1
\$\begingroup\$

I see many people are concerned about EMI issues, particularly noise due to this phenomenon. Many SMPSs are enclosed in boxes which looks like Faraday cages: enter image description here

But how is EMI identified or investigated or measured practically? What characteristics of the measurements let one to conclude that the measured data belongs to EMI? Is this a difficult practice only done by RF engineers or are there also easier ways/methods/circuitry?

edit: By the way here:Switching power supply using faraday cage and EMI protection? it is claimed that that the meshed SMPS case in the above picture has nothing to do with the use of Faraday cage and will not block EMI.

\$\endgroup\$
  • 2
    \$\begingroup\$ Have you done any research on this at all? \$\endgroup\$ – Andy aka Nov 22 '16 at 10:14
  • \$\begingroup\$ @Andyaka i was looking for something easy to monitor, and i wanted to have some initial overview not to go in wrong directions. many papers on this subject requires huge expertise and knowledge in the field. however i just encountered this for instance:saelig.com/MFR00154/M00154003.htm which made me to think maybe a loop antenna and an RF amplifier i could build a simple one. the tools are very expensive. \$\endgroup\$ – user16307 Nov 22 '16 at 10:52
  • \$\begingroup\$ With a calibrated antenna \$\endgroup\$ – Voltage Spike Nov 22 '16 at 17:56
3
\$\begingroup\$

What characteristics of the measurements let one to conclude that the measured data belongs to EMI?

It does not work like that. It is not that the product is measured, some result comes out and that that is called "EMI".

As any measurement you start with a specification and/or requirement. For example you want the product to meet ITU EMI requirements. We then need to know what these requirements are so we get the documents concerned and perform the measurements described there.

One of these measurements can be to use a broadband antenna connected to a spectrum analyzer and measure (in a shielded chamber) what signals (EMI) is generated by the product under test (while it is operating of course).

In the requirements it is stated that in a certain frequency band those emissions cannot exceed a certain fieldstrength. So you measure what they are (at a specified distance) and determine if that meets the requirement or not.

If the level of these emissions are not low enough then something must be done. The design must be changed or you could simply choose not to fulfill the requirements but that might mean your product cannot be sold in certain regions.

Of course some knowledge about RF and radio waves is required but I would not say that this area is limited to RF experts only. With some training most engineers could do this. Improving the product's design for lower EMI emissions might require more knowledge and experience but that is still in the realm of non-RF engineers. There are no transmission lines to be power matched and resonators to be tuned so in my opinion this is not comparable to RF design.

\$\endgroup\$
  • \$\begingroup\$ But imagine in a room there are many instruments. And then you have a particular device in the same room which you suspect of EMI noise is effecting it but you want to be sure. Then you have the data of that noise. Now it comes to verify if there is such EMI on the air. So how can one measure the magnitude/existence of that possible EMI? I gave this example to clarify my question. For example I don't have a broadband antenna or a spectrum analyzer but a scope. \$\endgroup\$ – user16307 Nov 22 '16 at 8:37
  • \$\begingroup\$ You cannot reliably measure the magnitude of EMI in the air without an antenna and a spectrum analyzer. As scope is usually not sensitive enough to detect EMI emissions unless they are very strong and low frequency (which they are usually not). To identify the suspect device you have to use trial and error, switch it off (assuming this does not affect the measurement) or shield it, change the position of the suspect device. \$\endgroup\$ – Bimpelrekkie Nov 22 '16 at 8:49
  • \$\begingroup\$ Oh I was thinking making my own near field probe something like this: emcesd.com/sloop7.jpg And then place the probe next to a device and then amplify it with an RF amplifier and data log it. I have no idea where to start with. What do you think? \$\endgroup\$ – user16307 Nov 22 '16 at 8:51
  • \$\begingroup\$ For my purpose it would be enough to see if it is increasing or decreasing. I mean the amplification doesn't have to be clean. \$\endgroup\$ – user16307 Nov 22 '16 at 8:54
  • \$\begingroup\$ Yes, you can use such a loop but the signal which can be picked up by such a loop is extremely small. I have build and used such loop antenna myself in the past but I used them with a spectrum analyzer and I used them to identify EMI sources on a PCB, so right above the components. Most equipment is shielded so the levels of EMI outside the instrument is much-much lower than the levels you would see at the PCB. I doubt if you will be able to detect anything with such a loop. Even with a good spectrum analyzer, let alone with a scope + amplifier. \$\endgroup\$ – Bimpelrekkie Nov 22 '16 at 9:00
2
\$\begingroup\$

EMI is measured in special rooms called "RF anechoic chambers" There are strict regulations regarding unwanted electromagnetic emissions, and all commercially-produced electronic devices must pass emission certification, to meet FCC and other international standards. This is true for all developed countries.

The anechoic chambers serves two porposes, (1) they shield the test environment from outside emissions from broadcasting stations, TV and radio, and (2) provide clean space with no reflections (no echo), to determine real emissions from a tested device, emission spectrum, amplitude, direction, and polarization. These are pretty expensive facilities equipped with carefully calibrated antennas, rotating test tables, and highest-quality measuring instruments. Every serious electronic corporation has these rooms, and there are certified laboratories who rent these rooms (with equipment and RF engineers) on hourly/daily basis, for a hefty price.

However, before taking the actual EMI test, design engineers usually conduct preliminary testings of their products with a set of near-field probes, aka "sniffers", connected to less-expensive spectrum analyzers in an ordinary lab. These small antenna-sniffers allow to locate a component or area that is the main source of EMI, and do something about the source, bypassing, filtering, smoothing signal edges, shielding, etc. Alternatively, the sniffers are employed only when a final product has some design flaw (board level, enclosure, etc.) and failed final certification test at a customer's site.

Theoretically, all EMI emissions could be calculated with modern 3-D simulation technology, but modeling all geometrical complexity is really resource consuming. RF engineers are some special people who got to know details of antenna theory and developed a sense of which electro-magnetical elements will emit what and how, without solving Maxwell Equations. This is still an art of black magic, even more than the art of high-speed design.

\$\endgroup\$
0
\$\begingroup\$

But how is EMI identified or investigated or measured practically?

  • With a set of typical tools, an RF Eng can identify easily areas of compliance risk for egress and ingress for both conducted and radiated

What characteristics of the measurements let one to conclude that the measured data belongs to EMI?

  • EMI has many signatures which are easy to recognize with experience
    • without experience you cannot tell if you are watching Bugs Bunny on the cartoon TV channel or someone jamming your lab . lol.
    • since signal gain in dBi is proportional to frequency on loop size of antenna, official sites use a calibrated biconical dodecahedral antenna at 10m or so in a calibrated quiet test site.
    • detection is trivial, while calibration is not.

Is this a difficult practice only done by RF engineers

  • The ease is inversely proportional to your decades of experience

or are there also easier ways/methods.

  • there are many ways to test for ; ingress such as a xenon strobe nearby with typical AC and DC cables and egress with suitable sniffer probes or scope probe loop for near-field rough detection.
  • calibration is hard, unless you have a network analyzer with calibrated antennae for each band
  • round hole slotted shields are excellent for microwave but the edge slots can make excellent antenna for SMPS impulse currents at 1/4 wavelengths causing egress failures, amplified by external cables
  • comparison with known "suitable" passed and failed units with test reports , makes it easier.
  • there are 2 categories: susceptibility or ingress, and unintended radiation or egress.
    • This applies to BOTH load and line induced transients
    • and applies to BOTH **CONDUCTED and RADIATED*
    • applies to BOTH magnetic <=30 MHz and electric fields>30MHz
    • and applies to both Quasi-Peak and Peak
    • and applies to Continuous Wave, CW and modulated wave and Impulse
    • it also applies to 10-15kV ESD tests (conducted and plane wave radiated) ,
    • also ~3~6 kV dc HIPOT 1 second WITH secondary grounded (worst case)
    • so there are many combinations of above to test !!

It can be as easy as looking for an FM signal or TV signal or as hard as solving "where is that smell coming from? "

No two experiences are the same.

  • I have done this process on about a dozen certifiable products in both Class B and A shield rooms, Lingren(tm) cages, which I have bought, are essential for ingress tests, and RF quiet field test sites for proper measurements or use digital Network Analyzer for uncalibrated reference tests. I have done many more times in-house and like me, there are many consultants available who are willing to fly and test

  • Tempest levels apply to Mil-Spec requirements.

  • Shield Rooms are not necessarily the best way to do it, because it will cause standing waves at 1/4 wave wall reflections and PSU connections to DC gnd and cable step load noise affects "everything"
  • conducted AC gnd leakage tests are easy to do, HIPOT is also easy to do and also easy to do the wrong way!

  • Anechoic Chamber is only for Microwave band as absorption is ineffective in lower RF band

  • Magnetic Interference tests to 30MHz are done with current probes.

  • Would you want a newbie analyzing your cancer risks?

    • or a "traffic ticket" lawyer to handle your divorce??

Final thoughts

Nothing beats "good" RF Test, Measure and R&D experience.

  • it is easy to spend $20k+ on all tests and fail then have to do it all over again.
  • The penalties for not doing it, can cost you your business
\$\endgroup\$
  • \$\begingroup\$ "Anechoic is only for Microwave." Yes, but the gear needed for sub-microwave is usually the same gear that does microwave so either way you always end up in an anechoic chamber :). \$\endgroup\$ – Wossname Nov 24 '16 at 9:52
  • \$\begingroup\$ not if you want to avoid harmonic standing waves \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Nov 24 '16 at 10:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.