Grounding shield of 4-20mA signal adds noise?

[nwboson]

VFDs are notorious for injecting noise since they use 'chopper' transistors to switch an inductive load (the motor), you have significant L* di/dt which means potential for lots of voltage spike. If you temporarily disconnect the VFD grounds, does the problem go away?

I normally build my panels with 3 separate GNDs: analog ground, digital ground and VFD ground. They are completely separate buses, except each one has one and only one wire connecting the respective ground bus to earth. This prevents any current flow between grounds but places each ground at earth potential.

 

[kolyur]

I just got back from a three week trip and had asked our electrician to install a ground rod in the floor next to the panel while I was away. It made a world of difference with the noise issues. The 4-20mA shield is tied to ground now and the fluctuations are minimal.

However, one issue that remains is that the VFDs seem to add a slight offset to the analog signal. I hesitate to call it noise because it's very stable and predictable. The magnitude is proportional to the number of drives running, topping off at about 200 counts with all motors on (~0.125 mA). I could compensate for it in my program but I'd really like to know what's causing it. Has anyone experienced this?

 

[OkiePC]

I have seen this with analog signals on a PLC-5. The cause was noise. It was affecting the average. I was seeing upwards of 50v spikes, tiny ones on the scopemeter, coordinated with one VFD running. We had not terminated it's shield at the VFD end, it was ready to be landed. Getting rid of the noise got rid of the 0.1 or so volt offset I was seeing and allowed the machine to cycle properly.

 

[jdbrandt]

Just use ungrounded, and unshielded twisted wire, 3-twists per inch, and all your noise problems go away...guaranteed (*).
I worked for a company who actually studied this, and made it their corporate standard, back-in-the-day.

 

[seppoalanen]

Shielding is to reduce noise issues, if shielding makes it worse, why would you use shield?

On another note. We run grounding on all piping etc. Zero noise problems.

Analog (4-20mA) cables don't need shields as You and "jdbrand" #19 told. Wires are twisted and it is enough.

Your "anather note" is good way kill noise sourses as compensation currents.

Another anotehr note is correctly motor cabling from VFD.
Use Symmetric cable. Connect shield (and/or 3 symmetric PE-wires) only in motor and VFD's "PE", not anywhere else metal construction inside the cubicles etc. Symmetric cable "sucks" harmonics to VFD's DC-circuit.

 

[GeoffC]

I agree with Seppoalanen the VFD cable should be grounded at the motor end and at the VFD itself, nowhere else in the panel. You stated earlier that the cable was grounded in some terminals in the panel - this is the first thing I would be changing

 

[Valley Ind]

I've run into this sort of problem too.

The grounds coming back from motors controlled through a VFD were tied to the same self-grounding ground terminals as everything else.

The fix was to run each motor ground directly back to its corresponding VFD and then each VFD was directly grounded to the one main cabinet ground.

In the case were a VFD powers more than one motor, run individual motor grounds back to isolated ground terminals which are jumpered together, then run a single wire from that set of terminals back to the VFD and then a single ground wire from the VFD to the main cabinet ground.

 

[TurpoUrpo]

Yeah. I did not even think about anyone could do it any other way . We run all motor cabling (that has VFD) from VFD to Motor straight up.