1
\$\begingroup\$

I have an electronic system design with lots of off-board data/power connections to various peripheral components (sensors, LEDs, motors). This is all contained within a large enclosure that is intended to be the "primary" EMC/EMI shielding.

Since no enclosure is perfect, to help with EMC and EMI adherence we are looking to use shielded cables for these cable runs inside the machine. They are carrying (in various combinations) differential I2C and SPI, DC currents of a few amps and some miscellaneous digital signals (enables , chip selects etc). There are also 4 wire stepper motor cables.

However, I'm unsure about terminating these cables onto the motherboard PCB. Using full 360° shielded connectors is both cost and space prohibitive.

Under what conditions is it acceptable to use shielded cable with non-shielded connectors (for example standard 0.1" headers), terminating the shield drain wire through one of the header pins? How badly will this affect the EMI/EMC performance of the cable/connector combination?

I realise the answer to all EMC issues is "it depends", but I'd really appreciate some guidance on this. It's really not my area of expertise.

The machine is powered from an external 12V DC power brick. It will be in an office/laboratory environment (non medical, non life critical etc.)

\$\endgroup\$
6
  • 3
    \$\begingroup\$ You should probably connect the shield/screen to chassis earth. You should also use differential receivers and balanced transmission of signals and data. Other than that, there is nowhere enough details in your question to give any more help. \$\endgroup\$
    – Andy aka
    Commented Nov 13, 2021 at 11:04
  • \$\begingroup\$ Using a shielded cable without proper termination of the shield is just a feel good exercise. Under this condition the shield can actually act as an antenna. \$\endgroup\$
    – SteveSh
    Commented Nov 13, 2021 at 11:15
  • \$\begingroup\$ And using a pigtail to terminate a shield is another thing you don't want to do. A pigtail can significantly reduce the shielding effectiveness (dBs of shielding) of the shield. \$\endgroup\$
    – SteveSh
    Commented Nov 13, 2021 at 11:17
  • \$\begingroup\$ I added some extra context about the signals being carried. \$\endgroup\$
    – jfowkes
    Commented Nov 13, 2021 at 11:29
  • \$\begingroup\$ A good old ribbon cable with one GND wire between each signal wire is a pretty good compromise. \$\endgroup\$
    – bobflux
    Commented Nov 13, 2021 at 12:24

2 Answers 2

1
\$\begingroup\$

You have not provided any details on the target requirements and very few on the actual design of "the machine". From what I see so far, it looks more like you are afraid of shadows than trying to solve any real problem.

Let's start with basic. The FCC part 15 in US and IEC 61000 / CISPR 32 standards in Europe are most common documents that govern EMI/EMC. So, before you start asking questions you should decide on what class of device you are making and what kind of standards it should conform to.

Now, the standards define two kinds of EMI, radiated and conducted.

The conducted EMI is usually dealt with by filtering on the power supply and data lines and have nothing to do with shielding.

To suppress radiated EMI you should first identify potential sources. Since you did not mention any high frequency radiating components your biggest concern would be those stepping motors, their power wires and switching circuitry. Note, that they also contribute heavily to conducted EMI. It is rare to see shielded motor cables. The usual approach is to add ferrite rings and lots of bypassing capacitors where necessary.

Another typical source that you did not mention is power supply. These are heavy polluters and unless you are designing your own, it pays to find properly certified part from reliable supplier.

Comparing to the motors and power supply, the digital signals are usually insignificant and easy to deal with using proper enclosure. The noisiest are typically oscillators for MCUs and DC-DC converters. This is where you have to focus your EMI-reducing efforts.

As for data lines, it is important to understand that shielding of digital connections is not meant to protect environment from the switching noise, it is to protect data from the environment. There is no reason whatsoever to shield these inside the enclosure. If you peek into any modern certified consumer device you will rarely find any shielded connections. Most likely it will be antenna coax or mic-level audio. For any external shielded connections (e.g. USB or long runs of load cell cables) the typical approach is to place sockets on the PCB and then mount PCB right at the enclosure openings. In some cases small boards are used with pre-amplifiers or transceivers that convert signals into something less susceptible to the noise that can be further routed to main PCB with unshielded wires.

For the internal connections, if you have multiple data lines combined in the same cable, the best approach is to use FFC/FPC with standard non-shielded connectors while making every other wire a ground conduit. For the differential signals a simple twisted pair should be sufficient. Again, this is to protect from interference within the enclosure, not for EMI reduction.

Finally, if your "the machine" has mains power supply, then proper grounding will go a long way in reducing radiated EMI.

\$\endgroup\$
5
  • \$\begingroup\$ Yeah ok "afraid of shadows" may.be a fair assessment but also I'd rather be paranoid and ask questions now than pay later with a failing design. Your answer is very informative and clarifies a lot of what I had gathered from other sources, thanks! \$\endgroup\$
    – jfowkes
    Commented Nov 13, 2021 at 20:01
  • \$\begingroup\$ Oh and I added some extra context about power supply and usage. \$\endgroup\$
    – jfowkes
    Commented Nov 13, 2021 at 20:10
  • \$\begingroup\$ External power supply should remove a lot of worries. Although be careful where you get them, I recall seeing an in depth power supply tear-down article somewhere with pictures of PCBs and DSO screenshots straight out of horror show. \$\endgroup\$
    – Maple
    Commented Nov 13, 2021 at 20:26
  • \$\begingroup\$ Oh yeah, we're being very cautious about all this, we'll be sure to source something that's good quality. Part of the reason for the question is trying to eliminate as many unknowns and risks as we can. \$\endgroup\$
    – jfowkes
    Commented Nov 13, 2021 at 20:42
  • \$\begingroup\$ Also, if you target laboratory environment you should also verify their own requirements, as many labs have specific needs. For example, the grounding may be mandatory, and your power brick most likely will not support this. See an interesting discussion here: electronics.stackexchange.com/q/207139/187920, including EMI considerations \$\endgroup\$
    – Maple
    Commented Nov 13, 2021 at 20:59
0
\$\begingroup\$

If you use a shielded cable to pass through the wall of a conductive enclosure, and EMI is an issue, then you must connect the cable shield to the enclosure wall.

Ideally you would do this while maintaining the full 360 degrees conductive shield around the cable. Typically this is done with a shielded connector, this is why shielded connectors are made.

If you have to compromise on the size and shape of the shield connection, perhaps gathering the shield wires together into a single bunch and connecting it at one point, then the more you compromise, the larger will be the antenna effect of the cable and shield.

Measurement on a real system will tell you whether you have compromised too much.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ The cables in this case do not pass through any walls, it's just connections between PCBs inside an enclosure. \$\endgroup\$
    – jfowkes
    Commented Nov 13, 2021 at 12:10

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.