0
\$\begingroup\$

I'm currently designing a new metal detector product and took it for pre-compliance as I've finished the hardware design and am currently finishing up on the software. I've never developed a product before (previously I've only contributed to prototype designs) so I'm not 100% clear on how to fault find in case of EMC failure. Specifically the test that my product failed was on:

Radiated Field Immunity (EN 61326-1:2013 / EN 61000-4-3:2006).

My product is in a stainless steel enclosure with a diagonal 4.3" (approx 110 mm) rectangular cut out for an LCD display, two 22mm hole cut outs for buttons. Coming out from the bottom of the enclosure is 2x 5m coax cables to connect a transmit and receive coil and 1.8m mains power cable.

Initially during this test, the device failed at all frequencies betwen 470MHz to 520MHz. I then changed the 1.8m mains power cable to a same length shielded cable and also connected this shielded cable to the enclosure. This reduced the number of failures to just specifically those two frequencies (rather than the whole range between). Knowing that the device fails at these two particular frequencies, and knowing that adding a shielded mains power cable greatly improved the pass rate, what information do I have here in order to investigate further? I feel like adding more shielding is a band-aid approach and would like to remove the shielded mains cable completely if I can determine and fix the root cause. Any help or suggestions would be great and I can supply further information if needed.

The page report:

enter image description here

\$\endgroup\$
1
  • \$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. \$\endgroup\$
    – Null
    Jan 3, 2023 at 15:57

2 Answers 2

3
\$\begingroup\$

First of all, note that some industrial applications might actually require 100V/m susceptibility (such as land-moving machinery, stone crushers, drilling rigs etc).

I also find it weird that general EMC standards are used on what's clearly a RF product. You should probably be using some EN/ETSI ones to test the intentional radiation as well as the unintentional. How about:
ETSI EN 303 454 "Short Range Devices (SRD); Metal and object detection sensors in the frequency range 1 kHz to 148,5 kHz;". Sounds good to me, though your test house should know these things.

So your pre-compliance test is using relatively very low requirements with 10V/m susceptibility and in a reasonably small frequency range. So if you fail the EMC test even with such lax requirements, I would suspect that you have some severe problems in your product beyond shielding/encapsulation.

PCB layout, power electronics layout and grounding would all be the usual culprits to investigate. In case of switching buck regulators, investigate if the frequency might have something in common with the problematic frequencies. Or just provide more details about the supply electronics in a separate question and some power guru could do a design review (layouts have to be done very carefully).

Though my general experience of EMC testing is to not stare blindly at which specific frequencies you were having problems, since it's fairly complex to translate those (and harmonics) to something corresponding to your product.

Also if you have problems with susceptibility you might have problems with emissions too. Particularly in case of RF products as is the case here.

Regarding the encapsulation:

My product is in a stainless steel enclosure with a diagonal 4.3" (approx 110 mm) rectangular cut out for an LCD display, two 22mm hole cut outs for buttons

You will need a conductive EMC gasket around the opening and around the enclosure lid as well. The LCD would ideally be covered in some screening material, since LCDs are quite often the culprit in failing EMC.

Connectors should ideally be metallic/conducting or otherwise you may need to add additional filtering. For tougher EMC requirements you'd typically have "pi" filters on all I/O signals.

Connect your power ground to chassis near where it enters the enclosure.

Shielded cables are "band aid" indeed, as are ferrites on them etc. You should address the root cause, noting that all cables may act as antennas.

\$\endgroup\$
5
  • \$\begingroup\$ As i mentioned in my question, I've no experience in EMC so have no choice but to follow the suggestions of the test house. I can suggest to them your recommendation for EN/ETSI 303 454. Regarding encapsulation, you basically suggest to screen the system as much as possible and to reduce entry paths for noise ingress. Is this a normal approach to solving EMC immunity? What if my system didn't have the luxury of being contained in a steel box? \$\endgroup\$
    – ChrisD91
    Jan 3, 2023 at 13:12
  • \$\begingroup\$ @ChrisD91 Note that I have absolutely no idea about applicable standards for metal detectors, that one was just something I found after brief poking around (it is not harmonized under RED). There's also "EN ETSI 300 330 for SRD in range 9kHz to 25MHz and inductive loop systems-..." which is harmonised and sounds applicable. The key here is that some sort of standard saying "SRD" (short range device) should be used. Your product isn't sorting under the EMC directive but under the RED directive. \$\endgroup\$
    – Lundin
    Jan 3, 2023 at 13:18
  • \$\begingroup\$ @ChrisD91 The encapsulation suggestions were just made since you already have one. And it shouldn't be needed for just 10V/m if your product is well-designed. However the RF paths in particular might be sensitive and maybe there's no other way around shielding at least those. Hard to tell since you didn't include any details. \$\endgroup\$
    – Lundin
    Jan 3, 2023 at 13:20
  • \$\begingroup\$ (Also disclaimer: I'm not an actual EMC expert, I'm actually not an EE but a computer engineer/software dude. But I've been project manager during some 10-20 EMC tested projects including real nasty MIL-STD, automotive, industrial etc. Meaning that I end up being the one who have to study all the bureaucracy and standards since nobody else wants to touch that boring stuff :) ) \$\endgroup\$
    – Lundin
    Jan 3, 2023 at 13:24
  • \$\begingroup\$ OK thanks for the suggestions! It's much appreciated. \$\endgroup\$
    – ChrisD91
    Jan 3, 2023 at 13:36
1
\$\begingroup\$

Sounds like your cable is acting as a receiving antenna for field energy that ends up being conducted into your product as common mode noise. That's a common scenario for RF immunity failures in the hundreds of MHz range as these frequencies have wave lengths in the meter range. That's also a typical length for cables, thus acting as antennas at these frequencies.

Typical fixes often times include common mode filtering of all signals entering your device. That might include ferrites around external and/or internal cables and/or common mode filtering directly on your PCB.

Sometimes it's sufficient to apply common mode filtering to the inputs of high gain amplifiers only, for example. Because high gain amplifiers might have a tendency to rectify and amplify common mode RF noise.

Observing the 1 kHz AM frequency and/or a DC offset at any point of your signal chain is very likely caused by such rectifying effects.

\$\endgroup\$
3
  • \$\begingroup\$ Sometimes it's sufficient to apply common mode filtering to the inputs of high gain amplifiers only Filtering common mode noise on the board is rather strange IMO. To what reference potential would you then filter it ?? IMO Common-mode filters should filter to the chassis if avaiable. That helps it stay entirely in common-mode and doesn't get rectified by various units to diff-mode which then sets up the functionality. \$\endgroup\$
    – tobalt
    Jan 3, 2023 at 12:45
  • \$\begingroup\$ 10V/m is the industrial level of the 61000-4-3 standard. My experience of EMC testing has shown me that cables cause problems by allowing common mode current to enter the system. \$\endgroup\$
    – Vincent
    Jan 3, 2023 at 14:02
  • \$\begingroup\$ A satellite system (first piece) we built failed its EMC testing - radiated emissions I believe. Root cause was leaking at one of the connectors. So to what @Vincent said - cables & connectors tend to be weak points. \$\endgroup\$
    – SteveSh
    Jan 3, 2023 at 15:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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