EMC Radiated Susceptability of Cable - Antenna problem

[human7]

I am trying to calculate the voltage induced onto a shielded cable by a 200 V/m external electric field. I found an antenna equation View attachment power received.bmp that relates the power received to the incident electric field, but I need to define an antenna gain (G) for the cable. I am not sure how to get that. Am I on the right track? Any help would be appreciated.

 

[berkeman]

I am trying to calculate the voltage induced onto a shielded cable by a 200 V/m external electric field. I found an antenna equation View attachment 39723 that relates the power received to the incident electric field, but I need to define an antenna gain (G) for the cable. I am not sure how to get that. Am I on the right track? Any help would be appreciated.

What is the application? How is the shield terminated? What frequency range are you working with? How long is the cable, and how is it terminated (both in common mode and in differential mode)?

200V/m is a darned strong field. What is the source? Is this in a real-world installation, or in a shielded anechoic test chamber?

 

[human7]

What is the application? How is the shield terminated? What frequency range are you working with? How long is the cable, and how is it terminated (both in common mode and in differential mode)?

200V/m is a darned strong field. What is the source? Is this in a real-world installation, or in a shielded anechoic test chamber?

Application: Shipboard equipment.
Outer Shield Termination: 360 deg termination on both ends to grounded eclosures
Frequency Range: 2 MHz to 18 GHz
Cable length: 10 m max

200V/m comes from the RS103 (MIL-STD-461) levels specified for above deck equipment. The potential source is radar.

We have used EMI best design practices, but I need to show at least on paper that our system should not be susceptible.

 

[cmb]

The question relies on the termination to ground. The cable has an impedance which almost wholly depends on the kit it is connected to and its routing. 'EMC of a cable' generally makes no sense.

If the outer shield is a perfect conductor, the end terminations are perfectly connected to a perfect ground, and there is no split, junction, hole or other orifice, then the cables inside are fully protected. The issue for EMC comes down to how close to 'perfect' are you, which is not a question that can be addresses by some hand-wavy concepts/maths.

You either argue that your screening is sufficient, or you take it to a test lab.

With RADAR the issue is more that it is direct radiation of the electronics attached to the ends of the cables, not the cable itself.

 

[human7]

The question relies on the termination to ground. The cable has an impedance which almost wholly depends on the kit it is connected to and its routing. 'EMC of a cable' generally makes no sense.

If the outer shield is a perfect conductor, the end terminations are perfectly connected to a perfect ground, and there is no split, junction, hole or other orifice, then the cables inside are fully protected. The issue for EMC comes down to how close to 'perfect' are you, which is not a question that can be addresses by some hand-wavy concepts/maths.

You either argue that your screening is sufficient, or you take it to a test lab.

With RADAR the issue is more that it is direct radiation of the electronics attached to the ends of the cables, not the cable itself.

If we assume a shielding effectiveness for the cable screen of 80 dB, could we not calculate the maximum voltage induced onto the cable?

 

[cmb]

If we assume a shielding effectiveness for the cable screen of 80 dB, could we not calculate the maximum voltage induced onto the cable?

No, it is still a resonance dependent on the terminations to ground at the cable ends. So, for example, if one wire in the bundle has 50 ohms at each end and another has an inductive termination, and yet another is grounded, and there are a shielded twisted pair in there, it is a biiig calculation of multiple capacitive and inductive interactions. There are ways to do it, but it's not like adding up some simple sums, especially at radar frequencies where the cable length will be many wavelengths long... that'll be multiple modes to boot.

I'll tell you, you could even test it in several test chambers and you'll still get several results, in a range, probably over 5-10dB, dependent on the test set-up.

I'm not saying that there are not ways and means to approximate, and maybe you'll find "a" means to do that which you feel comfortable with, but if you got a reply here then I'd say it'd be worth only as much as you paid for it.

You can't even use bulk current injection because the frequencies you need to test for radar susceptibilities is too high for BCI techniques.

I did this kind of work simulating wiring harnesses in cars and I could throw some ball-park figures at you, but the issue isn't this, the issue is, for example, that you have a particular resonance at a frequency where the kit at the end of the wires has a susceptibility. The cable can never fail, only the electronics at the end of it. If that is filtered and hardened enough at the cable's resonances, then it'd be a different answer to a home-brew board with no filitering.

It's a how-long-is-the-string question, I am afraid.

 

[berkeman]

I agree 100% with cmb's responses. Well stated.

 

[sophiecentaur]

The best thing to do is to use overkill. i.e follow standard practice and use everything with the correct spec for the installation type. That shouldn't be too hard to find out.
Unless you are prepared to do a lot of chasing round to measure your final, installed system performance and to change things later then you should do exactly what the equipment manufacturers and installers do. They can't afford to come back and fix 'simple' things like RFI. It's usually cheaper to do things once, properly. (Boring answer, I know but it does make sense)
P.S. You can't expect to 'calculate' an answer from any information you are likely to be able to find. You could take a sample of the (cheaper?) cable you want to use and measure its characteristics in situ, but that's more or less the same as testing the whole installation.

 

[human7]

Thanks for the responses.

 

[yungman]

I did a lot of CE certifications that included emittion and susceptibility. On top of the shielding grounding that described here, I put a ferrite toroid on the cable to make sure whatever induced to one of the conductor would induced onto the other. This is to create common mode so the differential voltage will be minimized.

Practical method aside, I am studying EM and I have a question regarding to the mechanism of how voltage is induced into the coax. I want you guys to comment on this:

According to Maxwell

\nabla \times \vec E = -\frac{\partial \vec B}{\partial t} \;\hbox { and }\; \nabla \cdot \vec E = \frac{\rho_{free}}{\epsilon}

From this, with varying E, you induce varying charge density onto the outer shield which create a varying current onto the shield. But the inner conductor is partially shielded by the shield don't see as much E so the current induced is not as much. Therefore there is a difference in the two currents which create the voltage into the coax.

At the same time, with a 200V/m varying E there MUST be B associated with the E by the Maxwell

\nabla \times \vec B = \mu\vec J +\frac{\partial \vec E}{\partial t}

But B induce equal current in both conductor and this is common mode and don't matter.


Is this the mechanism?

 

[yungman]

I guess I should post this question in Classical Physics section where people there really don't care as much in solving the problem than the analyzing the cause of problem!

 

[sophiecentaur]

I guess I should post this question in Classical Physics section where people there really don't care as much in solving the problem than the analyzing the cause of problem!

It depends what you want, I suppose.
You say that you use ferrite torroids. One effect of ferrites (the appropriate grade) is to absorb energy at high frequencies. Very useful for screening.h.

 

[berkeman]

I guess I should post this question in Classical Physics section where people there really don't care as much in solving the problem than the analyzing the cause of problem!

Actually, I didn't expect a question in your post, and only skimmed the first sentence or two...

Technically that is a thread hijack...

 

[cmb]

FWIW, here are some graphs from an EMC conference paper I gave many years ago, on a method for using statistical means to manage the spread of different transfer functions for 'test wires' in a simplified vehicle body shell.

What I did was develop means to automatically simulate a variety of conditions (specifically, this graphic is showing variations in transfer function for radiated angles/polarisations and wire position with the simulated body shell), and also statistical techniques for analysing both the magnitude and frequency differences (and, further, also discriminate the 'numerical errors' from the simulation method itself [FDTD was used for these simulations]).

Wires were ~0.5 to 5 x wavelength, in the frequency range simulated for.

The second graphic shows how you can then plot and inspect the range of data for mag and frequency differences, which you can use to then draw generalised conclusions about the relative immunity and transfer impedances between different harness layouts.

Now that I see these plots again, and come to think about it (this was 15 years ago!), the general 'rule-of-thumb' for the frequencies of interest in automotive (well, at the time, the spectrum of interest has gone up since then due to 1.8/2.5 GHz mobile) was 1mA/[V/m], which you can see is a 'general rule' that is actually supported by these simulations.

So, take it or leave it; 'an' answer is 1mA/[V/m]. (As mentioned, it is worth what you've just paid for it, but in any case, the issue rests on what the electronics does at the ends of the cable, not what the cable does.)

 

[cmb]

But B induce equal current in both conductor and this is common mode and don't matter.


Is this the mechanism?

A local near-field magnetic field (like an induction coil/probe) may induce motion in localised surface charges in both conductors, but, as mentioned, what they then go on to do is a function of how the parts are coupled to ground. If both inner and outer are relatively well coupled by low impedance, then, yes, they will mostly look similar and you could presume a common mode response. But if the inner is high impedance both ends whilst the outer is grounded, then you might expect differential modes to build up. Hence - it's more to do with the end terminations than the cable itself.

I am not so sure about what the reaction is to a propagating electro-magnetic wave. In general I would hazard a guess at an answer and say that the presence of the grounded shield is such a big discontinuity in the impedance path of the radiated signal [above some frequency] that the energy is essentially all dumped as surface current and dissipated. However, going down in frequency there will be some frequency where a propagating wave would begin to penetrate a conductive shield and behave as the magnetic case described in the other paragraph.

I will add big caveats on the above that it is just 'an interpretation' to initiate your own line of thinking, and is not a definitive answer.

 

[yungman]

It depends what you want, I suppose.
You say that you use ferrite torroids. One effect of ferrites (the appropriate grade) is to absorb energy at high frequencies. Very useful for screening.h.

I assume we are talking about RF radiation here.

 

[yungman]

Actually, I didn't expect a question in your post, and only skimmed the first sentence or two...

Technically that is a thread hijack...

I thought this thread is coming to an end already, I just inject in a question, don't mean to hi-jack the thread. In fact, I post my question on the Classical Physics section and start a discussion there.

In real life, particular in the middle of an expensive CE certification, we just slap anything on to get the noise down, no one has time to stop and think why. We have a box of ferrites that split open and can be snapped on the line and look at the spectrum analyzer to see whether it will lower the emmition.

 

[sophiecentaur]

I assume we are talking about RF radiation here.

Yes. Or at least the locally induced fields and the currents flowing along the outer surface.

It's a common method in supply leads and on circuit boards. Very effective at uhf and microwave frequencies.

 

[yungman]

I want to share something funny stories...well not so funny if you are in the middle of it against time( money)!

One time we were in the CE certification test which we have have a big system that we had to rent the whole facility...lots of money a day. We encounter a noise peak and we were stuck on it and nothing seemed to help. We narrow down to one cable and we physically inspected that it was a shielded cable, we went round and round to like 1am in the morning. we got so desperated, finally I told the technician to open the connector....here we found a need shield connected to a 4" piece of small wire before connecting to the ground pin. The assembler just stuffed the wire inside the connector. You measure, the shield was tied to ground, you look, there was a shield! Got rid of the wire and it passed with flying color!

One other time, we were struggling on a system for the longest time just to find out the pannel BNC connectors were actually an insulated connector. the lip of the connector ( the outer flange that the cable side snap onto) were not tied to the base that connect to the metal rack! We spent hours on this, it was not obvious at all the the eyes, I thought of it and inspected them multiple time and we all missed that. We were laughing after this...almost to tears!

CE test got a lot easier now a days, in the mid 90s when the requirement first imposed on products, there were very few places that perform the test, there were waiting period. You book the time slot and if you cannot pass, you had to move out as someone else always inline after you. This was the most dreadful part of the product development.

 

[berkeman]

I want to share something funny stories...well not so funny if you are in the middle of it against time( money)!

One time we were in the CE certification test which we have have a big system that we had to rent the whole facility...lots of money a day. We encounter a noise peak and we were stuck on it and nothing seemed to help. We narrow down to one cable and we physically inspected that it was a shielded cable, we went round and round to like 1am in the morning. we got so desperated, finally I told the technician to open the connector....here we found a need shield connected to a 4" piece of small wire before connecting to the ground pin. The assembler just stuffed the wire inside the connector. You measure, the shield was tied to ground, you look, there was a shield! Got rid of the wire and it passed with flying color!

One other time, we were struggling on a system for the longest time just to find out the pannel BNC connectors were actually an insulated connector. the lip of the connector ( the outer flange that the cable side snap onto) were not tied to the base that connect to the metal rack! We spent hours on this, it was not obvious at all the the eyes, I thought of it and inspected them multiple time and we all missed that. We were laughing after this...almost to tears!

LOL Been there a few times myself.

[hijack]

There was a time a while back when I was using 2-3 different EMI test ranges (indoor, outdoor) for some qualification work, and I started to suspect some calibration differences between the labs, because I was getting some unexpected results by several dB. So I built a simple battery powered reference transmitter, and started taking it to the labs. I'd have them do a quick scan of my reference device before we started scanning the products, so I'd have a baseline for that day's tests.

Well, one of the labs was not very happy with my request to scan my reference device before starting the regular tests. They eventually said okay, and we got that data and started scanning the products. Except we got really weird results at a few harmonics, something I'd never seen before for this product. It took me about half an hour to figure out that it matched harmonics from my reference transmitter.

I'd accidentally left the reference device turned on when I put it in my toolbox out under the scanning table...

[/hijack]

 

[cmb]

we just slap anything on to get the noise down, no one has time to stop and think why...

...This was the most dreadful part of the product development.

Yeah, this is what most people seem to do when they come for test, but it also seems to be what they prefer to do!

After I completed my University Research Fellowship in EMC simulation, I went on to try to develop it as a service offering (with a sister firm of one of the partners funding the project). What I found was that no-one seems to wants to pay any 3rd parties for support for 'front-loaded' EMC mitigation work - I guess too many folks are too proud to admit they don't know what they are doing, and figure in any case they just need to slap something into their kit if it fails... I gave up trying to sell EMC simulation and modelling support.

I'd suggest you can generally tell how experienced an EMC engineer is by how much they claim they don't know about EMC - the more experienced they are, the more they realize how little they know of the subject!

 

[yungman]

LOL Been there a few times myself.

[hijack]

There was a time a while back when I was using 2-3 different EMI test ranges (indoor, outdoor) for some qualification work, and I started to suspect some calibration differences between the labs, because I was getting some unexpected results by several dB. So I built a simple battery powered reference transmitter, and started taking it to the labs. I'd have them do a quick scan of my reference device before we started scanning the products, so I'd have a baseline for that day's tests.

Well, one of the labs was not very happy with my request to scan my reference device before starting the regular tests. They eventually said okay, and we got that data and started scanning the products. Except we got really weird results at a few harmonics, something I'd never seen before for this product. It took me about half an hour to figure out that it matched harmonics from my reference transmitter.

I'd accidentally left the reference device turned on when I put it in my toolbox out under the scanning table...

[/hijack]

The open range was the kicker. That was the only place available when it first started. It was March and was cold! And on top, the area was known to have mountain lions! Luckily we only did that twice and there were more test site and we had the choice.

 

[yungman]

Yeah, this is what most people seem to do when they come for test, but it also seems to be what they prefer to do!

After I completed my University Research Fellowship in EMC simulation, I went on to try to develop it as a service offering (with a sister firm of one of the partners funding the project). What I found was that no-one seems to wants to pay any 3rd parties for support for 'front-loaded' EMC mitigation work - I guess too many folks are too proud to admit they don't know what they are doing, and figure in any case they just need to slap something into their kit if it fails... I gave up trying to sell EMC simulation and modelling support.

I'd suggest you can generally tell how experienced an EMC engineer is by how much they claim they don't know about EMC - the more experienced they are, the more they realize how little they know of the subject!

I actually got very interested in EMC. A lot of people find it boring. I remember in the early 2000s, my company hired a big shot in EMC to give all of us seminar on EMC, I was so into it and I could see the others all falling asleep. I even studied a book on EMC, how to look for ground image current and make sure there is no disruption. I was actually contracted to work on EMC layout problem for KLA Tencor.

As the frequency of operation getting faster, it is getting more and more critical, a few inches can change from a good ground to an open circuit.