Practical tips for troubleshooting EMI issues

Throughout the world, almost all governments attempt to control the harmful electromagnetic interference (EMI) that emanates from electronic goods produced in their countries (see Figure 1). Very specific rules and regulations cover the design of electronics in order to provide a level of protection and safety for users.

This is a good thing, of course. But it does mean that companies must expend a great deal of time and effort on product design and testing in order to minimize their EMI signatures and pass official EMI certification tests. The bad news is that even after employing good design principles, selecting high quality components, and carefully characterizing the product, when it comes time for compliance testing, an EMI failure can still throw a major monkey wrench into the launch schedule if the testing does not go smoothly in all phases.

Often companies try to protect themselves from this scenario by performing “pre-compliance” measurements during the design and prototyping stages. It’s much better to identify and remediate potential EMI issues before the product is ever sent out for compliance testing.

Of course, most companies’ labs do not contain the test house conditions needed for making absolute EMI measurements. The good news is that it is entirely feasible to identify and resolve EMI issues without duplicating test house conditions. This article discusses some of the techniques you can use to reduce the risk that a product will fail the final full EMC compliance evaluation at the test house. It also includes an example on determining signal characteristics and coincidence in order to zero in on a source of EMI emission.


Figure 1: The changing voltages and currents that make up signals result in electric and magnetic fields.

Understanding the EMI Report
Before diving in to the troubleshooting techniques, a word about EMI test reports would be useful. At first glance, the EMI reports appear to provide straightforward information about a failure at a specific frequency. It might look like a simple matter to use the report data to identify which component of a design contains the offending source frequency and apply some attenuation in order to pass the test on the next pass. However, while many of the test conditions are explicit in the report, some important things to think about may not be so apparent. Before sifting through the design to try to determine the source of the problem, it can help to understand how a test house produces the report.

Consider the EMI test report in Figure 2, which shows a failure at around 90 MHz.


Figure 2: This EMI test report shows a failure at around 90 MHz.

The corresponding tabular data report, shown in Figure 3, details the values for test frequency, measured amplitude, calibrated correction factors, and adjusted field strength. The adjusted field strength is compared in the next column to the specification to determine the margin, or excess, shown in the far right column.

In the margin column shown, you can see that there is a single peak that is above the limit for this specific standard, at 88.7291 MHz, with a -2.3 margin difference from the spec.


Figure 3: This tabular data corresponds to Figure 2. It shows a failure at 88.7291 MHz, but there are factors that make it doubtful that this is the exact frequency.

Page 2

Article page index:

  1. Introduction
  2. You’re finished, right? Not so fast.
  3. Finding the EMI Emanation – Where to Start?
  4. Near field measurements used for EMI troubleshooting
  5. Tracking Down an EMI Emission Source

You’re finished, right? Not so fast. Don’t let all those digits leadyou to believe that this is the precise frequency of the problem EMIsource. In fact, it is very unlikely that the frequency given in thetest report is exactly the frequency of the source. According to theSpecial International Committee on Radio Interference (CISPR), whenperforming radiated emissions testing, different test methods must beused depending upon the frequency range. Each range requires a specificresolution bandwidth filter and detector type, as shown in Table 1. Thefilter bandwidth determines the ability to resolve the exact frequencyof interest; this means that the frequency ranges vary in how well theycan hone in on the offending source.


Table
1 : CISPR test requirements vary with frequency range and impact frequency resolution.

It’s important to notice here that for some frequency ranges, theCISPR test requirements call for the use of a detector type –quasi-peak (QP) — that can obscure the actual frequency. Usually, theEMI department or the external labs begin their testing by performing ascan using simple peak detectors to find problem areas. But for signalsdiscovered that exceed or are close to the specified limits, they alsoperform QP measurements. QP is a method defined by EMI measurementstandards that serves to detect the weighted peak value of the envelopeof a signal. It weights signals depending upon their duration andrepetition rate, so that more emphasis is placed on signals that couldbe interpreted as “annoying” from a broadcast perspective. Signals thatoccur more frequently will result in a higher QP measurement thaninfrequent impulses. In other words, with problem signals that happenmore frequently, it is possible that the absolute amplitude of theoffending signal will be masked by a QP measurement.

The good news here is that peak and quasi-peak scans are still usefulfor pre-compliance testing. An example of peak and QP detection is seenin Figure 4. Here, a signal with an 8 μs pulse width and 10 msrepetition rate is seen in both peak and QP detection. The resultant QPvalue is 10.1 dB lower than the peak value.


Figure 4: Comparison of peak detection and quasi-peak detection.

A good rule to remember is that the QP value will always be less thanor equal to the peak detect value, never larger. So you can use peakdetection to do your EMI troubleshooting and diagnostics. You don’t needto be accurate to the same degree as an EMI department or lab scan,since the measurements are all relative. If the QP value in your labreport shows the design was 3 dB over and your peak detect value is 6 dBover, then you know that you need to implement fixes that reduce thesignal by 3 dB or more.

Test houses usually perform the scans for the EMI report underspecial conditions that your company lab may not be able to replicate.For instance, the device under test (DUT) may be placed on a turn-tableso that information can be collected from multiple angles. This azimuthinformation is quite useful as it will indicate the area of the DUT fromwhich the problem is emanating. Or the EMI test house might make theirmeasurements in a calibrated RF chamber and report the results as ameasure of field strength.

Fortunately, you do not need to duplicate test house conditions inorder to troubleshoot EMI test failures. Instead of using absolutemeasurements that are performed in the highly controlled EMI testfacility, troubleshooting may be performed using the information in thetest report, a good understanding of the measurement techniques used togenerate the report, and relative observations taken around the DUT toisolate sources and gauge the effectiveness of remediation.

Article page index:

  1. Introduction
  2. You’re finished, right? Not so fast.
  3. Finding the EMI Emanation – Where to Start?
  4. Near field measurements used for EMI troubleshooting
  5. Tracking Down an EMI Emission Source

Finding the EMI Emanation – Where to Start?
Nowit’s time to move on to the work of zeroing in on the unwanted EMIsource. When we look at any product from an EMI perspective, the wholedesign can be considered a collection of energy sources and antennas.Common (but by no means the only) sources of EMI problems include:

  • Power supply filters
  • Ground impedance
  • Inadequate signal returns
  • LCD emissions
  • Component parasitics
  • Poor cable shielding
  • Switching power supplies (DC/DC converters)
  • Internal coupling issues
  • ESD in metalized enclosures
  • Discontinuous return paths

To determine the source of energy on a particular board and theantenna at the heart of a particular EMI problem, you need to examinethe periodicity of the observed signals. What is the RF frequency of thesignal? Is it pulsed or continuous? These signal characteristics can bemonitored using a basic spectrum analyzer

You’ll also need to look at coincidence. What signal on the DUTcoincides with the EMI event? It is common practice to use anoscilloscope to probe the electrical signals on the DUT. Examining thecoincidence of EMI problems with electrical events is arguably the mosttime consuming process in EMI troubleshooting. In the past, it has beenvery difficult to correlate information from spectrum analyzers andoscilloscopes in a synchronized way.

However, the introduction of the mixed domain oscilloscope (MDO) haschanged that by providing synchronized time-correlated views andmeasurements. This instrument, shown in Figure 5, simplifies the EMItroubleshooting process by making it relatively easy to see what signalcoincides with which EMI event.


Figure 5: A mixed domain oscilloscope (MDO) combines a spectrumanalyzer, oscilloscope, and logic analyzer in a single unit thatproduces synchronized time-correlated measurements from all threeinstruments. Shown here is the Tektronix MDO4000B.

An MDO combines the capabilities of a mixed signal oscilloscope with aspectrum analyzer. With this combination, you are able to automaticallydisplay and trigger on analog signal characteristics, digital timing,and bus transactions as well as RF. Some MDOs also have the ability toacquire or view both frequency spectrum and time domain traces,including RF Amplitude vs. Time, RF Phase vs. Time, and RF Frequency vs.Time. An RF Amplitude vs. Time trace is shown in Figure 6.


Figure 6: This shows an MDO’s time-correlated view with an RF Amplitude vs. Time trace.

Making Relative Measurements with Near Field Probing
Whilecompliance-testing procedures are designed to produce absolute,calibrated measurements, troubleshooting can be performed in large partusing relative measurements of the electromagnetic fields emanating froma DUT. In particular, you can zero in on the source of the energy byprobing the behavior of wave impedance in the near field, using thespectrum analyzer capabilities and RF channel of an MDO. You do this atthe same time that you are probing the signal with passive probes on oneof the oscilloscope’s analog channels in order to uncover a signal thatcorrelates with the RF.

First, though, it helps to have some background about theelectromagnetic field regions that are to be probed. Figure 7 shows thebehavior of wave impedance in the near and far fields, and thetransition zone between them. You can see that in the near field region,fields can range from predominantly magnetic to predominantly electric.In the near field, where non-radiative behaviors are dominant, waveimpedance depends upon the nature of the source and the distance fromit. In the far field, the impedance is constant, and measurements dependnot only upon activity that is observable in the near field, but alsoupon other factors such as antenna gain and test conditions.


Figure 7: This shows the behavior of wave impedance in the near and farfields, and the transition zone between them. Near field measurementsare the ones used for EMI troubleshooting.

Article page index:

  1. Introduction
  2. You’re finished, right? Not so fast.
  3. Finding the EMI Emanation – Where to Start?
  4. Near field measurements used for EMI troubleshooting
  5. Tracking Down an EMI Emission Source

Near field measurements are the ones used for EMI troubleshooting,since they allow you to pinpoint the sources of energy without requiringspecial conditions in a test site. However, compliance testing isperformed in the far field, not the near field. You typically won’t usethe far field, because it’s complicated by too many variables: Thestrength of the far field signal is dependent not only on the strengthof the source but also the radiating mechanism as well as any shieldingor filtering that may be in place. As a rule of thumb, remember that ifyou are able to observe a signal in the far field, then you should beable to see the same signal in the near field. (However, it is possibleto observe a signal in the near field and not see the same signal in thefar field.)

Near field probes are essentially antennas that are designed topick-up magnetic (H Field) or electric (E Field) variations. In general,near field probes do not come with calibration data, so they areintended for making relative measurements. If you’re not familiar withthe probes used to measure H Field and E Field variations, it helps tounderstand a bit about near field probe designs and the best uses ofeach:

H-field (magnetic) probes have a distinctive loopdesign, as shown in Figure 8. It’s important that the H-field probe isoriented so that the plane of the loop is in line with the conductorbeing evaluated, thus positioning the loop so that the magnetic fieldlines of flux pass through it.


Figure 8: Position an H-field probe in line with the current flow so that the magnetic field lines pass through the loop.

The size of the loop determines the sensitivity as well as the areaof measurement, so care must be taken when using these types of probesto isolate a source of energy. Near field probe kits will often include anumber of different loop sizes, so that you can use a progressivelysmaller loop size in order to narrow the area of measurement.

H-field probes are very useful for identifying sources with relatively high current such as:

  • Low impedance nodes and circuits
  • Transmission lines
  • Power supplies
  • Terminated wires and cables

E-Field (electric) probes function as small monopoleantennas and respond to the electric field or voltage changes. Whenusing these types of probes, it is important that you keep the probeperpendicular to the plane of measurement, as shown in Figure 9.


Figure 9: Position an E-field probe perpendicular to conductors to observe electric fields.

In practice, E-field probes are ideally suited for zeroing in on avery small area and identifying sources with relatively high voltages aswell as sources with no termination such as:

  • High impedance nodes and circuits
  • Unterminated PCB traces
  • Cables

At low frequencies, the circuit node impedances in a system can varygreatly; knowledge of the circuit or experimentation is required todetermine whether an H-Field or E-Field probe will provide the mostsensitivity. At higher frequencies, these differences can be dramatic.In all cases, making repetitive relative measurements is important sothat you can be confident that the near-field emission results from anychanges implemented are accurately represented. The most importantconsideration is to be consistent in the placement and orientation ofthe near field probes for each experimental change.

Article page index:

  1. Introduction
  2. You’re finished, right? Not so fast.
  3. Finding the EMI Emanation – Where to Start?
  4. Near field measurements used for EMI troubleshooting
  5. Tracking Down an EMI Emission Source

Tracking Down an EMI Emission Source
In thisexample, an EMI scan of a small microcontroller indicated an over-limitfailure from what appears to be a broad-band signal centered around 144MHz. Using the spectrum analyzer capabilities of an MDO, the first stepis to connect an H-field probe to the RF input to localize the source ofthe energy using relative near field measurements.

As discussed above, it’s important that the H-field probe is orientedso that the plane of the loop is in line with the conductor beingevaluated. Moving the H-Field probe around the PCB, you can localize thesource of energy. By selecting progressively narrower aperture probes,you can focus the search in a smaller area.

Once the apparent source of energy has been located, the RF Amplitudevs. Time trace, as shown in Figure 10, plots the integrated powerversus time for all signals in the span. Using this trace, one canclearly see a large pulse in the display. Moving spectrum time throughthe record length, it is apparent that the EMI event (which is the wideband signal centered around 140 MHz) directly corresponds to the largepulse. To stabilize the measurement, turn on the RF power trigger, andthen increase the record length to determine how often the RF pulse isoccurring. To measure the pulse repetition period, enable themeasurement markers and directly determine the period.


Figure 10: The MDO’s RF Amplitude vs. Time trace (top graph) shows asignificant pulse at 140 MHz. The spectrum display (bottom graph) showsits frequency content.

The next step to positively identify the source of EMI is to utilizethe oscilloscope portion of the MDO. Keeping the same setup, turn on theanalog Channel 1 of the oscilloscope and browse the PCB looking for asignal source that is coincident with the EMI event.

After browsing signals with the oscilloscope probe for a while, thesignal in Figure 11 was spotted: in this case, a power supply filter. Itcan be clearly seen on the display that the signal connected to Channel1 of the scope is directly correlated to the EMI event. Now, an EMIremediation plan can be developed and the problem corrected beforeattempting certification tests.


Figure 11: Using a passive probe on one of the oscilloscope’s analog channels uncovers a signal that correlates with the RF.

Conclusion
Failing an EMI compliance test can puta product development schedule at risk. However, pre-compliance testingcan help you troubleshoot EMI problems before ever reaching that stage.Instead of making absolute measurements in a highly controlled EMI testfacility, you can use the information in the EMI test report to makerelative observations that can be used to isolate sources and gauge theeffectiveness of remediation.

Effective EMI troubleshooting generally involves near-field probingto look for relatively high electromagnetic fields, determining theircharacteristics, and using a mixed-domain oscilloscope to correlate thefield activity with circuit activity to determine the EMI source. Thetroubleshooting techniques outlined here can effectively help youisolate the offending source of energy so that you can remediate itbefore submitting the design for EMI certification.

About the author
As Mainstream Technical Marketing Manager for Tektronix, Varun Merchantsupports mid-range wireless and power conversion product offerings.Before joining Tektronix, he worked for several year in productdevelopment and design in the semiconductor industry. He holds an MSEEfrom the University of California, Santa Barbara and an MBA from theStern School of Business, NYU.

Article page index:

  1. Introduction
  2. You’re finished, right? Not so fast.
  3. Finding the EMI Emanation – Where to Start?
  4. Near field measurements used for EMI troubleshooting
  5. Tracking Down an EMI Emission Source

Leave a Reply

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