Filters above 100MHz need shielding?

Hi
Page 30 of this
says...

Filters that must provide significant levels of attenuation at frequencies above 100MHz, must employ shielding techniques as well. They will not be able to achieve the required performance otherwise.
Is this true?
Why does this common mode choke purport to be able to provide attenuation above 100MHz….
Wurth 744841414

…it doesn’t have shielding

Hi,

the example shown in Fig. 23 shows the benefit of a shielding using a 3-terminal PI-filter. Also the 3-terminal PI-filter without a shield attenuates high frequencies, but less compared with the one using a shield. As you can see from the provided attenuation vs. frequency curve included in the Wurth 744841414 datasheet its attenuation will decrease also for higher frequencies (depends on configuration).

The conclusion is, both the 3-terminal PI-filter without a shield and the Wurth 744841414 provide attenuation at higher frequencies (>100 MHz). What matters is your required/aimed attenuation.

BR

Note, one (center) terminal of the 3-terminal PI-filter is electrically connected to the shield. The CMM choke you are reffering to, does not provide this option.

BR

Shielding of filters may be necessary in some situation. I smell an error of reasoning when you assume that a shielded filter would solve the experienced EMI problems. You have interfernces that are apparently bypassing the filters and shieldings of your design in various ways. It's unlikely due to a single unshielded common mode filter.

if the filter is too near the noise making devices - shielding is a must.

Hi,

before considering shielding you have to determine, or make a reasonable assumption, which kind of interference you are dealing with, magnetic or electric. Different materieals are used for both kind of interferences to maintain effective shielding. Last paragraph on page 8 in [1] might be of interest for you. Of course the layout itself would be a good starting-point to figure out possible mutual interferences and course of radiation.

[1] https://www.analog.com/media/en/training-seminars/tutorials/MT-095.pdf

BR

Thanks for the document, i read it, but am still unsure of the best way to tackle our radiated EMC issue.
The attached is our setup, and we just need to reduce rdiated emissions between 30MHz - 1GHz.
Please advise if you know of any good radiated EMC documents related to this?
All the ones i have read do not give much incite. They just trot out the theory without saying what to
actually do in practice. Either that or they give some ridiculously expensive solution involving 360 degree total metal connection with gasketting etc etc

Hi,

it seems you are struggling with this issue since quite a while, reading your last threads. In one of those I recommended to determine which of your systems (A or B) causes the emission, reply #9 in [1].

Nevertheless, a book which might be of interest is "Electromagnetic Compatibility Engineering" by Henry W. Ott, here is a google book link [2]. Starting from page 140 with "Distributed Systems".

[1] https://www.edaboard.com/threads/ra...t-frequencies-between-different-smpss.396742/
[2] https://books.google.at/books?id=2-...e&q=ott electromagnetic compatibility&f=false

BR

In one of those I recommended to determine which of your systems (A or B) causes the emission, reply #9 in [1].

Click to expand...

Thanks, The emissions come from both enclosure A and B......slightly more from B than A.

As you can see, box B contains a 10W DCDC Buck, but this enclosure B is open at the front. So when we take it for radiated emissions testing, it fails. Obviously the (screened) cable will likely be responsible for some of this failure, but also the semi-open enclosure we also think acts like an antenna. Anyway, we wonder about some shielding directly over the SMPS?

The top (inductor side) of the 10W DCDC buck faces the enclosure opening, so we wonder if we can actually put a simple shield over the top of the PCB? …as follows….just make a PCB the same size as the 10W Buck PCB (8cm x 3cm), and simply use PCB pillars to mount it above the 10W Buck PCB like a canopy……have this “canopy” PCB simply with a total copper pour over it, and connect this copper pour to the 0V of the 10W Buck PCB.

…What do you say? Surely it must reduce radiated emissions? It cannot posibbly act like an antenna itself since it will be connected to 0V GND of the 10W buck PCB.

The “canopy” idea is like the attached, but far simpler and cheaper.

What do you think?

The system is as in post #6 above

The document doesn't seem to describe how the attenuation is actually measured, so impossible to evaluate it.

Can you post any of your Data

treez said:

Hi
Page 30 of this

https://interferencetechnology.com/wp-content/uploads/2016/10/2016_IT_EMC_Filters_Guide.pdf

says...

Filters that must provide significant levels of attenuation at frequencies above 100MHz, must employ shielding techniques as well. They will not be able to achieve the required performance otherwise.
Is this true?
Why does this common mode choke purport to be able to provide attenuation above 100MHz….
Wurth 744841414

…it doesn’t have shielding

Click to expand...

That CM choke will attenuate common mode current but to be really effective you have to provide an alternative low impedance path for the common mode current to flow instead of the wiring. You will also get some diffential inductance from that CM choke at 100Mhz but it needs a differntial capacitance to be effective. The type of capacitor you use will be critical to optimal filtering for 100Mhz you would need a decent ceramic or Film Capacitor to provide the decoupling as a electrolytic will be of no use at that frequency.

Hi,
We have got to pass radiated EMC (30MHz to 1GHz), but in a “real” situation where we cannot afford “proper” RF EMC techniques…and in truth…”real” EMC techniques just wouldn’t be mandated, or realistic.
Our setup is as attached.
The following doc calls for a “RF Reference plane” for use with EMC filtering and shielding….


…That sinks us from the start. We have two panels, whose metal enclosure is earthed, but that earth comes from the earth wire in the long 3-core power cord…..so that connection is so high Z to RF that its NOT a connection at all. So we have no RF reference plane!!!…and our metal chassis is basically floating at RF !!!

So straight away no official EMC literature is of any help to us!!!

Does anyone have an “EMC guide for real-life, cost constrained situations” document, instead of these “perfect EMC” guides which tell how to do EMC for super-expensive medical devices or military devices with RF gasketted connectors etc? I’ve read tons of them and they are no use.

I mean, Feedthrough capacitors, etc etc aren’t much use without an RF reference plane.

Take the blue cable between the two enclosures in the attached.
What are the options?...........

(a)…Two core cable unscreened?
(b)…Three core cable with third connector connected to chassis at each end?
(c)…Two core cable with screen…screen left floating…it may as well float…both our chassis’s are already floating at RF?
(d)…Two core cable with screen connected to chassis at “A”?
(e)…Two core cable with screen connected to chassis at “B”?

…and what if we just don’t bother connecting our chassis’s (A and B) to earth at all?....after all, the power cord earth connection is NOT a connection at RF…its too high Z. And then what should we connect our Y capacitors to?....there is no Earth. How are the common mode chokes working at all?

No EMC literature discusses these real options. Do you know the real answers?

treez said:

Hi,
We have got to pass radiated EMC (30MHz to 1GHz), but in a “real” situation where what we cannot afford “proper” RF EMC techniques…and in truth…”real” EMC techniques just wouldn’t be mandated, or realistic.
Our setup is as attached.
The following doc calls for a “RF Reference plane” for use with EMC filtering and shielding….

https://interferencetechnology.com/wp-content/uploads/2016/10/2016_IT_EMC_Filters_Guide.pdf

…That sinks us from the start. We have two panels, whose metal enclosure is earthed, but that earth comes from the earth wire in the long 3-core power cord…..so that connection is so high Z to RF that its NOT a connection at all. So we have no RF reference plane!!!…and our metal chassis is basically floating at RF !!!

So straight away no official EMC literature is of any help to us!!!

Does anyone have an “EMC guide for real-life, cost constrained situations” document, instead of these “perfect EMC” guides which tell how to do EMC for super-expensive medical devices or military devices with RF gasketted connectors etc? I’ve read tons of them and they are no use.

I mean, Feedthrough capacitors, etc etc aren’t much use without an RF reference plane.

Take the blue cable between the two enclosures in the attached.
What are the options?...........

(a)…Two core cable unscreened?
(b)…Three core cable with third connector connected to chassis at each end?
(c)…Two core cable with screen…screen left floating…it may as well float…both our chassis’s are already floating at RF?
(d)…Two core cable with screen connected to chassis at “A”?
(e)…Two core cable with screen connected to chassis at “B”?

…and what if we just don’t bother connecting our chassis’s (A and B) to earth at all?....after all, the power cord earth connection is NOT a connection at RF…its too high Z. And then what should we connect our Y capacitors to?....there is no Earth. How are the common mode chokes working at all?

No EMC literature discusses these real options. Do you know the real answers?

Click to expand...

What do you mean by EMI for Real Life? As far as Passing EMI is concerned you have to design your equipment to meet your appropriate standard in the manner dictated by the standard. As far as you are concerned that is your real life.

Its impossible for anyone to help you as your not providing the data to be able too. Everything you say at present suggests you are not fully understanding your problem so are just adding components and hoping it will do something.

My advice would be to forget screened cables etc for the moment as thats going to add further cost and probably still not address your problem. An ungrounded screened cable will still radiate.

Your best bet for the moment would be to split the system into 2 and deal with Box 1 first using a resistive load at the load. Once you do this, Measure the conducted emissions on each of the input power and earth wires seperately and also measure as a wire bundle. Do the same for the secondary side also. This will tell you an awful lot as to your emissions and the type of emissions in the wiring caused by Box 1.

Box 1 would need to be screwed down onto the ground plane using a low impedance bond for EMI which would probably be allowed by your standard anyway. With the metal box at ground potential then the only thing left to radiated would be the cables or gaps in the box allowing rf to escape.

With the conducted data you can address the conducted emissions first. Which with knowledge of the emissions and your current circuit you can design the filter (if necessary with our support) then move to Box 2.

Your best bet for the moment would be to split the system into 2 and deal with Box 1 first using a resistive load at the load.

Click to expand...

Thanks, we did this, we found that there is signficant emissions coming out of Box A.
We also disconnected A from the mains, and put a 12V battery in A, and used that to power B....it gave slightly more emissions than the "resistive load" case above.
So we know that there are significant emissions coming out of Box A and Box B.....A bit more more from B than A

My advice would be to forget screened cables etc for the moment as thats going to add further cost and probably still not address your problem.

Click to expand...

Having screened cable (screen connected to chassis A and B) significantly improved emissions, when compare to non-screened 2 core cable being used between A and B. Academic EMC theory would tell us that this would not be the case, but it was...showing how inapplicable is modern academic EMC theory to this case.

An ungrounded screened cable will still radiate.

Click to expand...

Thanks...everything in our system is going to radiate, because there is no "RF reference plane", and having one would be impractically costly. Nothing is earthed, due to the unavoidably high Z of the earth cable in the mains cable to the product. We have no "ground"...and no EMC tutorials address our case.

We are seeking a "real" document, which confess's that our situation is pretty much irrelevant to modern academic EMC theory, and tells us how to tweak or bodge it so we pass the radiated emissions test anyway.

With our situation, the only way to calculate the solution would be to have a 3D model of the EM field over the whole system, and then calculate the solution...but that would be ridiculously expensive and impractical.

Box 1 would need to be screwed down onto the ground plane using a low impedance bond for EMI

Click to expand...

I agree that modern academic EMC theory calls for that...but it is impractically expensive. We have no "ground plane"...having one would be impractically expensive.

With the conducted data you can address the conducted emissions first.

Click to expand...

Box A on its own has passed conducted emissions. We are now doing radiated emissions.

treez said:

Thanks, we did this, we found that there is signficant emissions coming out of Box A.
We also disconnected A from the mains, and put a 12V battery in A, and used that to power B....it gave slightly more emissions than the "resistive load" case above.

Click to expand...

Thats a good start, Stay on Box A with a resistive load for a while. Don't jump around doing what if scenario's, address box A emissions first for both the power lines and the output side. You will need to address both prior to moving to looking at box B.

Now you know where some of your problems are coming from you need to determine if the noise is Common Mode or Differential Mode using the tests i gave you in my previous message. This will help to determine the type of filter you need and the amount of filtering you need for Box A.
.

hats a good start, Stay on Box A with a resistive load for a while. Don't jump around doing what if scenario's, address box A emissions first for both the power lines and the output side. You will need to address both prior to moving to looking at box B.

Click to expand...

Thanks, Box A is an established unit that passes conducted and radiated when used on its own (with no cable coming out of it as we are doing now).
We now want to have Box B connected to it as shown, but this is causing the radiated emissions problems to rear its ugly head.

We simply cannot make significant changes to Box A...its an established unit used in many products. About the most we could do to A is add a common mode choke to the cable as it leaves A to go to B.

Box B is the new bit, and this is where we need to concentrate our efforts because its too expensive to go making too many changes to box A.

treez said:

Thanks, Box A is an established unit that passes conducted and radiated when used on its own (with no cable coming out of it as we are doing now).
We now want to have Box B connected to it as shown, but this is causing the radiated emissions problems to rear its ugly head.

We simply cannot make significant changes to Box A...its an established unit used in many products. About the most we could do to A is add a common mode choke to the cable as it leaves A to go to B.

Box B is the new bit, and this is where we need to concentrate our efforts because its too expensive to go making too many changes to box A.

Click to expand...

It sounds more and more like Box A is going to be a majority of the problem. Cables are antennas so if it wasn't tested with cables attached thats major deviation from the original Pass especially on radiated. Without filtering the secondary side outputs of a flyback contain the turns ratio * primary switching current in terms of secondary side emissions which will easily cause major issues.

If you double the length of the inteconnecting cables you'll probably see the radiated emissions get even worse. For sure you'll probably need to filter the output of Box A if it wasn't originally designed to come out of the box. 100Mhz is approx a 3m Wavelength which i bet you used a cable length thats a perfect antenna for it or its harmonics.


In the hope that your noise is differential you could try a small PI filter on the output of Box A (it won't be much bigger than a CM choke) - to get 40dB attention @ 150kHz you'd want about 150uH with a 680nF either side. See if it makes a difference, it should give good attenution right up through the frequency range, add a couple of 470pF Ceramics in parallel to the 680nF facing the output too as these will be better at high frequency atteniation at 100Mhz

Thanks, i thought problems above 10MHz were always Common Mode?
The skin depth at 100MHz for copper is 5um, and so 100MHz diff mode would see very high resistance from the output of the flyback SMPS?

treez said:

Thanks, i thought problems above 10MHz were always Common Mode?

Click to expand...

Not necessarily, it could be one or the other or both. If your PSU are generating the frequencies and there is a path for them to come out on the cables they will just take the path of least impedance, the type of noise depends on how its coupling and where its generated, rectifier diodes, switching mosfets and transformers are the usual suspects. They typically create noise at much lower frequencies but the harmonics nearly always create problems up at higher frequencies. You can kill it at source through damping or filter it.

You normally use a CM Choke and some Y capacitors to form an LC filter to the Chassis Earth to create a common mode filter. The capacitors form a Low impedance path for the current to flow in a controlled manner and the CM Choke provides an high impedance path to prevent the current going out on the cables. Differential Mode is similar but in this case an inductor in the line provides the high impedance to differential current flowing in the wire and the capacitors provide the low impedance path.

Your Box B will probably need a LC Filter on its input as opposed to a PI filter so the element facing the Box 1 will be an inductor, this will prevent Box B acting has a RF load to any potential RF currents getting out of Box 1

Thanks,
As discussed, the attached is our setup going for radiated emissions testing. (It’s a kiddies play panel for hospitals and doctors gaming.)
What do you think to the following edicts on the radiated emissions testing ?.....(30MHz to 1GHz)

A…..Adding an RC snubber from switching node to ground at the Buck SMPS may well have a good effect. However, the sudden discharge of the snubber capacitor simply means there will be another radiated emissions problem, but above 1GHz, where formal testing does not take place?

B…….If we use common mode chokes with Y capacitors to the chassis, then any failure peaks that we may get are just as likely to be coming from resonance in the common mode LC filter, as from genuine noise emissions? As such, tweaking the Y capacitor value to try and bring peaks down is likely to be futile, as it will just mean getting a resonant peak at a different frequency?

C……The cable connecting units A and B could be made of 2 core cable (not twisted) with a metal foil screen…or twisted pair with a foil screen. Can you confirm that 2 parallel conductors covered in metal foil screen would radiate more than twisted pair inside a screen?

D…….If I do a radiated emissions test in a proper radiated EMC chamber…but of a battery supplying a Buck converter in a metal chassis via a cable (ie, a cable from the battery to the chassis containing the buck)…then will the emissions be worse if the chassis is connected to earth ground?……or will emissions be worse if the chassis is not connected to earth ground at all?

E……The buck converter in the attached supplys some radio pagers via a wire loom. (5V, 0.5A to each pager). Suppose we put a common mode choke at the output of the buck, just before the connector that connects to the loom…..is there any circumstance where adding such a common mode choke could make radiated emissions worse?

F….For the blue cable in the attached…..which is best for low emissions?...
a)….2 core cable (unscreened)
b)….twisted pair (unscreened)
c)….3 core cable with 3rd connector connected between the two chassis’s (unscreened)
d)… 3 core cable with 3rd connector connected between the two chassis’s (screened)

G.......Why on the entire web are their No forums, or sub-forums, for people to ask questions about EMC testing of products?....is it because the actions taken to get kit through EMC testing is the deepest secret of any company?

H....For good EMC performance, why is it said that a capacitor with high SRF is needed to be connected right at the input/output connectors? (diff mode)

To answer G, there are forums such as this to ask questions but to get answers you have to provide data (schematics, EMC sweeps, test method, specification etc)so we know what your problem is otherwise its impossible to help you.

To answer H, a capacitor with High SRF has low inductance and as such will remain a capacitor accross a broad spectrum. A highly inductive capacitor with Low SRF will be high impedance to high frequencies and therefore not a capacitor.