Current transformer for coupling noise into a cable

[Nikhil N]

I want to couple white noise to a communication cable. I am using a 50/5A CT. My idea is to connect the noise in the secondary and to couple that noise into the cable. I want to know will my idea work? Does the CT what I am using will be suitable?

 

[Paul Colby]

what kind of cable? coax?

 

[Nikhil N]

what kind of cable? coax?

Yah coax, I am now using cat5

 

[Paul Colby]

so i expect the answer is no since a balanced cable will link no flux.

 

[Nikhil N]

so i expect the answer is no since a balanced cable will link no flux.

But I have seen people using current probs for BCI(Bulk Current Injection) tests..

 

[Paul Colby]

In cables like CAT5 and coaxial cables the transformer winding will excite the common mode while the cable usually is used for differential mode.

 

[Paul Colby]

What I mean by this is the transformer winding will induce current in the same direction in both inner and outer conductors. This induces 0 volts across the line or no signal.

 

[Nikhil N]

In cables like CAT5 and coaxial cables the transformer winding will excite the common mode while the cable usually is used for differential mode.

Actually I have two devices which are communicating with modbus protocol, I need to analyze the efficiency of the communication under the effect of noise, which are very common in industries. For that I need to inject the noise into the cable either capacitively or inductively. As CT can be easly used I am using that for this purpose

 

[Paul Colby]

Okay, The cable is passed through the winding hole in the device you've shown. Get a scope and a signal source and do a test. If it works you're golden. If not then what I've said holds.

 

[Nikhil N]

Okay, The cable is passed through the winding hole in the device you've shown. Get a scope and a signal source and do a test. If it works you're golden. If not then what I've said holds.

Ohk..Thanks

 

[Paul Colby]

Of course if one splits the cable conductors and passes only inner (or outer) conductors through the CT then the difference mode would be excited and the signal will be injected. Somehow I expect this is not what you're after.

 

[Nikhil N]

Of course if one splits the cable conductors and passes only inner (or outer) conductors through the CT then the difference mode would be excited and the signal will be injected. Somehow I expect this is not what you're after.

Actually in industrial environment the noise/interference will be couple as common to both line, I can not go for differential mode . In fields we have already faced issue of data loss, so I need to know how much is the noise effect on the data loss. I can not perform this now in industrial environment.

 

[Paul Colby]

If you need to measure common mode noise then your method will work just fine.

 

[Baluncore]

For that I need to inject the noise into the cable either capacitively or inductively. As CT can be easly used I am using that for this purpose

A CT will not be operating as a CT in that application. It will be working more like a voltage transformer with a one turn secondary.

The CT will inject common mode noise onto the bundle of conductors in the cable.

A CT usually has a laminated core that is designed for the bandwidth of the power being measured. You need to use a transformer that will pass the bandwidth of the noise you want to inject. Consider using an iron powder or ferrite toroid.

 

[jim hardy]

Common mode interference gets transformed into differential mode by unbalanced impedance to earth. So an experiment will show how well your CT works on that particular system.

Baluncore is right, for "white noise" you want a wideband current transformer that uses a core suitable for a high frequency .
We used a big one around the neutral of our main generator to listen for 1 mhz noise , evidence of internal arcing.

Search on wideband current transformer turns up several vendors
http://www.lilco.co.uk/ was first on mine

but i'll bet you could make one out of the inductors inside a line filter.
Take apart a dead PC power supply. They're rich with inductors.

old jim

 

[Svein]

Noise injectors...

Way back when I started out in professional electronics, our main noise generator was the old Weller soldering irons. Turning them on and off generated wideband noise.

Another version thought up by one of the guys: Get hold of a coarse file and a high voltage capacitor (about 10nF). Put on insulating gloves, connect one mains phase to the file and the other to one lead of the capacitor. Rub the other lead of the capacitor against the file.

I also once found a switching supply with too long leads in the PCB layout. A short-circuited oscilloscope probe detected noise spikes one meter away.

 

[berkeman]

Yah coax, I am now using cat5

Cat5 is twisted pair, not coax, AFAIK.

https://en.wikipedia.org/wiki/Category_5_cable

It sounds like you are wanting to do a conducted common-mode noise test similar to the industry standard EN 61000-4-6 test:

http://www.elmac.co.uk/pdfs/PNP_61000-4-6.pdf

 

[tech99]

I have used a ferrite RF toroid surrounding a cable to measure common mode RF current. It should work in reverse.The toroid requires a few turns of wire depending on frequency of interest. You can calibrate it using a signal generator or a transmitter. It is possible to split the toroid so it can be placed around a cable, then clipped together. The CT looks to be designed for power frequencies.

 

[Nikhil N]

A CT will not be operating as a CT in that application. It will be working more like a voltage transformer with a one turn secondary.

The CT will inject common mode noise onto the bundle of conductors in the cable.

A CT usually has a laminated core that is designed for the bandwidth of the power being measured. You need to use a transformer that will pass the bandwidth of the noise you want to inject. Consider using an iron powder or ferrite toroid.

Thanks for the answer, the CT what I have is 50/5A, 50/60Hz. What is mean by 50/60Hz which is written over the CT? Does it mean the bandwidth is 60Hz? My noise will have frequency upto 20KHz, in that case will this transformer works?

 

[Baluncore]

Does it mean the bandwidth is 60Hz? My noise will have frequency upto 20KHz, in that case will this transformer works?

Your CT has a steel laminate core that is designed to work on a 50/60 Hz power supply. It will probably work only up to the 5th harmonic at 300 Hz. You will need to find a core that is designed to work at 20 kHz and build your own transformer. Find one in a switching power supply or the horizontal output transformer of an old CRT TV.

If the ends of the cable are transformer coupled you should inject a common mode voltage difference into the bundle. The turns ratio of the voltage transformer will depend on your signal source and the voltage you want to induce. You may gain an advantage by putting more than one pass of your cable through the transformer.

 

[tech99]

Thanks for the answer, the CT what I have is 50/5A, 50/60Hz. What is mean by 50/60Hz which is written over the CT? Does it mean the bandwidth is 60Hz? My noise will have frequency upto 20KHz, in that case will this transformer works?

My experiments were much higher in frequency but used a ferrite for that frequency. I think you could try the CT under known conditions and see if it is working. Put the CT on a length of wire forming a big loop, and connect this loop to a scope. Connect a signal generator to the CT coil and see if you have transformer action happening.

 

[tech99]

Thank you, Baluncore, for practical information.

 

[Nikhil N]

Your CT has a steel laminate core that is designed to work on a 50/60 Hz power supply. It will probably work only up to the 5th harmonic at 300 Hz. You will need to find a core that is designed to work at 20 kHz and build your own transformer. Find one in a switching power supply or the horizontal output transformer of an old CRT TV.

If the ends of the cable are transformer coupled you should inject a common mode voltage difference into the bundle. The turns ratio of the voltage transformer will depend on your signal source and the voltage you want to induce. You may gain an advantage by putting more than one pass of your cable through the transformer.

If I am operating with the CT with noise upto 20KHz, will it cause heating problem ? Other than heating problem will it cause any problem?

 

[Baluncore]

If I am operating with the CT with noise up to 20KHz, will it cause heating problem ?

With the old CT you will have trouble with voltages over about 1 volt. There will not be problems if you use the right core material and turns ratio.

At the moment we do not know what you mean by noise, or what V RMS amplitude it has.

 

[Nikhil N]

Your CT has a steel laminate core that is designed to work on a 50/60 Hz power supply. It will probably work only up to the 5th harmonic at 300 Hz. You will need to find a core that is designed to work at 20 kHz and build your own transformer. Find one in a switching power supply or the horizontal output transformer of an old CRT TV.

If the ends of the cable are transformer coupled you should inject a common mode voltage difference into the bundle. The turns ratio of the voltage transformer will depend on your signal source and the voltage you want to induce. You may gain an advantage by putting more than one pass of your cable through the transformer.

Does In the second paragraph what you have suggested mean by the following diagram?

 

[Baluncore]

The noise injected there is differential noise, not common mode noise. I would want to know why only slave 4 is isolated with a voltage transformer.
If you inject a current into the bundle then you must ask where the induced currents will go. You need a return circuit for current.
Earth or ground is not everywhere the same voltage. There can be big differences in voltage between different sites.

 

[Baluncore]

Modified to inject a common mode voltage.

 

[Nikhil N]

View attachment 112743

Modified to inject a common mode voltage.

Thanks for the modified circuit , Can you please suggest me the type of voltage transformer too with its ratings, if noise is of 10V (max peak-peak), Bandwidth - 100MHz ?

 

[Paul Colby]

View attachment 112743

Modified to inject a common mode voltage.

It's unclear to me that this modification does as advertised. Wouldn't an ideal center tapped transformer generate no flux in the core if the master line is balanced. Ideally this will not inject either a differential or common mode signal in slave 4. The transformers I've wound on toroids of this type tested at 90dB of isolation.

 

[Nikhil N]

It's unclear to me that this modification does as advertised. Wouldn't an ideal center tapped transformer generate no flux in the core if the master line is balanced. Ideally this will not inject either a differential or common mode signal in slave 4. The transformers I've wound on toroids of this type tested at 90dB of isolation.

what you have said is correct, if the line is balanced it won't generate flux . That means, we should go for a toroid type transformer right? Can you suggest me how can I make a toroid by myself for my purspose? (Noise used is of 10V(max), 100MHz(B.W.))

 

[Paul Colby]

From a topology point of view one could wind a transformer using the transmission cable as a secondary and a single conductor as a primary. The noise would be injected in the primary. Your initial concept works from a topology point of view. The 100MHz bandwidth is centered on what frequency?

 

[Baluncore]

Wouldn't an ideal center tapped transformer generate no flux in the core if the master line is balanced.

I thought you were testing to see if the line was balanced and therefore immune to CM noise.

 

[Nikhil N]

I thought you were testing to see if the line was balanced and therefore immune to CM noise.

Actually my experiment is..The master and slaves are communicating using modbus protocol, and I need to analyze the effect of noise on data loss in this communication. The line length is about 1000ft and communication standard used is RS485.

 

[Paul Colby]

Okay, so maximum is 12Mbps so much less than 100MHz bandwidth centered on your baud rate should work, right?

 

[berkeman]

Actually my experiment is..The master and slaves are communicating using modbus protocol, and I need to analyze the effect of noise on data loss in this communication.

That still sounds like a low-frequency version of EN 61000-4-6. We do tests like that all the time here in our lab.

Driving the center tap does cause a common-mode signal to be propagated down the wire. The fluxes do cancel, but the center tap is galvanically connected to both wires. The common-mode signal propagates down the wires to the receivers, which tests their common-mode rejection at their inputs. Especially for a DC-connected network like this one appears to be, the CM rejection will be pretty low. Nothing like it is for a transformer-connected multi-drop network.

EDIT -- the Leakage Inductances (Lk) will limit the bandwidth of this drive technique, but for the frequencies you are asking about, and the modest common-mode capacitance, it shouldn't be much of a problem. It's worth calculating it, though.

 

[Paul Colby]

Driving the center tap does cause a common-mode signal to be propagated down the wire.

For sure. Just move the center tap to the device under test rather than what's shown.

 

[berkeman]

For sure. Just move the center tap to the device under test rather than what's shown.

But the network devices shown are DC-connected to the twisted pair network. The only transformer shown is the coupling transformer...

 

[Paul Colby]

View attachment 112743

Modified to inject a common mode voltage.

The thing with the blue circle is what I was referring to. Seems to me the center tap is shown on the wrong side but what ever. I'm going to bow out. Just don't have the expertise nor can I tell what the OP is really after. The charm of the inductive injection coil he originally posted would be not having to introduce new components into the system. If your going to modify the circuit I'm sure there 1000 ways to do it with a simple network.

 

[Baluncore]

If your going to modify the circuit I'm sure there 1000 ways to do it with a simple network.

I agree.
To select one of those 1000 circuits we need to know why the test is being done and what will constitute a pass mark.

 

[berkeman]

The charm of the inductive injection coil he originally posted would be not having to introduce new components into the system.

Exactly. That's why the EN 61000-4-6 industry-standard test uses a broadband CM noise injection coil.

 

[Nikhil N]

That still sounds like a low-frequency version of EN 61000-4-6. We do tests like that all the time here in our lab.

Driving the center tap does cause a common-mode signal to be propagated down the wire. The fluxes do cancel, but the center tap is galvanically connected to both wires. The common-mode signal propagates down the wires to the receivers, which tests their common-mode rejection at their inputs. Especially for a DC-connected network like this one appears to be, the CM rejection will be pretty low. Nothing like it is for a transformer-connected multi-drop network.

EDIT -- the Leakage Inductances (Lk) will limit the bandwidth of this drive technique, but for the frequencies you are asking about, and the modest common-mode capacitance, it shouldn't be much of a problem. It's worth calculating it, though.

Actually I don't have much knowledge about these, But I have one more doubt, the slave and master connected only using two wires and there is no ground wire for a reference. So in that case if I am using a center tapped transformer where will I connect the ground of my noise source ?

 

[Nikhil N]

From a topology point of view one could wind a transformer using the transmission cable as a secondary and a single conductor as a primary. The noise would be injected in the primary. Your initial concept works from a topology point of view. The 100MHz bandwidth is centered on what frequency?

I don't know at what frequency it is centered on now. I will check its spectrum and let you know that

 

[Paul Colby]

Unclear I can be of much further assistance. A measurement spec was given so I'd suggest following that if this is a work related question. One note, if the bore of the toroid is large enough the transmission line could be looped multiple times through upping the generated noise voltage and modifying the turns ratio.