Cable construction selection best practices for avoiding noise in control circuits 1

[wroggent]

I am looking for general advice, rules of thumb, or best practices but I also have a specific project I am looking for guidance on.

Overview of my project: I need to install instrumentation and control cable from an existing DCS cabinet to a new junction box. Cable distance will end up being 700 feet or so. From the junction box others will installing cable to instrumentation and valves which I believe may be another 50 feet or so of cable length. The instruments are loop powered 4-20mA devices. The other party will be installing shielded twisted pair cable for the 4-20mA devices. From the junction box to the DCS cabinet we will install a multiconductor twisted pair cable with individual and overall shields. There are a few solenoids valves that we will be controlling from the DCS. The position of the valves are signals that are being brought back to the DCS. These discrete input and output signals are 120Vac. From each valve to the junction box the other party is proposing to use an untwisted multiconductor cable with an overall shield (cable will contain wires for valve open command, valve close command, common for valve open and close commands, valve opened indication, valve closed indication, and common for valve opened and closed indication). The cables from the junction box to the valves will not be in close proximity to "power" cables. The cables from the DCS area to the junction box area will be installed in a new underground duct bank. This duct bank will also have (pretty low) power cables in other conduits.

Where I am unsure what to do: Unlike the seemingly universally accepted practices for analog signals, I can't seem to find any hard-and-fast rules about what kind of cable to use for discrete I/O nor any methods to make this determination based on specific circumstances. Anecdotally, at my first job I was doing some work at a newly built plant. That particular installation could not have been worse: control and power cables in a shared cable tray; unshielded untwisted multiconductor cables for discrete I/O; among the power cables were random lay single core cables providing power to relatively large 480V motors from VFDs. There were several valves that had position indication brought in to the DCS (frequently appearing as "ZSC" and "ZSO" on P&IDs in the industry), just as I am dealing with now on my project. At that plant the position indicators never worked due to the noise - the valves were always shown as opened and closed simultaneously. This is the issue I wish to avoid plus I have concerns about being to open and close the valves. Also anecdotally, future plant builds used individually shielded twisted pairs for discrete I/O and power and control/instrumentation cable was kept in separate trays, and we never experienced similar issues.

Following is a list a related and semi-related threads, some of which contain similar issues. (Mainly providing for future researchers on this topic that may come across this thread)

https://www.eng-tips.com/viewthread.cfm?qid=100098

https://www.eng-tips.com/viewthread.cfm?qid=447254

https://www.eng-tips.com/viewthread.cfm?qid=356589

https://www.eng-tips.com/viewthread.cfm?qid=214666

https://www.eng-tips.com/viewthread.cfm?qid=191392

https://www.eng-tips.com/viewthread.cfm?qid=308092

There are several mitigating techniques available but I don't know what is necessary or how to justify their expense. The following all seem like prudent things to do but with varying additional cost to the project:
1. Separate power and control/instrumentation duct banks
2. Use only steel conduits (rather than PVC as intended)
3. Do not use "common" wires for the solenoids or valve position signals
4. Use individually shielded and twisted pair cable for the discrete I/O

This is not my area of expertise although as mentioned above I have experience with things gone wrong. My coworker whose area of expertise includes this topic does not seem concerned with doing what the other party is doing - just use a cable with an overall shield. The purpose of this thread is to seek some additional opinions.

So, what kind of cable to you use for discrete I/O?

Other considerations: Given the distances and control voltage, would you expect capacitive coupling to be a problem? Would a DC control voltage help (or be necessary) and/or be advised in this situation?

Also, I don't have IEEE 518 and it is a withdrawn standard, but is it still worth owning/purchasing?


Edit:
Reading for future researchers:

https://isajobs.isa.org/standards-a...ractices-for-process-instrumentation-cabling/

http://www.smar.com/en/technical-ar...ize-their-effects-in-industrial-installations

https://electrical-engineering-port...n-enter-a-signal-cable-and-its-control-part-1

https://electrical-engineering-port...n-enter-a-signal-cable-and-its-control-part-2

http://www.moeller.net/binary/ver_techpapers/ver949en.pdf

 

[7anoter4]

At first, our practice in power stations, was running instrument cables on separate cable trays solid bottom and covered usually of steel [or aluminum] at the uppermost position, above the low voltage power cable trays-one feet distance- and at the lower level medium voltage shielded cables- shield grounded both ends.
Outdoor run was in duct banks usually of 4-6" steel or cast iron at the uppermost level ,lower rows low voltage power cables and lower medium voltage. Here only 10" was the distance between rows.
The standards as IEEE 422, 518 and other are now updated in IEEE 63[2017?]. I have not with me this standard and I cannot detail any thing from here.
For shield grounding see IEEE 1100 /2006 [mainly ch.8]

 

[itsmoked]

Run analog signals in sheilded twisted pairs with no common returns. You can run them together.

Conduit or not so PVC is fine.

No closer than 6" from low voltage 24VDC or less, this includes relay driving and switch contacts.

No closer than a foot from "AC power wires".

Do not run them within 2 feet of any wires having to do with VFDs.


You can violate all this and it may still work depending on exactly what's being sensed and what else is going on, but the likelihood of issues will be greater and the more signals and digital I/O the more fleeting and puzzling problems will be to fix.

You should have no capacitive coupling issues following the above recipe and none between analog signals at all. If you have high speed signals that requires you pay close attention to the sheilded twisted pair cable you select because too much capacitance and the analog signal will be washed out. Best to avoid high pair cables if this is the case.

If you want big multipair cable runs that's fine but you need different cables to carry the different species above. Do not run digital I/O in the same big cable with analog.

I find running single sheilded twisted pairs to be more useful than a big fat cable and usually a whole lot cheaper and more available. This is because the price of cable plunges with quantity and you could buy 20,000ft of cable you will use there and at future jobs as compared to having to buy just a 1000ft roll of crazy expensive multipair cable throwing way 300ft, or worse, saving it for a million years hoping you need some 40 pair cable for something.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[7anoter4]

See also:

https://electrical-engineering-portal.com/cable-spacing-as-a-means-of-noise-mitigation

 

[dpc]

120 V AC solenoid valves can have significant inductive spike when de-energized. I would keep that wiring separate from any instrumentation wiring. 4-20 mA current signals are pretty robust. Shielded twisted pairs should provide adequate isolation.

 

[bboosstteerr]

Hi OP,

I'm curious whats your best practice for the analog signals? any references you used, thank you. sorry oot.

 

[wroggent]

bboosstteerr,

Above 7anoter4 mentions IEEE 442 so I purchased that. Paragraph 6.4.2 suggests 4 options for reducing noise in instrumentation circuits: provide physical separation from any noise source, install instrumentation cable in a magnetic conduit, use shielded twisted pair cable, install instrumentation cable separate from all power and control cables. Paragraph 8.3 essentially says that shielded cable is common for instrumentation cables (and typically feature twisted pairs). This standard defines instrumentation as "..variable current or voltage signals (analog)..."

In the bibliography for IEEE 442 one of the references is IEEE 1143 Guide on Shielding Practice for Low Voltage Cables. I've not yet purchased 1143 but looking at the table of contents it may offer more information on the treatment of cables for analog signals.

itsmoked mentioned some items that align with my experiences:
-Analogs should use shielded twisted pairs
-Do not use common returns
-Do not use other signal types/voltages in a multiconductor cable with analogs

Be sure to use single point grounding for analogs unless the circuit length becomes significant compared to the wavelength length (1/20th).

From the ISA link in the OP, the NSL-1 classification applies to analog signals.

API RP 552 and PIP PCCEL001 look like they could be valuable references.

 

[wroggent]

Clarification to the OP - my use of the word "discrete" in discrete I/O is to mean on/off signals. Some folks refer to these as "digital".

 

[bboosstteerr]

wroggent,

thank you for awesome answer Sir, sorry I can't contribute to the thread due to my experience only 1 years in instrumentation