EMI(Electro Magnetic Interference) Magnitude and Frequency

  • Thread starter Nikhil N
  • Start date
  • #1
81
2
I want to analyze the efficiency of my data communication between slaves and master in the presence of interference. The protocol what I am using is modbus, with baud rate of 9600 , I am using 1000 ft cable. What I am struggling to do is, how to couple the noise into the cable in a way it is coupling in real time industrial environments. Also I dont know the voltage levels and frequencies of interference coming from motor and compressor and all which are common in industries. Any body knows the EMI level and frequencies of the interference common in industrial environments ?
 

Answers and Replies

  • #2
davenn
Science Advisor
Gold Member
9,538
8,567
Any body knows the EMI level and frequencies of the interference common in industrial environments ?

it's going to vary substantially depending on the environment, no two are going to be the same or any anything like each other
every bit of electronics is going to produce it's own specific range of frequencies at what ever level
( the only thing you can tell is for gear like that is that it is supposed to conform to what ever the emission standards are in your country and there should be a label on electronic gear stating its conformance)

electric motors are wide and varied in their power and design and their RF noise emissions will also vary depending on load and other variables... they are not my specialty but I would assume they should have an EMI rating sticker as well ?

@jim hardy ..... any thoughts ? :smile:


Dave
 
  • #3
tech99
Gold Member
2,154
799
I would suggest using the standard test specifications for "conducted interference" used for product testing. For instance, the European CE Mark test specifications or the FCC equivalent.
 
  • #4
jim hardy
Science Advisor
Gold Member
Dearly Missed
9,847
4,883
@jim hardy ..... any thoughts ? :smile:


I used to walk around the power plant with a plain old pocket transistor radio tuned between stations.
That is a "Poor Man's RF Noise Detector".

Electric motors by themselves are well behaved unless the insulation is wet or damaged in which case they produce a lot of static. Same effect as driving under power lines, your car radio picks up corona from the insulators. A healthy machine doesn't emit significant RF.

Line frequency of course is too low for the radio to hear. So i wound a coil for a flux detector, and connected it to a portable oscilloscope.
I taped to a piece of cardboard 10 turns encircling 1/10 square meter so as to get readout in webers/sec.
I found motors also well behaved at line frequency. They're designed to keep the magnetic flux inside the iron so their 60 hz emissions are also minimal.
Immediately adjacent a single phase transformer my coil could pick up tens of millivolts from leakage flux, but it was not sinusoidal instead peaky looking i assume due to the flux running up to knee of saturation curve each cycle. That'll be rich in harmonics of line frequency.

The only places my coil measured significant line frequency magnetic flux were adjacent the main generator leads that carried about 20 kiloamps at 60hz , and near some large air core inductors in the switchgear carrying just a few kiloamps. My coil produced ~two volts in both places.

The computer room produced a roar on my cheap transistor radio and millivolts of wideband noise on my coil. Computers themselves are noisy, that's why they use differential RS422-like drive for lines of any length.

We didn't have any large VFD electric motor drives so i can't tell you how they behave.
We did have a 400 amp SCR controller but it switched at zero crossing so was "RF-Quiet" . Thank Goodness !

Long cables that run clear across the plant sometimes arrived in control room with as much as a few (worst i recall was seven) volts of 60 hz on them. Grounding the shields usually reduced that enough so it wasn't troublesome.

What DID drive our computers nuts was arcing at relay contacts. That induces short bursts of high frequency noise into all nearby cables and on our simulator it would corrupt the differential I/O bus . My coil saw several volts adjacent a visibly arcing relay contact. We found the arc suppression snubbers at its contacts failed open, replacing them solved that problem.

So to your questions

Motors and compressors when running are benign, unless their insulation system is in trouble.
Arcing of any sort is NOT benign.
So route your Modbus cables away from motor starters, relay cabinets and phase angle control SCR systems.
Pay attention to grounding of shields.
Placing your MODBUS cables in steel conduit would help immensely. Insulating that conduit from structural steel and grounding it only at receiving end sounds extreme, but that's how our reactor instruments were done.

My plant had a special instrument ground cable encircling control room . It was tied to the plant ground mat well below the building so that power system ground currents would not flow in instrument ground conductors.
Dedicated lightning protection system ground rods were separate and went well below the plant ground mat so as to make an equipotential surface underneath the whole plant, a "magic carpet" to ride smoothly over lightning if you will.

So to your questions -
In my plant most noise was a few tens of millivolts or less with some exceptions up to a few volts.
EPRI did a research project in mid 70's measuring noise in plant instrument systems. Their first question was "Where's all your 60 cycle interference?"
That felt good, for we'd invested a lot of work during startup chasing and eliminating noise.
The researchers nicknamed us "That plant with the quiet signals".
So -a few volts might be more typical.

Spurious noise from switching will give you the most trouble. That's hard to capture . So try to avoid sources of it. When you find a source, fix it.

Sorry for boring ramble, but all i can do is share my experience. Hopefully it'll be food for thought in your testing.
 
Last edited:
  • Like
Likes GrahamN-UK, Nikhil N and davenn
  • #5
jim hardy
Science Advisor
Gold Member
Dearly Missed
9,847
4,883
If you're just doing practical testing, i'd try a hobbyist Tesla Coil
and a 250 watt soldering gun with a coil of big wire replacing the tip.

Former will make plenty of RFI
latter will make a goodly line frequency magnetic field
both of which you can measure with everyday test equipment

If either upsets your MODBUS then you have reason to do more formal testing.

I've even used a cellphone to test a circuit . We replaced an old discrete component opamp big as a handkerchief with a modern IC .
Placing a celllphone nearby saturated the old one. New one was affected not enough to measure.
Since those days(1990's) that facility has got a proper RFI test rig with a big horn antenna .
 
Last edited:
  • #6
81
2
I used to walk around the power plant with a plain old pocket transistor radio tuned between stations.
That is a "Poor Man's RF Noise Detector".

Electric motors by themselves are well behaved unless the insulation is wet or damaged in which case they produce a lot of static. Same effect as driving under power lines, your car radio picks up corona from the insulators. A healthy machine doesn't emit significant RF.

Line frequency of course is too low for the radio to hear. So i wound a coil for a flux detector, and connected it to a portable oscilloscope.
I taped to a piece of cardboard 10 turns encircling 1/10 square meter so as to get readout in webers/sec.
I found motors also well behaved at line frequency. They're designed to keep the magnetic flux inside the iron so their 60 hz emissions are also minimal.
Immediately adjacent a single phase transformer my coil could pick up tens of millivolts from leakage flux, but it was not sinusoidal instead peaky looking i assume due to the flux running up to knee of saturation curve each cycle. That'll be rich in harmonics of line frequency.

The only places my coil measured significant line frequency magnetic flux were adjacent the main generator leads that carried about 20 kiloamps at 60hz , and near some large air core inductors in the switchgear carrying just a few kiloamps. My coil produced ~two volts in both places.

The computer room produced a roar on my cheap transistor radio and millivolts of wideband noise on my coil. Computers themselves are noisy, that's why they use differential RS422-like drive for lines of any length.

We didn't have any large VFD electric motor drives so i can't tell you how they behave.
We did have a 400 amp SCR controller but it switched at zero crossing so was "RF-Quiet" . Thank Goodness !

Long cables that run clear across the plant sometimes arrived in control room with as much as a few (worst i recall was seven) volts of 60 hz on them. Grounding the shields usually reduced that enough so it wasn't troublesome.

What DID drive our computers nuts was arcing at relay contacts. That induces short bursts of high frequency noise into all nearby cables and on our simulator it would corrupt the differential I/O bus . My coil saw several volts adjacent a visibly arcing relay contact. We found the arc suppression snubbers at its contacts failed open, replacing them solved that problem.

So to your questions

Motors and compressors when running are benign, unless their insulation system is in trouble.
Arcing of any sort is NOT benign.
So route your Modbus cables away from motor starters, relay cabinets and phase angle control SCR systems.
Pay attention to grounding of shields.
Placing your MODBUS cables in steel conduit would help immensely. Insulating that conduit from structural steel and grounding it only at receiving end sounds extreme, but that's how our reactor instruments were done.

My plant had a special instrument ground cable encircling control room . It was tied to the plant ground mat well below the building so that power system ground currents would not flow in instrument ground conductors.
Dedicated lightning protection system ground rods were separate and went well below the plant ground mat so as to make an equipotential surface underneath the whole plant, a "magic carpet" to ride smoothly over lightning if you will.

So to your questions -
In my plant most noise was a few tens of millivolts or less with some exceptions up to a few volts.
EPRI did a research project in mid 70's measuring noise in plant instrument systems. Their first question was "Where's all your 60 cycle interference?"
That felt good, for we'd invested a lot of work during startup chasing and eliminating noise.
The researchers nicknamed us "That plant with the quiet signals".
So -a few volts might be more typical.

Spurious noise from switching will give you the most trouble. That's hard to capture . So try to avoid sources of it. When you find a source, fix it.

Sorry for boring ramble, but all i can do is share my experience. Hopefully it'll be food for thought in your testing.
Thanks for sharing your experience and Its interesting to know these kind of experiences .
 

Related Threads on EMI(Electro Magnetic Interference) Magnitude and Frequency

  • Last Post
Replies
3
Views
1K
  • Last Post
Replies
8
Views
6K
  • Last Post
Replies
2
Views
1K
  • Last Post
Replies
1
Views
2K
Replies
6
Views
2K
Replies
3
Views
3K
Replies
7
Views
2K
  • Last Post
Replies
4
Views
6K
  • Last Post
Replies
11
Views
1K
Replies
6
Views
2K
Top