors, and this switch was by far the shortest cable length since it is physically close to the electronics.

Without any shields connected, and having connected the other switches, I need to set the debounce to about 2000 to make the system usable, but even then I still get random switches being triggered.

What I don't follow is that connecting the shields to mains GND introduces a massive amount of noise, and makes it impossible to use the machine even with a high debounce setting. As I mentioned earlier in this post, every cable in the machine is shielded, including the stepper cables, with the shield being connected to earth at the control board end, and left unconnected at the relevant switch/stepper. I have a fairly complex setup with the cables whereby I have a 4-core cable which is used to connect three of the switches to the board i.e. at the end of the 4-core cable it branches into three 2-core cables, the three switches sharing common 0v (pc). However, I have attempted to keep as much shielding around these connections as possible.

Here's a closeup of the connection board. All green wires are to earth. The single green wire on the RHS of the photo goes to mains earth.


Stepper driver cables are shielded, the shield is terminated at the stepped motor where the connection is made.


All limit switches are like this - the shielding is terminated within the heatshrink.



So currently I'm getting better (though dodgy) performance with my grounds disconnected! Have I misunderstood something somewhere?!

Also is the value of the pullup resistor critical? The mach3 manual shows 470ohm, I have seen other sites showing the higher 10Kohm value, which is what I've used.

[mcyr]

You may have gound loops. Try disconnecting the sheilding from ground. I don't think sheilding is necessary at all and causes more problems than it solves. Instead, put a 0.1uF capacitor across your limit switch connections, as close to BoB as possible. Reset your debounce to 0 and see what happens. If it's better but you still get occasional glitches, try a slightly bigger capacitor.

Good luck,
Mike

[mcyr]

Oh, and try to physically separate the stepper motor cables from the limit switch lines as far apart as possible, if you can. A little experimentation here will tell you a lot.

Mike

[Richster]

Thanks for the info - I've just added some photos too which might help show what's going on. A few further thoughts:

- I don't have a BOB, but instead have connected to direct to parallel. What are the implications of this?

- With the unit turned off, just connecting the mains earth connection is enough to introduce havoc! I suspect I have some loops, I need to read up on exactly what that is. Of course I can test the switches with just the PC turned on.

- I have physically put all my cables very close to one another, I was hoping I might get away with this my shielding everything? I would be loathed to have to change this, since it's all rather neat at the moment. Would rather find a way of using the existing connections.

[mcyr]

Are you still seeing a bunch of flickering on the limit switch LED on the diagnostic screen? If so, try the 0.1uF cap across the two pins for your input/grd.

[Richster]

I've just tried a few more tests.

Firstly, I've never seen anything visually on the diagnostic screen, on the pin status LEDs. Maybe it's too quick?!

Something else odd I noticed...

With the machine powered off, and the earth disconnected, all switches are detected correctly.

However - when I connect the earth, then I loose the use of the x-axis limit switch. The reason that's odd is this is chained in series with the y-axis limit switch, on the same pin, and the Y one continues to function normally!

Connecting the 24V to the driver board with the earth connected is then enough to trip the unit, regardless of the debounce setting

[Richster]

No it's not odd- it was shorting to earth at one of the switches! I think I'm getting there now....

Right, having sorted that one, all switches are now working, but it still works better with the grounding disconnected rather than connected. I need to have my debounce setting around the 2000 mark as well but am still getting the occasional trip - I wouldn't want to go much high than that since the time lag would then start to add up. It seems odd though that it would work consistently on zero before with just one relatively short limit switch connection.

The capacitor idea sounds promising - I don't have access to any right now, but that could certainly be one to try? I guess the idea behind that is it flattens out the spikes which are causing the triggers?

[mcyr]

Exactly, it's a filter to dampen out the voltage spikes which are triggering the software limit switches.