Advertisement

Blog

EMC Problems in Modern Electronics

For many types of electronic devices, the Federal Communications Commission just requires Part 15 certification. For devices that fall under the Part 15 rules and regulations section, emissions are checked if there are any clocked circuits that conceivably might radiate RF that could cause problems with other devices. Susceptibility (i.e., is the device likely to be interfered with rather than to interfere) is also checked to a nominal 5 volts/meter field strength.

While it may seem reasonable to test at this level of EMC/RFI (electromagnetic compatibility/radio frequency interference), one must examine this more closely, to see what really may be required. For example, I have seen residential housing subdivisions built near 100,000 watt FM and TV transmitters. In this case, many of the items in the homeowner's house would likely malfunction. RF-based wireless remotes for TVs, DVRs, DVD players, and set-top boxes would fail routinely. “Garage door openers” would likely revert to the old-fashioned version — get out of your car and manually raise the door with your arm.

Fortunately, RF energy radiated from a line source (or something close to a line source) diminishes with the square of the distance — so one fourth the energy at twice the distance. However energy can, unfortunately, be concentrated and effectively increase in strength due to reflections and antenna gain.

If one looks into the details of various RF sources a bit more, one learns that a smartphone for any of the 2.5/3/4G networks will produce 40V/m at the end of the phone opposite the antenna. The fairly common two-way radio or walkie-talkie used by safety forces and in commercial settings is often a 5W digital radio. Setting this on the dash of your car while data is flowing can expose the instruments in the vehicle to fields that are as much as 200V/m. Set the handy-talkie near a laptop, tablet, or phone, and problems may occur. An additional complication: Signals from multiple devices can on occasion combine to voltage levels equal to the square of the number of carriers times each carrier's peak power.

How does the RF find its way into the susceptible equipment? Isn't it shielded? At frequencies above 1 GHz, the effectiveness of the mesh shields on coaxial cable and shielded twisted pairs starts to diminish rapidly due to skin effect and the physics of microwaves. Also, energy can sneak in along with other signals and then re-radiate once in an enclosure and cause issues.

What can happen to these devices? For typical analog electronics, this energy can be rectified to DC and cause bias points to shift, rendering circuits inoperable. Op-amps' inputs can develop a mysterious offset voltage and produce an error in their outputs. Linear voltage regulators can be pushed out of regulation. Switchers can switch and regulate at the wrong frequency and voltage.

For the more purely digital devices, there can also be problems. Clock signals can develop timing and logic errors, resulting in improper operation or multi-clocking on a single (now non-monotonic) edge. This can even cause carefully constructed software to malfunction (the logic can become highly illogical). Bits can flip, tasks can hang or die in the code, and variables and calculations can change.

One is left with a serious dilemma. For some of the older classes of logic devices that operate at higher supply voltages, there is higher noise immunity. But the tradeoff is shorter battery life and less functionality. Other systems that use the much more common lower voltages use redundant error correction and detection circuits. Refer to applications information for the new MCUs from Spansion and TI. These offer improved resistance to bit-flipping. They employ methods with appropriate levels of paranoia in the firmware to catch and clean up problems — or restart upon serious errors.

What are your experiences in this area? Have you experienced any erratic operation of electronics due to RFI?

Related posts:

29 comments on “EMC Problems in Modern Electronics

  1. Davidled
    November 20, 2013

    I wonder how to solve the issue if new EMC issue is occurred, when old PCB board would be retrofitted by updating the component. That could be a challenge task. There is no way to change routing with via. In some case, new component has the same footprint as that of old. But, new component would generate more EMC issue. Does datasheet provide some clue?

    For EMC related RF, I think that more medical electronic has more concern for EMC related to RF, because EMC and radiation would harmful for human body.

  2. samicksha
    November 21, 2013

     I guess EMC problems are not due to defects in the transmitter, and so improving radio transmitter design may not help, i have seem people often cure this defect by improvement in the RF shielding and the RF filtering of all power.

  3. Pho99
    November 21, 2013

    Have often seen where shielding might be good for say 50 V/m  — immunity / error protection above that level had to be done via special coding in firmware (check sums, parity, CRC, etc) for those rare but still present peaks that might on rare chances hit 800 V/m or more.

     

  4. Vishal Prajapati
    November 21, 2013

    I can understand the scenario of RF interference in the analog circuits to change the offset voltages and produce errorsome results but I wonder it can also flip the bits stored in the RAM or flash. Does it affect differently in both RAM and flash? Can anyone comment the scenario of RF interference in digital circuits?

  5. Davidled
    November 21, 2013

    I wonder how shielding levels is related to check sums, parity and CRC in the firmware.  Sums, parity and CRC would be used in the firmware. Does shielding level is related to this code in the firmware?

  6. Pho99
    November 21, 2013

    The most critical are the clocks in digital circuits — errors in the clocking can result in metastability in the CPU, and also during RAM and Flash operations — RAM is somewhat sensitive (myself had a PC that would not run vista, but would run Linux with a 3G modem stick plugged in a while back)  Flash itself is fairly immune, but if the CPU experiences metastability then the Flash write routine could get called by mistake for example)

  7. Pho99
    November 21, 2013

    Sheilding usualy does very little for signals conducted into the enclosure as part of a design.   Many of those other techniques also solve a host of other issues such as detecting electro-migration faults, flash bit bake-out faults, conducted suceptibility issues, and even ESD related corruption and upset.

  8. Davidled
    November 21, 2013

    It sound like these phenomena that you comments, could be occurred in the worst case analysis. I wonder what many of those other techniques that you illustrated, means. I need to get the clarification that I am missing.

     

  9. samicksha
    November 22, 2013

    connector penetration through shield and poor shield integrity are just another emc problems which can be avoided during commissioning…

  10. eafpres
    November 22, 2013

    When designing things like RF transmit systems you face the problem that you intentionally are generating RF and radiating. Generally the solution is to use shield cans with very small openings for the RF lines which lead to the antennas. There must be filtering on those lines to reduce out of band interference. The antennas are an easy way for unwanted EM energy to get in. For that we often used ferrite cores on the coaxial cables.

  11. Pho99
    November 23, 2013

    One of the most amusing projects to be on was a 60 Watt VHF/UHF radio for aircraft — for the first prototype both the Transmitter and Reciever teams left all the internal shield walls off of those subsystems — Now the DSP/Logic board was in it's own metal can with it's own EMI connector to the rest off the system, and the Audio board had EMC treatments on all active and passive components — the reciever leaked LO energy at such a high level it was noticed by the customers, and the transmitter did not make it to type approval testing without several more rev's of shielding.

  12. jkvasan
    November 25, 2013

    Shielding itself would work like a powerful antenna, if not grounded properly. While designs concentrate on shielding the parts, sometimes, due to improper grounding, the desired effect may not be achieved.

  13. PCR
    November 26, 2013

    Vishal I think that its affect only for the RAM But I am also not sure. I also want to know the scenario. 

  14. Vishal Prajapati
    November 27, 2013

    How to calculate the size and slits in the RF shield? Does it have any special theory or it uses the antenna calculation theories only? What part of the RF circuit should be shielded?

  15. jkvasan
    November 28, 2013

    Vishal,

    I may not be the best person to answer this question. All I understand is everything is dependent on the wavelength of the radiation that you would like to protect from and there are some calculations originating from that direction.

  16. Victor Lorenzo
    November 28, 2013

    @Daej: “I wonder how shielding levels is related to check sums, parity and CRC in the firmware.  Sums, parity and CRC would be used in the firmware. Does shielding level is related to this code in the firmware?

    Shielding is by itself not in direct relation to CRC errors, but mitigates severe PCB traces design errors.

    Under high intensity external electromagnetic fields (EMI), the induced voltage in one PCB track (e.g. data line from external flash to MCU) is directly dependant on the area covered by the signal path for that track. The larger the area the higher the induced voltage. And that voltage (noise, glitch, spike, etc) can cause one severe bit error. In some systems it should be detected and properly managed by firmware, like in modern car engine control units.

  17. Victor Lorenzo
    November 28, 2013

    @Vishal, “I can understand the scenario of RF interference in the analog circuits to change the offset voltages and produce errorsome results but I wonder it can also flip the bits stored in the RAM or flash “,

    You'll also find conductors (tracks, metallizations, wirebondings, etc.) inside the chips, and on every conductor exposed to one varying magnetic field (specially strong ones) a voltage is induced.

  18. Victor Lorenzo
    November 28, 2013

    @Pho99, “Have often seen where shielding might be good for say 50 V/m  — immunity / error protection above that level had to be done via special coding in firmware (check sums, parity, CRC, etc) for those rare but still present peaks that might on rare chances hit 800 V/m or more

    There are two more countermeasures used for that, the watchdog (widely used nowadays) and the redundant CPU, used in some systems with very strong security requirements.

  19. Victor Lorenzo
    November 28, 2013

    @eafpres, “Generally the solution is to use shield cans with very small openings for the RF lines which lead to the antennas.

    Conducted emissions should also be taken into account, specially in the powerlines.

  20. eafpres
    November 28, 2013

    For the slits the equation is basically like a cutoff frequency related to opening size. However there can be conducted interference so what you see in test may not match the calculation exactly. Also consider some designs allow use of one piece shield with no openings–the shield is soldered all the way around. In my experience the designers try to get entire RF section including filters etc inside the shield.

  21. eafpres
    November 28, 2013

    I agree conducted EMI must be addressed by other means. One common way if RF choke in either common mode or differential mode, using soft ferrite like chip beads or array. In some cases even narrowband ceramic filter or SAW filter may be needed.

  22. Vishal Prajapati
    November 29, 2013

    @eafpres, thanks for your suggestion. It looks like you are suggesting to keep only antenna outside the shield and all others including LO, IF and other filters to be inside the shield. As you suggested conducted EMI can be reduced by ferrite beads. Can it be applied to power lines also?

  23. eafpres
    November 30, 2013

    Hi Vishal–To use ferrites for power line applications you have to use differential mode.  This means the power and return lines must pass through the same core.  The way that works, is that the time-dependent power currents cancel each other and the core attenuates noise currents that are one one line or the other (but not both).  Since the power currents cancel, the net attenutation of them is 0.  Thus, the ferrite core selectively attenuates noise.  In many cases a cylindrical core is used, but for board-traces there are SMT designs that can handle 2 or more lines.

  24. yalanand
    November 30, 2013

    @Murray, Thanks for the post. I agree with you that radiation and interference is the serious issue at present days. Not only at FM transmitters ,mobile  phone antennas too causing this interference problems  to electronic devices.  More over this radiation is causing health problems  too.

  25. BillWM
    November 30, 2013

    Cell phone radiation is non-ioninzing radiation — meaning it cannot cause chemical bonds to be broken — it can still produce a small effect due to heating and stiring of the molecules  — the heating due to a cell phone is less than 1/1000th the heating due to a microwave oven.   Statisticaly the largest single risk is distraction while doing a safety critical task like driving — texting and driving, or carrying on an emotional conversation while driving etc.

  26. Davidled
    December 1, 2013

    ->Under high intensity external electromagnetic fields (EMI), the induced voltage in one PCB track (e.g. data line from external flash to MCU) is directly dependant on the area covered by the signal path for that track.

    On this type condition, I highly recommend that ground isolation would be beneficial for noise and radiation suppression. For example, high voltage and low voltage could be separated in the ground view point, if high intensity external electromagnetic is caused by high voltage current closed to data line.

       

  27. Victor Lorenzo
    December 1, 2013

    To use ferrites for power line applications you have to use differential mode ” This type of ferrites are specially usefull in the case of common mode conducted emissions (common mode noise) to avoid the power line to act as an antenna. For other noises you will need line filters (ferrites and capacitors combined to form a low pass filter).

  28. Victor Lorenzo
    December 1, 2013

    @William: “the heating due to a cell phone is less than 1/1000th the heating due to a microwave oven

    Microwave ovens use to output about 1kW, class 1 3G phones put about 2W on the air.

    In Spain it was found that some correlation exists between cancer incidence (leucemia) and long term expossure to radiation coming from massive 2G/3G mobile antenna sites (usually over the buildings).

  29. BillWM
    December 1, 2013

    3G — to get the full power one must be holding the phone during a data call –  Know what you mean about base-stations and occupied buildings — just should not be permited for buildings occupied by people — each carrier added significantly increases peak power — plus the limits were set for an 8 hr working shift , not 14 hrs in a dwelling — a bad call on the limits by regulators — one can easily put a tower on a rooftop to get the antennas further from the occupants

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.