[mjohnsonsa]

I am using shielded wire between the drivers and the steppers as well as the VFD and the spindle. My question is what do I do with the shielding in the cable? Do I ground it on the sending end, receiving end, both ends or neither end?

Thanks!

Matt

[mcphill]

Ground on 1 end

[Al_The_Man]

If equi-potential bonding is done on the machine to the star point ground, then both ends can be connected to ground.
Al.

[dharmic]

^^ it often won't be though, especially on cheaper machines.

I'd pin all your earths back to one star point. Earth from incoming mains cable goes there, as do any earth points from power supplies, VFD, stepper drivers, logic boards and all the shields. This avoids ground loops and the potential differences they cause.

oops - missed a bit. The far end of the shields? Leave them disconnected.

[interzis]

^ right!

mobile...

[mjohnsonsa]

This is interesting. I was told previously to keep D/C grounds separate from A/C grounds and in fact to keep D/C grounds from different power sources also separated (e.g., PC USB 5v ground separate from toroidial 5v ground). Do you guys feel this previous advice is wrong?

- Matt

^^ it often won't be though, especially on cheaper machines.

I'd pin all your earths back to one star point. Earth from incoming mains cable goes there, as do any earth points from power supplies, VFD, stepper drivers, logic boards and all the shields. This avoids ground loops and the potential differences they cause.

[dharmic]

Matt, I guess the answer is "it depends". Not sure it's so much AC/DC separation as much as partitioning high power and small signal areas. But consider: they must be grounded in common at some point. If it's not in the machine, it's in the powerstrip feeding the computer and power supplies on a small machine, or in the switchboard on a larger machine. Somewhere, they all hook up. I'm a fan of keeping the runs (and therefore impedances, and therefore the voltage differentials for given current draws) to a minimum from that one point where everybody agrees what 'zero' is.

But, sure, if you have a massive but crappy VFD shoving a tonne of noise back down into the earth, attempting to isolate that isn't a bad idea. Just be careful when you do, though: the motor windings will inductively couple to the casing and induce currents (and therefore voltages) there which, if not pinned to a common ground, can cause serious mischief.

Exactly what was happening on my router. Motor etc were 'isolated' as was all the logic board but charge built up enough over several hours of running that I got a significant belt when I touched the case trying to work out why the computer would lose the BOB whenever I turned the spindle down. Sorted out the earthing with a big ol' fat spark as I connected the earth strap and since then it's not missed a beat.

[Al_The_Man]

There are several papers out there published by the likes of Siemens, the PDF used to be on their web site, unfortunately since they revamped their web site it does not show up, but also details the now preferred method of grounding both ends..
It covers grounding and equi-potential bonding, of which the aim is to prevent ground loops and most industrial CNC systems do this, this allows the preferred choice of grounding both ends of shielded cables.
Equi-potential bonding also avoids problems caused by spurious noise.
http://www.cnczone.com/forums/phase-...se-issues.html
Al.