data-acquisition.us

[Guest]

There is an application where there exists a 4-20 mA transducer at the end of a 250 ft long cable. This entire setup up is in a "harsh" electrical environment (i.e., lots of RF and generator noise created by gigantic generators, motors and nasty power).

In this application, the engineer is using a 4-conductor (2 twisted pairs, one pair not used) cable with a foil shield and a drain wire; the engineer has connected the shield at one end only (it is floating at the sensor side) and has connected the other end of the shield to Analog Ground. He is also using a 400-Ohm sense resistor and a unity-gain follower before getting to the A/D.

This setup works well in the lab, but fails miserably at the field location. The DAQ system goes wild with noisy (as in white noise up to the rails of the A/D input levels) data.

The engineer has tried: Common-mode chokes on the 4-20 mA lines, he has made sure that he is not connecting the Analog Ground to the Case Ground (Earth).

Questions:

- Should he connect the shield to the Analog Ground of the 12V DC power supply (and, therefore, the AGND of the sense resistor, Buffer Amp and the A/D as well) or to the Case Ground which is tied to the safety Earth of the AC power supply and the EARTH/GND pin of my 110VAC connector?

- Is it true the shield itself only protects against capacitively-coupled nois? The common mode chokes should clean up any common mode noise. So why is the engineer getting crappy data?

[daqmaster]

Advice:

The shield is probably bringing in all the interference so connect it to the ground (earth) at both ends and not your analog ground. Powering the 4-20ma loop from a separate, isolated supply helps as well as low-pass filtering on the A/D input circuit.

What type of instrumentation is the engineer using? Say it's NI SCXI chassis with the instrumentation modules, as another engineer has. This engineer experienced similar problems as above and found that the common-mode drain line was connected to chassis ground, which in turn was connected to earth ground. The shield drain wire was attached to the earth ground at the A/D converter module and left to float at the transmitter.



The pressure transmitter was connected to a pipe that was isolated from the earth ground by rubber hose and PVC pipe. The company was using a VFD drive to control the pump motor. Apparently, VFD drives produce a tremendous amount of electrical noise into the environment. The isolated section of piping, with the pressure transmitter piped to it, was picking up the noise from the VFD drive and sinking it down the signal wires, through the common mode drain resistors, and onto earth ground.

The solution was to earth ground the isolated section of piping. Once that happened, the signals ran smooth.

There may be a communications problem rather than a grounding problem.

Try using industrial-grade instrumentation cable from Belden or similar. Don’t forget to terminate both sides of the line with the same EOL resistor, matching the characteristic impedance of the cable. Normally, it is 100 Ohm nominal, not the 400R you intend to use. If it still doesn’t work, try an unbalanced connection with properly designed line drivers, or, use an off-the-shelf converter for RS-485 data transmission. That may certainly work.

Orig engineer asks:

If I connect the shield to Earth on both sides do I not just create a 500 ft antenna?

Analog 4-20 ma Loop....There is not digital data here.. how would RS485 apply?

The isolated section of pipe attached to the transducer (which was not connected to your shield) was picking up noise form the VFD and coupling it to ground through the signal wires -- correct? Once it grounded the transducer case itself, the problem went away? Correct?

Orig engineer reports that…

…his transducer's case is earth-grounded at the install point (because there a giant WECO hammer union attaching it to the rig itself). A recent mod was to add a common-mode line filter to AC power input line and add common-mode protection to the transducer signal wires themselves. Theoretically, these should remove any RF or EM coupled common mode noise into the signal lines. The shield is still conneccted to analog ground at the A/D.

He thinks these have cleared up the data to the point where everything works reasonably well right now. He is still having some problems with the digital side of things but all that seems to be protocol layer stuff -- nothing that a couple of months of recoding won't fix.

On the system design level, all the DAQ is custom. A 12V power supply (Lambda Switching Power Supply) is used to drive the transducer and a 400 Ohm sense resistor. The resulting signal is passed through a unity gain Buffer Amp (LTC1151) to a 16 bit A/D (AD976) to two processors: one which drives a thermal strip chart while the other converts the data into serial data to be sent to a laptop if needed. He cannot use off-the-shelf stuff because people on drilling rigs -- where he works -- have a tendency to wash down equipment with a fire hose once in a while -- so everything has to be sealed and protected very tightly.

The engineer is still confused -- he wonders if he should start completely from scratch and do the following:

1) Use an isolated AC-DC power supply to power my 4-20 mA transducer. As applied to a 24VDC power supply.
2) Connect a 400 Ohm sense resistor, to the return line so he can convert the loop current to a voltage between 1.6V and 8.0V. This means that even if he rails out the transducer, he still had 16V of compliance left to power the transducer itself.
3) Use a unity-gain follower to buffer the 1.6V to 8.0V signal to have a very low impedance output. He does this so he doesn't have to worry about loading the input of the A/D converter.
4) Where should he connect the shield? Basically these three choices, right?
a) The Neutral Line of the AC input
b) The Earth (ground) Line of the AC input
c) The Gnd of the DC Power Supply which is also the ground reference for Amp and A/D.
So Which one should he use?

Also what about driving the shield to prevent noise?
---
Back to advice:

As it's grounded at both ends does that matter? At the moment the "antenna" is connected to the engineer's equipment. As to where he should connect it, the rig ground, i.e. the metalwork and not the instrumentation.

Grounding the far end through a capacitor (1uF?ish) would reduce 50/60Hz ground loops but still get rid of higher-frequency garbage.

Has the engineer tried double shields?

In a similar situation (not as long though), inner shield was connected as follow:

Signal >>> Instrumentation Amplifier >>>> DAQ >>>> Inverting AMP >>> Shield opened at the other end.
Top layer shield was grounded both ends.

Could it be some problems with an un-isolated instrument. If the 4-20 ma output of the instrument is not isolated, then a ground loop exists. Try placing a loop-power optical isolator between the instrument and the converter.