Advertisement

Blog

EMI Noise, Part 2: Don’t Be the Victim

As mentioned in (EMI Noise, Part 1: Don’t Be the Problem), there is always a source, a path, and a victim when it comes to electromagnetic interference (EMI) and other types of electrical noise. This post focuses on the victim and what you can do to avoid becoming one. To recap, a variety of passive components can be used to suppress electrical noise, including: multilayer ceramic capacitors (MLCCs), feedthrough capacitors, specialized EMI filters, ferrite beads, and common mode chokes. These same devices are also used to reduce EMI radiation. Assuming smart circuit board layout has been implemented, each of these devices offers unique advantages depending on the nature of the electrical noise coupled to a system.

To solve electrical noise problems effectively, the source of the EMI, its coupling path to the victim, and the victim must be identified. So, before proceeding, let’s review some common sources of EMI:

  • Electric Motors
  • Lightning
  • Electric Power Lines
  • Mobile Electronics
  • Loose Wiring Harnesses
  • Garage Door Openers

There are four possible mechanisms for EMI to be coupled into a system:

  • Conducted coupling occurs when source and victim circuits share current paths.
  • Electric field coupling occurs when there is a difference in charge between two lines placed in the near field, a distance of less than half the wavelength of the radiated noise. This difference in charge induces a voltage between the two lines.
  • Magnetic field coupling occurs when current flowing through a wire induces a magnetic field that couples to a nearby wire, which in turns induces a current in that wire.
  • Radiated coupling occurs when far field — a distance of several wavelengths between source and victim — radio transmissions couple into a system.
  • Figure 1

    Source - Coupling Path - Victim.

    Source – Coupling Path – Victim.

    Identifying the source
    Identifying the source of EMI noise is not always intuitive. For example, if lightning strikes near your home during a storm and the lights go out, you can deduce that the lightning was the source of the problem. However, if it’s a clear day outside and you notice the television flickering, you may be tempted to think that the television is defective, warranting repair or return, but the problem could likely be due to a nearby EMI source, such as power lines, a new RF tower, or even construction equipment, all of which can produce disruptive noise.

    Identifying the current path
    When trying to identify the EMI coupling path, it is important to remember that current flows through the path with the least amount of impedance, which is the sum of the resistance and reactance of the current loop.

    Equation 1

    Impedance, in which Z = impedance, X = reactance, R = the conductor's resistance, f = frequency, and L = inductance of current path.

    Impedance, in which Z = impedance, X = reactance, R = the conductor’s resistance, f = frequency, and L = inductance of current path.

    The impedance equation above uses the inductive reactance, since the impedance of the current loop largely depends on the inductance of the path that the current takes. At low frequencies (typically below 1MHz), the path that the current takes is largely dependent on “R.” Since R>>2ΠfL, the current path will be through the path of lowest resistance.

    At higher frequencies (typically 1MHz and above), the path that the current takes is largely dependent on its inductive reactance. Since 2ΠfL>>R when “f” is high, the current will tend to return through the path of lowest inductance. At low frequencies, signals will travel through the low resistance path. At high frequencies, signals will travel through the low inductance path. As such, to determine the current path, just compare the impedance of the possible current loops at the operating frequency.

    Available EMI filtering solutions
    Capacitors, inductors, and shields are some of the many available tools that can help suppress EMI in a system. With regard to capacitors, standard MLCCs and feedthrough capacitors can both be used to decouple noise from non-critical and automotive systems. MLCCs are relatively low cost components, which can be an advantage in many applications, but feedthrough capacitors are typically preferred for EMI filtering due to their broad, high-frequency filtering range and higher-level filtering capabilities. Often more critical applications — including medical, aerospace, and space applications — require more specialized filters, which are available from several manufacturers.

    Figure 2

    AVX's W2F4 Series ceramic feedthrough capacitor arrays are ideal for high-frequency EMI filtering of multiple parallel lines.

    AVX’s W2F4 Series ceramic feedthrough capacitor arrays are ideal for high-frequency EMI filtering of multiple parallel lines.

    Inductive devices are also effective at suppressing noise. For example, ferrite beads exhibit low impedance at low frequencies and very high impedance at high frequencies to suppress high frequency noise. Common-mode chokes are also frequently used to filter common-mode noise out of parallel wires in applications ranging from high data rate USB and HDMI to power systems.

    Implementing EMI filtering
    (See Demcko, R.; Mello, C.; and Ward, B., “AVX EMI Solutions.”)
    In addition to the use of filters, effective EMI suppression also requires smart PCB layout techniques. Depending on the source, the coupling mechanism, and the victim, there are several layout solutions that can be implemented to reduce the amount of noise being coupled to the system. For example, EMI can be significantly reduced through PCB layout techniques including:

    • Using a MultiLayer PCB with uncut Vcc and Ground planes
    • Using proven decoupling methods
      1. Using a high frequency decoupling capacitor at each IC
      2. Using a high frequency decoupling capacitor at the regulator
      3. Connecting decoupling capacitors in the lowest inductance method possible
    • Keeping I/O traces short
    • Using minimal cable length for connections
    • Terminating high-speed lines
    • Shielding cables if necessary
      1. Grounding both ends of cables
      2. Considering multiple grounds on a ribbon cable
    • Protecting ESD sensitive components with a transient voltage suppressor
      1. Multilayer varistors (MLVs) clamp bi-directionally in the on-state and like EMI filters in the off-state
      2. Placing the MLV as close to the transient source as possible
      3. Considering a dedicated Vcc line to clocks
      4. Considering limiting the number of 90° trace features, since two 45° trace features typically radiate less
      5. Using balanced trace design if possible
      6. Filtering at connector pins if necessary

    This list is by no means exhaustive and doesn’t apply to every application, but it is a solid collection of good guidelines to take into consideration when laying out PCBs to minimize EMI noise.

    Conclusion
    In summary, to avoid becoming a victim of EMI: Identify its source, coupling path, and victim; utilize filtering solutions well suited to your application; and implement smart PCB layout techniques tailored to your system.

19 comments on “EMI Noise, Part 2: Don’t Be the Victim

  1. vasanjk
    October 31, 2014

    Hi Edgardo

     

    I would definitely say that this is one of those very informative posts with a practical and direct approach.

     

    By the way, in regard to Emi noise – both common mode and differential, how can a designer approach in a decisive manner?

  2. vasanjk
    October 31, 2014

    I would like to reframe my query.

     

    How to approach and handle both?

    Which one comes first or is it depending on the application?

     

  3. SunitaT
    October 31, 2014

    @Edgardo, thanks a lot for the informative post. I am curious to know how difficult or easy is it to find the Identify the source and the coupling path and is it necessary to implement all the EMI suppression methods ?

  4. chirshadblog
    October 31, 2014

    @sunita: Well its not easy indeed but it can be measured only when the situation demands 

  5. Davidled
    November 2, 2014

    It is a pure and zero EMI environments surrounding lumberjack house. It seems like modern house or building might have EMI pollution that unfortunately, people could not feel anything. In the future, when any house or building is in the market, EMI pollution level might be indicated in the mortgage paper.

  6. nasimson
    November 30, 2014

    @ Daej:

    > It is a pure and zero EMI environments surrounding lumberjack
    > house. It seems like modern house or building might have EMI
    > pollution that unfortunately, people could not feel anything. In
    > the future, when any house or building is in the market, EMI
    > pollution level might be indicated in the mortgage paper.

    Even the lumberjck house will have EMI from neighboring house or power transmission lines above or the cellular coverage in the near by.

    As far as the mortgage paper is concerned, you are  joking, right? Or am I missing something here?

  7. Sachin
    November 30, 2014

     Well its not easy indeed but it can be measured only when the situation demands

    @chirshadblog, I am not clear about situation demands. What kind of situation demand are you referring to ?

  8. Sachin
    November 30, 2014

    In the future, when any house or building is in the market, EMI pollution level might be indicated in the mortgage paper.

    @DaeJ, Is government framing new laws to disclose the EMI pollution levels in mortgage paper ? If that happens that would be a welcome move.

  9. Sachin
    November 30, 2014

    Even the lumberjck house will have EMI from neighboring house or power transmission lines above or the cellular coverage in the near by.


    @nasimson, true. We cannot escape from EMI interference. As you mentioned power transmission lines or cellular coverage in the near by places definitely creates EMI interference. I am not sure if there is a way out.

  10. exmenendez
    January 2, 2015

    @SunitaT0, The difficulty of idenitifying the source of and path of EMI varies by situation. It may not be necessary to implement every EMI suppression method in every situation, only the methods that the situation calls for.

  11. exmenendez
    January 2, 2015

    @vasanjk, Please know that every design is different so consider the parameters of your specific design. First and foremost, good circuit board layout techniques should be implemented into every design to avoid becoming and EMI source or a vicitim. Second, consider using EMI filters where necessary. Obvious places requiring common mode filtering will be where there are parellel high speed lines (USB, CAN, etc). As mentioned in the post, consider the possible sources and paths and where your design may be vulverable when designing your circuit.

  12. Aplonis
    September 4, 2016

    I have a situation at work where a horrendous amount of noise is appearing on my 4-20mA current loops. The signal sources are a pair of Emerson Micro-Measurement CMF mass flow sensors, which I am sure are not producing the noise. I've used those for years, and always they are rock solid. I checked them regardless, by unhooking and splicing in a 500 Ohm resistor then viewing on an oscilloscope, and they are fine. The CMF's are, however, built into a test stand custom designed by Flow Systems which I'm trying to modify because they never quite performed up to par.

    Flow Systems normally makes air-flow test stands. This one, I think, is their first try at controlling hydraulic flow. They had been governing flow via PID calculated inside of LabVIEW from values gained digitally via serial loops: RS-485 for flow, and HART (slow!) for pressure. Okay for air, since it is squishy and slow, but nowhere near real-time enough to do incompressible fluid. So first thing I do is dedicate a hardware PID controller (Watlow) and set that up for 4-20mA control so as to get the system closer to instantaneous. I switch out all of the sensors for 4-20mA as this has always served well before. But now I find there is all of this noise on the 4-20mA signals for mass flow.

    I was anticipating noise on the signals for pressure (pump ripple) and so have those signals on a voltage input via Sallen-Key OpAmp filters. I wasn't expecting any such on the current loops, though. Not at all. And it's horrendously large.

    The sensor is a National Instruments NI-9203 module. And once again, I've used those for years. And never has there been any such issue.

    The lines running between are shielded, twisted pair by Alpha wire. Again, this can't be the issue.

    The CMF can drive 820 Ohms and the NI-9203 has only 130 Ohms. So first thing I tried was to drop in a 680 Ohm resistor between. That knocked it down some, but not enough. So next I'll build an 2-stage, in-line RC filter comprised of a pair of 340 Ohm resistors and metal film capacitors to go in between. I doubt, however that it's going to be enough.

    I suppose that I might (like for pressure) convert from 4-20 mA to voltage across a 500 Ohm 0.01% resistor to get a range of 2 to 10 Volts and build yet another set of Sallen-Key filters. I could do that, however…

    It occurs to me, however, that surely there must somewhere exist an active filter circuit for current, even one specific to 4-20mA loops. But I can't seem to find any such. So my question is, have you a solution for that? Do you know of any OpAmp circuit for low-pass filtering whose output is 4-20mA and not volts?

    I would be most gratified if someone could please kindly answer just that. A current-output LP filter built out of OpAmps. Does one exist? Where's a schematic? Alternatives I already have, as stated above. I'm looking for only that. Thanks in advance.

  13. antedeluvian
    September 4, 2016

    @Aplonis

    There are several questions on your situation and sometimes it seems your observations fly in the face of conventional wisdom.

    I have a situation at work where a horrendous amount of noise is appearing on my 4-20mA current loops. ….. And it's horrendously large.

    Is it actually affecting the performance of the receiver? The 4-20mA loop. as with any current loop, has a good common mode rejection and as such you have to examine why there is an imbalance to cause the noise. 

     But now I find there is all of this noise on the 4-20mA signals

    What kind of noise, what frequency, what voltage and  how are you measuring it? What are you using for your negative of the measured signal?

    The CMF can drive 820 Ohms and the NI-9203 has only 130 Ohms. So first thing I tried was to drop in a 680 Ohm resistor between. That knocked it down some,

    Did it knock down the voltage across the 130 ohms or across the composite 810 resistance? and again how did you measure it? Is is AC or DC (or both)? Where is the 'scope probe connected?- remember that it is usually connected to earth-ground of the mains and if your system is grounded elsewhere you have a ground loop.

    I suppose that I might (like for pressure) convert from 4-20 mA to voltage across a 500 Ohm 0.01% resistor to get a range of 2 to 10 Volts and build yet another set of Sallen-Key filters

    In all likelihood that is already what is inside your NI-9203. And in terms of noise, a lower resistance is better than a higher one, but in my experience there is not a big difference in performance between 250R and 10R impedance when it comes to post filtering a current loop.

    Finally you refer to “pump ripple ” as noise. How does this noise correalate to the noise you are seeing in the other loop. Does it look the same- I am not suggesting that there is any cross relationship between the two, just trying to clarify your definition of noise.

    As an outsider looking in, my first suggestion would be to look for a ground loop issue, possibly induced by the scope, but just by both ends of the loop being connected to ground.

  14. Aplonis
    September 4, 2016

    @antedeluvian

    In answer to your questions about my post…

    Yes, it is affecting performance. Noise on the order of nearly 1 mA on a range of 16 mA meets my criteria for naming it horrendously large. I do understand the advantages of current loops. I like them therefor. I haven't calucated the period timing of the noise. I'll go in and do that tomorrow. As the plant will be shut down, and everyone gone, no TIG or EB welders running, no vacuum furnaces, nor anything else, I can eliminate those as the source. Although, to be certain, they would be all in the building next door. Nor have those been a problem in the combustion lab where I normally work which is not so very far distant. I suspect the noise to be 60 Hz, although it was hardly sinusoidal in aspect during consecutive scans of 50 samples each.

    I am measuring the noise by capturing scans of 50 samples on the NI-9203. Also by watching the drop across 500R and 820R inline resistors via a digital o'scope. Also by inserting a Fluke meter in-line on the mA scale and just noting the bar-graph fluctation. Not very quantitative, that last, but I had it, and so gave a look. Switching to AC scale on the Fluke, I see there too. The noise truly is there.

    By inserting an 820R inline with the loop, outside the 130R integral to the NI-9203, I observed the NI-9203 to report reduced noise. It held the same 12 mA which I was having the CMF-025 send (in test calibration mode) with noise upon it. I quantify the noise by taking a scane of 50 consecutive samples into an array, sorting the array, then subracting Min from Max to get the span.

    You ask what I'm using for my “negative of the signal” which somehow, I fail to grok. I measure the drop across an in-line resistor. I can do it with a Yokogawa DL820P digial oscilloscope. Now that on this particular scope, the probe shield is not at earth ground. It floats, and has a high isolation resistance from each of its other sixteen probes. It doesn't ground unless I choose it. The BNC jacks are special in that the probe-shield is recessed inside the jack (an annoyance most times other than this, since not many BNC cables connect to it well). I use the scope only so as to see what there might be to be seen. The scope sees it too. It hasn't affected the signal, I'm sure, because if it had, then the 12 mA would have fallen to zero, which it did not. How I'm quantifying the noise is by the span of 50 readings inside the NI-9203, as detailed above. That is where I need it to go away.

    Adding in 820R outside the NI-9203 affected the noise seen inside of the NI-9203 (across its internal 130R) by a significant, and repeatable, degree. By well more than half. The span of 50 readings fell off abruptly, just not enough.

    I shall certainly look for a ground loop. I did find that Flow Systems had, against recommendation, grounded the shield from the sensor to the transmitter at both ends. This I have already corrected, to null advantage. The shields for digital and analog cables are separate. I have tried clipping grounds to earth on each and both, to null effect. Other things I shall try yet more things tomorrow.

    My question, however, as yet unansered, is does there exist a low pass filter circuit that works for 4-20 mA the same as can be easily done for voltage. Do you know of such a circuit? If so where might I find a schematic?

  15. antedeluvian
    September 5, 2016

    @Aplonis

    My question, however, as yet unansered, is does there exist a low pass filter circuit that works for 4-20 mA the same as can be easily done for voltage. Do you know of such a circuit? If so where might I find a schematic?


    A suitably sized capacitor across the input of the NI-9203 should short circuit the AC current and bypass the input resistor, allowing only the DC through the internal 130R resistor.

    You may find some configuration of a “gyrator” that would suit your needs. Try googling it. I covered my experience with it in a blog “Combining power and data wires“, but I don't think the configuration shown will help, since it is sourcing the current and prevent signal feedback to the source.

  16. antedeluvian
    September 5, 2016

    @Aplonis

    By inserting an 820R inline with the loop, outside the 130R integral to the NI-9203, I observed the NI-9203 to report reduced noise. It held the same 12 mA which I was having the CMF-025 send (in test calibration mode) with noise upon it

    Where does the Watlow fit in this?

  17. Aplonis
    September 5, 2016

    The Watlow is an external consideration. It too will be employing 4-20mA both to receive a command from an NI module, and to send its own command to a Badger Research electrically controlled needle valve to throttle flow.

    Going back into work now. I'll try and check back here later. Thanks for your input so far.

  18. antedeluvian
    September 5, 2016

    P.S. If you use the capacitor, it should be across the total resistance (including the additional series resistance if you use it) or the signal will be smoothed and create a DC component.

  19. Aplonis
    September 5, 2016

    Solved,

    After going into work, and with all the industrial noise sources quiecent because of the plant being shut down for the holiday, the noice seen before was yet still there.

    After tracing every wire to do with 24 Volts, both the analog and digital supplies, all were found to match the print supplied by the vendor. Yet neither of these were tied to ground. Nor did any wire in the print showed where they they had ought to be. Both were floating free of ground, apparently by the vendor's design.

    I decided that couldn't be right. So I ran wires from the negative terminals of each directly to chassis ground via 18 GA yellow/green wire, and then did the noise fall down to more tolerable levels.

    Nevertheless, it still helps yet more having a cap and resistor inline with the 4-20 mA signal. So I will leave that in place until and unless there seems a reason to take it out.

Leave a Reply

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