In this mill, there will be many devices with "signal ground", such as encoders, ppmc control box, servo drives etc. I think that I should I tie all of those signal grounds, and the minus of the isolated DC from the servo power supply, to neutral (in a star configuration).
Is that a good way to go, or am I way off base with my thinking.
I think that the servo drives may have separate power and signal grounds. I would tend to tie all the derived grounds (isolated from the incoming power but referencing power) to the safety ground, and keep a separate signal ground bus.
Neutral can go a bit too high over safety ground under certain conditions of loading.
Your servo amplifiers (drives) are all being powered from a single power supply -- right? If you had a separate one for each amplifier, I would alternately connect them to one half or or the other of the incoming power (presuming 120 VAC input to the power supplies) so the load is balanced a bit better.
The old Bridgeport BOSS-3 took a *lot* of power for the steppers, and drove each axis from a different phase of the incoming three phase to balance that load.
Here, you have:
The VFD (across the full 240V)
The servo amplifiers (drives) which may be 120 VAC loads or 240 VAC loads. If 240 VAC -- no problem, otherwise split between 120 VAC halves of the incoming 240 VAC.
The computer (balancing one of the servo amps if split power supplies).
Lighting. (120 VAC)
Coolant pump (120 VAC?)
Tool changer? (perhaps another servo axis?) I don't have a tool changer -- sigh!.
that would certainly NOT be the way I would do it - signal ground should go to its own ground with no current carrying motors, solenoids, etc. Power ground is some thing else - they could be tied together, but don't put them all on the same ground stud
You'll have a boatload of signal ground wires by the time you're done. I always take a section of terminal strip and bond several together with little loops of white wire. Then one green wire to the common ground bond point on the machine. Do NOT put AC nuetral to this bond point. Also don't daisy chain your signal grounds, each device has a wire back to this strip. I always use white wire here and then don't bother with wire numbers.
Iggy, The purpose of all those signal grounds is to assure that the signal reference is quiet, eliminating voltage reflections from other device return currents, which can be seen as noise . Bonding them all together defeats that purpose. Obviously at some point, all those returns will tie to DC return (never AC neutral). Those tying connections must be very low impedance, typically
6/8 gauge wire. To simplify the statement, all the signal returns within an electronic chassis could bond to an insulated from chassis bus bar. That bus bas bar is then tied to power supply DC return with a heavy gauge wire as short as possible.Repeat that for every electronic chassis. Never, never, ever connect neutral to safety earth in ANY machine. This connection can only be made at the equalization bar at the power entry point for the building. DC return can be connected to the green safety earth wire at the machine power entry, but often is connected through a 10K ohm resister. Under no circumstances should DC return ever be connected to chassis. These simple rules eliminate noisy ground loops. Steve
When using an oscilloscope to trouble shoot there are lots of occasions where there are conflicts between signal ground and power ground connections. For safety's sake scope metal work must be connected to power ground, but for examination of low level signal circuits scope ground must be to signal ground.
Temporarily breaking the power ground and floating the scope metalwork is expedient but dangerous. A much safer solution is to power ground the scope through back to back diodes.
My scope has a 30Amp bridge rectifier mounted on its three pin power plug. The + & - rectifier pins are connected to each other. The AC pins are series connected between scope metalwork ground lead and the plug power ground.
For either polarity of current there are now two diodes in series beween scope metalwork and power ground. For up to a few hundred mV difference, the scope is now essentialy floated but it is still safely anchored to no more than a Volt or two from power ground.
In the 80's I worked on semiconductor production test equipment. In those days the computers that controlled them drew very substantial power, ours had about 100A of 5V from switching supplies. The measurement sensitivity was microvolts and picoamps; without all the shielding the machine could detect a person several meters away by the increased noise floor.
There were separate ground blocks for all digital supplies and all analog supplies, and we had to be careful to return as much of the digital supply current as possible to the digital return, likewise for the analog.
The measurement ground, HQC or High Quality Common, was partly isolated as Jim Pentagrid mentioned, with two parallel diodes and a
470 Ohm resistor connecting it to analog common. We had to account for any unbalanced current in or out of HQC. IIRC it was considerably less than 5mA.
It turned out that the biggest problems were the control signals from digital to analog or the measurement section. The current for heavy loads like the 2102 memory flowed in well defined loops whose radiating area could be minimized by twisting the power leads. Controls passed between sections and had to be handled individually, to trade speed for impedance. If necessary we used optical isolators for digital or reed relays for analog controls.
Be thankful you aren't designing the grounding for a digital radio that combines millivolt analog-to-digital converter sensitivity with high powered ECL logic. I've seen 3 Volt spikes between two points on a ground plane that someone else had designed to the DC-appropriate star topology. The data bus crossed the plane break in that area but its return path was about three inches away through the star's center.
As a general rule I was told to treat under 10KHz as single-point, over a Megahertz (100KHz?) as a common plane, in between takes more careful analysis, mostly knowing where the currents flow out and back.
In theoretically-oriented night school classes they taught voltage, the very practical Army Signal Corps taught us to pay attention to current flow.
First, as many others have said, do NOT use the power neutral (the white wire) for anything other than power neutral.
I often deal with this exact signal grounding issue in large electronic systems, where "large" means that the equipment is spread out in a large building. Another example is a large computer room with raised floor, such as those for Air Traffic Control systems, where there is always a substantial ground grid under the floor.
The standard solution is to bond the equipment racks and cases together with a grid made of heavy copper wire. This grid is called the "signal reference ground system" or the like, and is an extension of the ordinary safety (green or bare wire) ground system. Except during faults, neither green nor signal reference grounds carry significant current. During faults (like a power cross to an equipment case), thousands of amps may flow, at least until the power breaker opens. So, the reference ground must be built as heavy as the safety ground system, because it may serve as a safety ground as well.
One other thing to keep in mind. The National Electrical Code (NEC) requirement on safety ground impedance is that the resistance can be as high as five ohms, and no mention is made of a minimum test voltage. This makes perfect sense, as the NEC's intent is to ensure that the breaker will pop if there is a fault.
However, five ohms and oxide films requiring 50-100 volts to puncture, while acceptable to the NEC, make such a ground useless as a signal reference. So, the reference ground system must be made of copper and/or its alloys, with mechanical connection hardware optionally plated with silver and/or tin, and must not be made of aluminum or steel alloys of any kind (unless all connections are made by fusion welding, soft soldering, or brazing).
Another way to look at this is to observe that the signal reference ground system more than meets the NEC requirements for a safety ground, and so can serve both safety and signal-reference roles.
For far too much information, see TM 5-690 (Grounding and Bonding in ... Facilities), downloadable from .
War Story: In the late 1980s or early 1990s, I was working on the computers of an Air Traffic Control system in Canada. (We were *not* working on the system running real-world ATC, so no danger to air safety.)
The computers were VMEbus cards carrying 68000-series processors running at ~10 or ~20 MHz. I had a computer board out on a board extender, chasing some now long-forgotten problem.
When I went to clip the oscilloscope ground to the VMEbus ground, there was a tiny spark, and the ATC system crashed. Followed by my heart -- What did I do? What could I have done?? ..... Umm.... Spark? Between grounds?? This cannot be. Again touched the scope ground clip to the VME ground pin, watching closely. Spark! What the hell??? Cannot be...Try it again. Spark!!.... Repeat until eyes are believed.
Turned out that there was a few volts between the two grounds, at 180 Hz, with a source impedance of perhaps an ohm. Yes, 180 Hz. That turned out to be the key
- it was the charging pulses from all those DC power supplies fed from the three legs of three phase power lines adding up.
The cause turned out to be that on a single cabinet (out of many tens of cabinets), the white and green ground leads were reversed, injecting sufficient noise into the reference ground system to cause many kinds of flaky and hard to diagnose behaviour. Probably including the now forgotten problem that prompted me to be poking around with a scope in the first case.
Take a bar of copper -- say 3/8" square, drill and tap a bunch of holes along its length (probably best with a thread-forming tap), put in screws, and use these to attach ring or fork terminals to a common signal ground -- while isolating this ground bus from the chassis.
Or -- just go to the electrical department of your local Home Depot or whatever and pick up one of the ground busses there (intended to go in power distribution boxes) -- holes drilled through a bar, and screws at right angle to clamp them. Not as nice for the stranded wire as proper crimp ring or fork terminals would be, but not too bad either, and less work.
[ ... ]
Another alternative, if you have two (or more) inputs, with an A-B switch position is to run two identical probes to the two inputs, and let the scope display the difference between those two probes. One connects to the logic ground, and the other to the signal being checked. No need to float or semi-float the scope then.
No. The PPMC encoder board provides power to the encoders. For best noise immunity, ALL encoder wires should be connected to the PPMC encoder board, only. The only exception would be a cable shield, which might best be connected to the enclosure the PPMC boards go in.
The DAC board has individual grounds for the velocity command out. If your servo amps have instrumentation amps for the command input, then you want just a shielded twisted pair from the DAC board to the servo amp input. Again, connecting the ground of the DAC board to other grounds may introduce additional noise into the signal. These are 16 bit DACS, so the least significant bit is 20 V/65536 = 0.3 mV, so a little noise really can matter. If your servo amps DON'T have fully differential (instrumentation amp) inputs, then it matters less, as they already have some noise contamination from the drive system, but still, better to leave out extraneous grounds. The PPMC should be grounded at one point to the common ground. For ESD protection, the PPMC backplane is already grounded to the DB25 shell. You can tap off the single-point ground from one of the encoder ground pins, or the motherboard power connector.
This is not critical, but once you have PWM drives switching tens of amps at 80 V, with rise times about 100 ns, the noise issues can become significant.
One could not be a successful Leftwinger without realizing that, in contrast to the popular conception supported by newspapers and mothers of Leftwingers, a goodly number of Leftwingers are not only narrow-minded and dull, but also just stupid. Gunner Asch
I don't have an opinion on this but grounds can give you grief. More learned people than me have commented. Years ago, I built a 2 axis motion system to carry a 150# SRIM injection head between two presses. I believe I was using an IMC4xxx series motion control. The connectors at the encoders did not have the shields grounded and I
*assumed* my assistant tied the cabinet side grounds to the grounding bar. I drew the prints that way. Well after all sorts of hair pulling, scoping, and lost of sleep when the controller kept faulting out, I found he just stuffed the shields into the wireway and never tied them together and bonded them to the ground bar.
Worrying about grounds is a sign you are really doing your research on this. I'm impressed.
-- "Additionally as a security officer, I carry a gun to protect government officials but my life isn't worth protecting at home in their eyes." Dick Anthony Heller