Spectrum analysers are very useful tools for observing radio transmissions. One of the more common applications is to monitor a known channel or band of frequencies. In this note, we are going to show how to quickly capture an FM radio transmission and then configure the analyser to provide us with the frequency deviation of that signal.

NOTE: If you are unsure of the power of the signal that you are attempting to measure, it may be a good idea to add external attenuation to the RF input of the analyser to help prevent damage to the analysers sensitive measurement circuit. You are ultimately responsible for any damage caused by excessive signal power.

1. Connect the appropriate antenna to receive the signals of interest:

2. Press Frequency >  Center Frequency and set the center frequency to match the frequency of interest. Here, we are setting the center frequency to 100.7 MHz, a local FM radio station:

2. Press Span > Span and set the span to an appropriate level for your application. If you are looking at a single channel, you can set the span to 500 kHz or so. This will allow you to observe the channel and the full range of the frequency deviation:

3. If the signal is still very low or cannot be easily observed through the noise, you can lower the attenuation by pressing Amplitude > and lower the Attenuator setting. You can also enable the preamplifier.

NOTE: Use caution as lowering the attenuation and enabling the preamplifier makes the instrument more susceptible to damage due to overpowering the input.

After adjusting the preamp and attenuator, you can see the signal rise out of the noise:

4.  Now, we can activate another trace on the display and we can set it to be a Max Hold trace type. This trace type records the largest amplitude for each frequency bin and it will stay there until it has been manually cleared. Over time and successive scans, the Max Hold trace will show us the frequency deviation of our signal.

Note that trace 1 type is Clear Write. This overwrites each frequency bin amplitude with a new value for each scan:

The FM deviation of this signal is approximately 4 divisions.

The Span is 500 kHz and there are 10 divisions on the display. Therefore, each division is 50 kHz. So, the FM signal deviates around 200 kHz.

You can use markers for more accurate measurements:

Or adjust the span to see adjacent transmissions:

Electromagnetic interference (EMI) can cause a host of problems, especially when developing a product or attempting to pass mandatory electromagnetic compliance (EMC) tests. Garbled displays, bad data, or complete malfunctions can occur when a design is effected by EMI. To minimise the effects of interference, government agencies like the Federal Communications Commission (FCC) in North America create and enforce standards that set limits on the EM output of a product type. Testing to the specifications is commonly referred to as Electromagnetic Compliance (EMC) testing.

Many EMC test failures stem from the interaction of unintentional radio frequency (RF) emissions with a circuit or element within the design itself. The electric and magnetic fields that cause this interference are not visible to the unaided eye, which can present complications when trying to isolate the root cause and minimize the effects of the EMI.

• What is causing the issue?
• Where is the source of the signal or energy causing the radiation?
• How can I fix it?

Fortunately, there are simple tools and techniques that can help identify the sources of EMI. Once you can identify the source, you can begin to build up a list of solutions to the problems. These techniques are not part of the mandatory compliance tests required to pass EMC testing. Rather, these are pre-compliance test techniques that help identify potential areas of EMI as quickly as possible without the burden of expensive test equipment and setups.

In this application note, we are going to introduce some common pre-compliance test techniques for identifying potential problematic EMI sources using near-field and current probes. These techniques can save you time and money by isolating problem areas quickly, and with a little fixturing, you can create repeatable test stations to help correlate data. This knowledge can then be used to “design for EMC” in your future products.

NOTE: Pre-compliance tests are designed to help identify and resolve issues that may hinder passing full compliance tests. Pre-compliance testing is not a replacement for full compliance testing at a certified lab.

ELECTROMAGNETIC RADIATION BASICS

In electronics, EM radiation is most commonly caused by a current flow or voltage build-up along or through a conductor. This includes traces on a PC board, discrete wires, component leads/pins, connectors, or any other metal, including the chassis, rack, or product enclosure. Recall that EM radiation is actually a combination of electric and magnetic field components. It is described as the propagation of orthogonal time-varying electric and magnetic fields as shown in figure 1.

Figure 1: Electromagnetic wave propagation out of the page (top left), to the right (top right) and out of the page at an angle. Note that the E and H fields are orthogonal (90°) to one another.

While the electric (E) and magnetic (H) fields are created by the same phenomena, they physically behave quite differently in the environment. Magnetic fields are only created by moving charges (current). In most circuits, current is conducted by traces on the PC board, component pins/leas, and discrete wires. Therefore, the magnetic field tends to dominate the EM radiation produced by the traces and wires that route signals and power to different parts of the design.

Visualising the magnetic field can be a bit easier if you go back to your Physics texts. Recall that the magnetic field of an infinitely long straight wire can be calculated by applying Ampere’s law:

For a circular path centered on the wire, the summation becomes:

Where:

Figure 2 is a physical representation of this relationship. Note, this is also described by the “right-hand-rule” wherein if you were to point the thumb of your right hand in the direction of the current flow, then the magnetic field lines form concentric rings that wrap around the conductor in the direction of your fingers.

Figure 2: Magnetic field produced by a current

Unlike the magnetic field, electric fields can be created by moving or static charges. In this way, electric field effects dominate over magnetic fields when searching for EM radiation on surfaces like heatsinks or metal enclosures. The effects of the electric field also tend to dominate further away from the source (far-field). Far-field measurements are more susceptible to error due to environmental factors like radio stations, WiFi, and intentional RF. Far-field measurements, like those performed during radiated emissions portion of a compliance test, require more setup, equipment, and expertise than near-field.

By measuring the amplitude and frequency of the magnetic and electric fields that are generated by elements of a product, we can identify the areas that have the highest potential to cause EMI issues.

EQUIPMENT LIST

Here are the basic requirements for a near-field troubleshooting kit:

Spectrum Analyzer/EMI Receiver: Measures RF power with respect to frequency. The analyser should have a maximum frequency of at least 1 GHz, DANL of -100 dBm (-40 dBuV) or less, and a minimum RBW of at least 10 kHz.

Figure 3: A SIGLENT SSA3021X 2.1 GHz spectrum analyser.

Near-field probes: Commercial or handmade. Many are magnetic (H) field probes, but there are also electric (E) field probes as well.

Current probes: Commercial or handmade.

50 Ohm cable: Use a cable with connectors that mate to the near-field probes and the RF input of the spectrum analyzer. Many commercial probes can be purchased with a cable and any adapters that may be required.

PROBES

Since EMI cannot directly observed by the human eye, we need some tools to help. Recall that moving charges in a conductor produce magnetic and electric fields that radiate throughout space from the conductor. We can use these fields to induce a voltage in a circuit. Then, measure that induced voltage and therefore indirectly measuring the strength of the original field. The two most common types of probes used in EMI troubleshooting are near-field probes and current clamps.

Magnetic field probes and current clamps operate on a similar principle. The magnetic field that flows through the “loop” area of the probe induces a voltage that can be measured (figure 4). Larger loop areas pick up more magnetic flux, and are therefore better suited to finding smaller signals, but smaller loops offer better spatial resolution. Many kits come with multiple loop sizes (figure 5) to help strike the balance between sensitivity and spatial resolution.

Electric field probes do not generally have a loop area. They pick up the electric field similar to a monopole antenna. The rotation of an electric field probe is not critical as with the magnetic field probe, but the distance from the signal source is.
Here are some guidelines for probing:

• Measure the background radiation by powering off the device-under-test and monitor the analyser display. Note any RF that may be caused by background or environmental conditions and retest often.
• Probe displays, communications port terminals, and any cutout/air vent/seam of the enclosure. These are common problem areas.
• E and H field probes positioned closer to the signal source will measure higher amplitudes
• H field probes oriented perpendicular to the magnetic field will measure higher amplitudes than those oriented parallel to the magnetic field.
• Since probe positioning is critical to repeatable measurements, a non-conductive fixture (wood, plastic) to position the device-under-test (DUT) and the probe can be used. Remember, position and orientation are very important. A few millimeters or a few degrees of rotation can cause a big difference in the measured amplitude of a given magnetic field.

Figure 4: Magnetic field probe orientation and position affect measurement amplitude.

Figure 5: SIGLENT SRF5030 near-field probe kit.

Figure 6: Probing a PCB using a SIGLENT SSA3X and SRF5030 probe.

Cables and interconnects can make very effective (and unintentional) antennas if they are not shielded/grounded correctly. Small currents flowing on the outside of the conductor can easily cause radiated emissions that can exceed the set EMC limits. A current clamp can be used with a spectrum analyser to provide insight into the cause of radiating cables/interconnects.

Current clamps operate on the same principle as magnetic loop probes. They can be purchased or made by wrapping a few rounds of wire around a ferrite clamp and epoxy a BNC connector as shown in figure 7. Simply attach the clamp to the cable to be tested, connect it to the spectrum analyser input, and configure the analyser for the frequency span of interest.

Figure 7: A handmade current clamp.

Here are some guidelines for probing:

• If in doubt, add an external attenuator to the RF input of the analyser before you start. Power cables or expected high-power applications can have signals that will damage the sensitive RF input of the analyser.
• Test all of the cables that could be connected to the DUT. This includes the power cord, USB, Ethernet, and any other possible connections (figure 8)

Figure 8: Measuring the RFI of a USB cable connected to a scope.

• Current clamps, especially handmade, are susceptible to picking up environmental RF that can skew or overwhelm the signals that you wish to measure. Connect and arrange all cables, probes, etc.. and then measure the environmental RF by simply keeping the DUT powered OFF. Then, compare it to measurements made with the DUT ON. It may also be a good idea to retest periodically to account for any environmental changes.

Figure 9: Traces of the environmental pickup from a current clamp (Yellow) and with the DUT powered ON (Pink).

• If you have a failed radiated emissions report, start by looking for the failed frequencies or for the first few harmonics of those frequencies.

SCANS AND EVALUATION

It is highly unlikely that data collected during probing will directly correlate to radiated emissions test performance. But, by observing the RF output of cables, switching power supplies, displays, and cutouts, you can have information that can lead to faster troubleshooting if you do happed to fail.

Here are optional techniques that can help provide more insight:

1. Most spectrum analysers do not have pre-selection filters. If you are using a spectrum analyser without pre-selection filters, the peaks you observe may not be real. Analysers without pre-selection filters can create false peaks due to out-of-band signals mixing with the observed signals.

You can test the validity of a peak by adding an external attenuator (3 or 10dB should do). Real peaks will fall by the amount of the attenuator. If the peak falls by more than the attenuator, it is likely to be a false peak. Make a note of the false peaks for comparison with your compliance test results. You can also use pre-selection filters or an EMI receiver, but these tend to be cost prohibitive for most quick testing.

Figure 10 below shows a typical peak confirmation test. The yellow trace was collected without an attenuator. The Pink trace was collected with a 10 dB attenuator added to the RF input of the analyser. In this case, the peaks drop the same amount as the added attenuation. This helps affirm that the peaks are likely real and not products of out-of-band signals.

Figure 10: Comparison of two scans using the marker table function of the SIGLENT SSA3000X spectrum analyser. The Yellow trace was collected without attenuation while the Pink trace was collected after adding a 10 dB external attenuator.

2. Many spectrum analysers have Max Hold trace types that will continuously hold the highest amplitudes of each frequency scan. You can enable a single trace as Clear Write to show active RF performance and enable a second trace as Max Hold. This allows you to compare changes in the DUT to the “worst case” data collected and “frozen” using Max Hold.

3. You can use markers and peak tables to clearly indicate peak frequencies and amplitudes, if available.

Figure 11: SSA3000X analyzer with Peak table and markers activated.

CONCLUSION

• Magnetic fields are produced by current flow. Use a magnetic (H) near-field probe to identify EM radiation near traces, wires, and ribbon/flex cables.
• Electric fields can be produced by current flow or static charge build up. Use an electric (E) near-field probe to identify EM radiation on metal surfaces like heat-sinks, enclosures, display bonding/edges, and slots/cutouts.
• Use current clamps to identify potential radiation and resonance from cables, wires, and interconnects
• Displays, cutouts/holes/seams in the chassis, ribbon cables, and communications ports/busses are the most likely cause of radiated emission failures.
• Use conductive tape or aluminium foil to cover areas of “leakage”, making sure that the covering is grounded. Rescan with the tape/foil in place to see if it has mitigated the EMI.
• Poorly terminated cables and interconnects also cause radiated issues
• Frequently measure the background effects by removing power from the device-under-test and monitor the output on an analyser. Note any changes and their potential effects on the measurements.

With a few simple tools, you can implement an in-house pre-compliance test process that will minimize the total development time for your products, lower the cost of design, and decrease the amount of testing on future products.

Spectrum analysers are useful tools for broadcast monitoring, RF component testing, and EMI troubleshooting. There are a number of common adjustments available with many modern analysers that can optimize performance for a particular application. In this application note, we will introduce resolution bandwidth (RBW) and video bandwidth (VBW) and how they affect measurements.

Resolution Bandwidth (RBW)

Bandwidth is defined as the span of frequencies that are the focus of a particular event. For example, the bandwidth of transmission signal is the span of frequencies that the transmission occupies. The bandwidth of a measurement defines the range of frequencies that were used for the measurement.

In spectrum analysis, the resolution bandwidth (RBW) is defined as the frequency span of the final filter that is applied to the input signal. Smaller RBWs provide finer frequency resolution and the ability to differentiate signals that have frequencies that are closer together.

Why not use the smallest RBW setting for all measurements?

Sweep Time.

Sweep time is the length of time it takes to sweep the detector from the start to the stop frequency. Here is the equation governing the sweep speed:

In this formula, the meaning of the first factor is the number of frequency selections under SPAN, each step is 1 / k of RBW, to ensure the accuracy of amplitude measurement. The second factor means that each selection The time required depends on the smaller value between RBW and video bandwidth (VBW). Usually when we do not focus on noise, the VBW can be set to a value greater than or equal to RBW.

The time equation is reduced to:

That is to say, the scanning time is proportional to SPAN and is inversely proportional to the square of RBW. This means that if the RBW is reduced by 100 times the scanning time will be increased by 10000 times in the same SPAN

Smaller RBWs also lower the noise floor, but they extended the sweep time for a given span of frequencies. Select a spectrum analyser that has a large number of RBW settings, especially on the lower frequency end. You may not use 10 Hz RBW often, but it is very useful when you do. Adjustment is easy. Simply adjust the RBW to provide the proper balance between speed and resolution for your application.

Figure 1 is the measurement of two signals separated by 20 kHz. The traces were collected using RBWs of 30 kHz (Blue), 10 kHz (Yellow), and 3 kHz (Pink). Observer that while the frequency of these two similar signal measurement power is completely unchanged, the signal separation is only clear when the RBW is less than the frequency difference between the signals.

Figure 1: Spectrum analyser display showing two signals at three different resolution bandwidth (RBW) settings.

Shape and Shape Factor

The shape and shape factor of the RBW filter can also be an important selection. Many analysers have an RBW filter that has a Gaussian shape and a shape factor determined at the 3 dB point. The RBW value is the bandpass frequency of the filter 3 dB below the peak response of the filter. Recall that 3 dB is equal to 50% of the maximum. This is also referred to as the filters Full Width Half Max (FWHM) value. The 3 dB Gaussian filter is acceptable for many measurements, but for Electromagnetic Compliance (EMC) related measurements, a filter defined at 6 dB may be required.

The shape factor of a filter is the ratio of the response at two attenuation values. Typically, the highest attenuation is measured at 60 dB down. The lower attenuation value is either or 6 dB down. It is a measure of the sharpness of the filter response. If the ratio is large, the filter is not very “sharp”. This indicates that the filter spreads out over a large frequency range. If the ratio is small, this indicates a skinnier filter shape and sharper roll off. This aids in rejecting more out-of-band signals because they don’t “bleed” over. Figure 2 shows how the shape factors for both the 3 dB and 6 dB points are calculated for a given filter. For spectrum analysers, the 3 and 6 dB shape factors are similar, but the 6 dB filter has a steeper curve and has higher out-of-band rejection.

Shape Factor @ 3/60dB = (F6 – F1)/(F4-F3)

Shape Factor @ 6/60dB = (F6 – F1)/(F5-F2)

Figure 2: Gaussian filter showing 3, 6, and 60 dB points and center Frequency (Fc).

Phase Noise

Another factor that affects the frequency resolution of an analyser is the phase noise. This is observed as a widening and increase in the noise amplitude near the center frequency of the signal (figure 3). It is caused by the random thermal fluctuations of the oscillator used as a timing reference in the spectrum analyser circuitry. These fluctuations cause the phase of the output clock signal to vary with time, very similar to jitter in a time-based system. This widening can cover up any small signals that may be near the frequency of interest. For meaningful measurements, select an instrument with lower phase noise than the signal source you are measuring.

Figure 3: Spectrum analyser display showing phase noise effects of an input signal

Video Bandwidth (VBW)

Another factor that affects the displayed trace quality of a spectrum analyser is the video bandwidth (VBW). Video filtering is a time-domain low-pass filter, mathematically equivalent to the mean or average. The main effect of the VBW filter is to smooth the trace and decrease noise.

Strictly speaking, the VBW does not change the measurement results. It will not affect the “frequency selection, peak detection” of the measurement process. The VBW filter is applied after the data has been collected, but before the screen displays the trace. As can be seen from figure 4 below, when the VBW is large, noise makes small signal observation difficult. If we reduce the VBW, the small signal becomes much more clearly.

Figure 2: Smoothing effect of random signal with different VBW

Conclusions

Modern spectrum analysers offer flexible measurement capabilities. Select an analyser that provides an adjustable RBW/VBW (lower is better), lower phase noise than the signal you are testing. Adjusting the RBW can provide lower noise floor and fine frequency resolution, but the sweep time will increase dramatically. For noisy signals, you can lower the VBW to help smooth the trace and make signal identification easier, but this will also increase sweep time. If you are performing EMI measurements, a 6dB sharper filter option is required to increase peak detection accuracy.

Spectrum analysers like the SIGLENT SSA3000X Series have a number of available detector selections that can help you observe specific signals of interest. This operating tip provides a brief description of the available detector types and suggested usage.

BASICS:

The SIGLENT SSA 3000X Series spectrum analysers normally display power versus frequency on a Cartesian coordinate system as shown below:

The horizontal span (which composes a frequency span when the instrument is configured to sweep/not in zero span mode) is comprised of 751 discrete samples (also called “bins”).

The analyser sweeps from the start frequency to the stop frequency and fills each frequency “bin” with an amplitude value determined by the detector selected for that particular sweep operation.

THE DETECTORS:

Positive Peak: For each trace point, a Positive Peak detector displays the maximum value of data sampled within the corresponding time interval.

Positive peak detectors are the most commonly used detector type. They are perfect for measuring the peak power of a signal as well as determining the “worst case” signal in EMI applications.

• Negative Peak: For each trace point, a Negative Peak detector displays the minimum value of data sampled within the corresponding time interval.

Negative peak detectors are rarely used, but can be helpful when comparing positive and negative peak values when looking for CW and Pulsed signals. CW signals will have less difference         between positive and negative peak values for a given frequency bin. A pulsed signal could have a significant difference

• Sample: For each trace point, a Sample detector displays the transient level corresponding to the central time point of the corresponding time interval.

This detector type is applicable to noise, noise-like signals, or small amplitude continuous wave (CW) signals that are near the noise floor of the analyser.

• Normal: Normal detector (also called rosenfell detector) displays the maximum value and the minimum value of the sample data segment in turn; namely for an odd-numbered data point (bin), the maximum value (positive peak) is displayed; for an even-numbered data point (bin), the minimum value is displayed. In this way, the amplitude variation range of the signal is clearly shown.

• Average: For each trace point, an Average detector displays the average value of data sampled within the corresponding time interval.

• Quasi-Peak (QP): This is a mathematically weighted form of the positive peak detector. For a single frequency point, the detector measures the peaks within the defined detector dwell time. These peaks are then weighted using a mathematical model that simulates a tank circuit that meets charge/discharge rates specified by CISPR (an international organization that governs electromagnetic effects). The measurement time for QPD is far longer than Peak Detector.

For a given signal, the QP detector values will never be greater than the positive peak results.

Our friends at Tekbox have a nice app note that provides details on performing conducted noise measurements using a SIGLENT SSA3000X Spectrum Analyser:

Tekbox Conducted Noise with a LISN Note

Spectrum analysers like the SIGLENT SSA3000X and SVA1000X series are extremely useful instruments that can provide invaluable insight into broadcast signal performance, transmitter troubleshooting, and interference as well as RF device characterization and EMC testing.

But, like other spectrum analysers, they are very sensitive and can be damaged easily, if the proper precautions are not followed.

The instruments have standard protection elements that includes a DC blocking capacitor and an automatic attenuator that help to prevent damage from low frequency and higher powered signals. There is even an audible and visible warning if the ADC (Analog-to-Digital Converter) overload.

In addition to this, adding external attenuation and protection can be useful in further preventing damage, especially when connecting to unknown sources such as antennas, transmitters, and LISNs.

One of our customers, Mr. Jeff Covelli (WA8SAJ), is a HAM (Amateur Radio Operator) who recently shared a very simple protection box that can be useful for keeping that sensitive front end functioning when connecting to unknown signal sources.

He uses it to help protect the input on his SIGLENT SDG1032X signal generator as well as his SSA3021X spectrum analyser:

And he has a pretty impressive setup:

And here are the details:

At some point, I hope to characterize this setup and provide S11, S21 information.. but, if your signal is unknown, this will add an additional layer of protection when measuring an unknown signal for the “first time”.