A Bit Of The Logic Behind Grounding And Shielding....
For all practical purposes, ALL electrical conductors have resistance. When current is passed through those conductors, that resistance causes voltage drop. The greater the current, or resistance, the greater the voltage drop. If the current is pure, constant DC, this voltage drop will be constant. But, if we're operating in the real world, this is almost never the case. The voltage drop will add "noise" to the signal, and the shape of that noise will be more or less the same as the shape of the current. So, where, in an ideal world, you night have a 35V DC supply driving a constant 35V into your load (motor or electronics), you will instead have a noisier, lower voltage, signal present at the load. So, what happens if you run a wire from the power supply, to your motor drivers, and from there on to your electronics? The motor drivers will introduce a lot of noise into BOTH the power and ground wiring. This noise will then be passed along to the electronics, possibly leading to the electronics misbehaving, because the noise may "look" like a signal to the electronics. If, however, you run one set of wires from the supply to the motor drivers, and a completely separate set from the supply to the electronics, the electronics will not "see" the noise introduced by the motors and drivers, making the electronics operate more reliably. The whole point here is to provide a dedicated path from the power supply to each "load", so noise introduced in one part of the system does not get into other parts of the system.
Similarly, you want to ideally have one, and only one, path the power, and ground can take. Electricity will always take the path of least resistance. If you provide more than one path, SOME of the current will go down each path, creating the possibility for current to take a path that will introduce noise into places you don't want it. This is part of the reason you generally don't want to ground the shield on both ends of a shielded cable, as it creates the possibility for current to flow through the shield, rather than through the ground conductor, and that current can induce noise, through inductive coupling, into the signals carried on all of the conductors in the cable. Make the current go where you WANT it to go, don't let it find just its own way.
There are two generally "safe" ways to make ground connections. One is to bring all the ground wires to a single point, and tie them all together there. The other is to create a VERY low-resistance ground bus, using terminal strips, heavy ground cables, or a ground plate. The E-box for my big mill is a 24"x24"x12" steel box. In it, I've mounted a 24"x24"x1/4" aluminum plate, to which ALL the electronics are mounted. This box contains the entire PC (all removed from its case, and all the bits re-mounted to the door of the enclosure), the servo power supply (dual 72V@20A supplies), a giant E-Stop contactor to cut the AC to the power transformer, rectiifers and filter caps, all the servo drives, the VFD for the 3HP spindle motor, three breakout boards, relays for coolant, pwoer drawbar, etc, a Modbus controller, four fans, and some other odds and ends. Everything is open, and in rather close physical proximity, but from the very first time I powered it up, it has worked absolutely rock-solid reliably, with no glitches or flakiness whatsoever, despite having the VFD inches away from the PC, which is inches from the servo drives, etc..
The AC line (220V, plus neutral, plus AC ground) comes into one side of the box, through a fuse, then goes directly to a terminal strip. From the terminal strip, I branch off separate wires to power the VFD and servo power supply transformer. The AC ground connection is tied, through a heavy conductor, to the 1/4" plate to which most of the electronics are mounted. There are additional heavy conductors from the terminal strip to the enclosure itself - one to the main enclosure, one to the door, where the PC electronics are mounted. ALL DC ground connections are to the 1/4" plate. The separate ground connections for all the high-current AC devices ensures there can be no AC ground currents flowing through the 1/4" plate. The 1/4" plate itself acts as a very low resistance ground conductor. Since the currents drawn by all the electronics are relatively small, there is no real possibility of ground currents through the plate causing any measurable voltage drops, or noise injection, into any of the electronics.
All I/O cabling is shielded, with the shields connected ONLY at the BOB, or whatever device they connect to. Where appropriate (limits, E-Stop, etc.), I try to use higher voltage I/Os (limit switches, etc), and ALWAYS provide good, stiff pull-up resistors (100-300 ohms) on logic-level inputs, to make it much more difficult for noise to false-trigger an input signal. If you follow a similar, simple layout, you will rarely have problems with noise, ground loops, and other such problems.
Regards,
Ray L.