Radiated emission measurements


I have a problem with radiated emissions at 100+ MHz. I don’t want to go to the test site without checking it in my own lab. So I am using near field probes to pick up the noise

The TEM cell does not show anything, but I am guessing that’s because the problem is magnetic fields and the TEM cell works best with electric fields

I need to do measurements in the far field to be able to correlate to what will be measured in the 3m Semi chamber.

I have been using a log periodic antenna with external wide band amplifier into the spectrum analyzer, but I am not getting consistent measurements (not finding any peaks that are clear)

My best solution right now, is a near field probe with a precise direction and distance to the PCB between measurements… But again near field does not translate directly to far field

I am considering building a small shield chamber and getting amplifier with more gain

But before I do that, any recommendations?



Why do you think so?

Measurement done at 10m room at acredited test lab. Looks like coupling from primary side of transformer to secondary. I could add shielding layer, but a Bead the right place is a lot cheaper

Near field probes are very limited in usefulness in my experience and will often send you chasing after stuff that doesn’t even show in the far field. If you have a common-mode issue, near-field probes often show the noise on every metal surface of the EUT, which doesn’t help much.

A high-bandwidth current probe can be useful if cables are the culprit, but so is a good collection of snap-on ferrites.

A home-built shielded room is also probably a waste of time and effort, since reflections and standing waves may make those measurements useless. A ground plane in your OATS might help with measurement stability. As long as there are no ambients stomping directly on your frequencies of interest you can live with them, you just have to turn the EUT off and on a lot :slight_smile:

Meaningful quantitative measurements in a “home” OATS is an elusive dream. But since you have a baseline from the 10m lab, all you have to do is find the emissions in your own setup, mess with your equipment settings until it’s clearly visible and reasonably repeatable, then set about finding ways to reduce it by more than it was over-limit at the lab. Don’t forget to change polarization and azimuth often during your experiments; more than once I’ve thought I solved a problem only to find that it was still there on the other polarization or a different azimuth.


Must be a pretty big log-periodic antenna to get down to 100MHz :open_mouth: I use a bicon below 300MHz, or a combination bicon-log periodic for wideband.

Is this broadband noise (switchers etc) or narrowband from stable clocks? Mitigation strategy can be very different.

I find it useful to work much closer to the antenna for quick troubleshooting. The levels increase about 10 dB going from 10 m to 3 m and another 10 dB going from 3m to 1 m. Having levels 20 dB higher requires less preamplification. I generally go back to the correct measurement distance after resolving the issue to verify the readings. Also, if you are close, you can move the UUT right up to the antenna to prove the signal is coming from the UUT.

I keep the measured sample as a golden reference and troubleshoot on a different unit. That makes AB comparisons easy when the test setup changes.

Calibrating antennas and an OAT facility are very interesting science projects that take a lot of work. It took me about 2 years to learn all the tricks and get reliable and repeatable measurements. The company wanted an in-house facility.

1 Like

Absolutely, I never do more than 3m in house, and often closer if I’m still in the far field.

Unless the EUT is a $100,000, 800lb machine like the one I was working on today, lol.

Indeed a lot of work, and respect to you for achieving it! EMC is a sideline for me so I’ve never been able to justify the effort, and I’ve been able to deliver results for clients without it. Mind you, that was a lot easier when there was a chamber 5 minutes from my shop, but that’s sadly no longer the case.

Yes. I have one of those cheap ones from Aliexpress, but for lower frequency I had to spend real money, so have one that is good from about 150MHz (but no calibration certificate)

I am using a 250MHz clamp on current shunt that is designed for conducted cable measurement and those measurements can be translated with some uncertainty to radiated emission.

I was looking at a large peak from the 3m measurement at 30MHz, but then compared that to the conducted emission at 30MHz which was very low. That didn’t compute. So dug some more. The conducted measurement is done with LISN, while radiated seems to be conducted without LISN. Sometimes they use CMAD to remove line impedance resonance

So I did a resonance measurement of the cable with the DUT, not powered. And lo and behold, the cable shows a resonance peak at the frequencies I am working on. So next step is to reduce the source or the medium

I’ve seen this situation. Had an EUT a couple of months ago that passed conducted emissions but failed RE at 37MHz. It passed RE when we used a 10-foot line cord but failed with a 2m line cord. Common-mode noise was radiating on the line cord: 2m is 1/4 wavelength at 37MHz. They are now shipping the product with a giant ferrite on the line cord :frowning:

Be careful with extrapolating RE from cable C-M current measurements. I had a client years ago who did this. I asked about a particular 2-conductor unshielded cable, and he was adamant that it couldn’t be the problem - “I measured the C-M current on that cable and according to Henry Ott’s formula it cannot account for the radiated emissions”. He kept moving to more and more expensive connectors for a multi-conductor shielded cable that he was convinced was the source of the problem. I added a 10-cent ferrite to the unshielded cable and the problem went away.

I am using EMCview SW with a spectrum analyzer to get good measurements, but are checking in radiated room once per week to calibrate progress, since you are right, Conducted clamp measurements are not the same as radiated measurements

That is actually the same I am seeing. To understand the problem, I did the measurement of the cable. Even though it took a couple of hours, I know that to next time :wink:

It is a 100kHz switcher with spread spectrum. After this there’'s the typical load etc
Primary problem is ringing of the leakage inductance with the transformer parasitic capacitance. So I am making a new transformer with a shield to combat that. That is costly, so are trying out other remedies also

I think you wrote in a different thread that you did radiated measurements in the garage. Can you share a little more about the setup and which antenna you are using?

The radiated measurements in the garage were for antenna gain at 2.4 GHz. Those were performed with 18" spacing between antennas. I have set up many crude facilities in small labs, though. I will tailor the comments to 30 - 300 MHz emissions.

I use a standard Biconnical antenna like this SAS-540 Rigid Biconical Antenna (ahsystems.com). They do make a folding version as well. AH systems has usually been the most economical choice for new antennas for all my work. Biconnical antennas are pretty rugged. Since they are mechanical items, they will maintain their calibration unless they are physically damaged or burned with high power from susceptibility tests. If you find one on E-Bay it is should be ok unless it is burned out. They do need a preamp with most spectrum analyzers. The antenna is a little bulky. There are smaller antennas with built-in preamps. Tekbox makes some Amazon.com: TekBox TBMA1 30MHz 1000 MHz Biconical Measurement Antenna: Electronics. I do not have any experience with smaller antennas. I am always suspicious of preamps as they can get overloaded. So, I like the antenna separate from the antenna so I can remove it. If you want to build the biconnical, the basic information I contained in MIL-STD461. Julia has shown pictures of her antenna. It is a higher power Biconnical.

Most times, I do not use a ground plane. It has had minimal effect on the devices I measure. It is not necessary if you are just doing comparative measurements.

For the setup, I clear out the largest area available Setup the antenna, and measure. If I am testing against absolute readings, I scale the readings for the distance. I have used metal and wood tripods successfully. If you go for at least a 10 dB margin on the emissions level, the shortcuts work pretty well. The biggest issue I have seen is understanding and accurately tracking the correction factors for antennas, cables, preamps, and measurement distances. Measuring in the near field of an antenna will probably not scale well, but it does work for comparative measurements. For the near field, make sure the positioning of everything is identical between measurements to ensure a measure of repeatability. Also, do sweeps with the UUT turned off to know what the ambient signals are. Steel buildings seem to provide a measure of shielding as opposed to a wood house without metal siding.

Hopefully, this helps. I am happy to answer specific questions if you have any.