Feature Article
The Evolution of EMI Receivers
Reducing Time to Market and Visits to Testing Labs for New Products
By William Koerner
old dial radio on a wooden table with an aquamarine wall behind it
S

ince the first observations of interference from unknown events with AM radios in the early 1920s, the field of electromagnetic interference (EMI) has continued to evolve and involve more than AM radios. Today, any product with a power cord or that is battery-operated can and will generate electromagnetic fields. Electromagnetic compatibility (EMC) testing is required for any product that has electrical, digital, and/or radio components.

With the growth of the variety and volume of those products, the time to complete EMC testing typically takes longer, due to competition for lab time, and for surprises in tracking down short-burst or impulse-type emissions. The automotive industry, for example, requires exacting methodologies to measure all emissions accurately. Long test times impact test facility availability and potentially reduce the number of devices that are certified. It’s also easy to miss intermittent disturbance signals with conventional scans since an extended dwell time must occur at each frequency.

With the implementation of time domain functionality in EMI receivers and short-time FFT (STFFT) engines, EMI receivers now enable independent compliance test laboratories and in-house certification labs to shorten their overall test time, and for device manufacturers to quickly troubleshoot intermittent and impulse signals during design validation and pre-compliance testing.

This article will provide a short history of radiated EMI testing, a discussion on the evolution of EMI receiver designs, and a look at the newer time domain scan and FFT capabilities to meet EMI measurement requirements. We’ll also discuss EMC standards such as CISPR 16-1-1 and MIL-STD-461 and highlight how you can easily reduce receiver scan and test time from multiple hours to seconds. Finally, we’ll identify those areas where this makes the biggest difference, and when you may not need to consider adapting the newer technology.

History of Radiated Emissions Testing
How It All Began
In preparing to write this article, I wanted to do a little research on how the use of electromagnetic waves came about, and how it was discovered that it created some unintentional issues. In reading the “Empire Of the Air” by Tom Lewis,1 it was interesting to discover that, besides the well-known inventors and scientists like Marconi and Tesla, others such as Henry de Forrest, David Sarnoff, and Edwin Armstrong played major roles in the growth of the use of electromagnetic waves for wireless transmission in the late 1800s. Perhaps the first documented case of electromagnetic interference occurred in September 1901 when the competing wireless telegraphs of de Forrest and Marconi jammed each other during the International Yacht Races,2 resulting in neither inventor being able to report the results of the race.

Their work aligns with the discovery of solar activity creating “phantom telegraph operators,” in which radiated emissions are picked up by the long parallel transmission wires that generate telegraph output without telegraph input,3 as well as the growth of broadcasting and the use of electronic equipment in commercial and military applications. As a result of these developments, some sort of rules or regulations would become necessary to prevent radio interference or equipment malfunctions.

Beginning of Regulatory Oversight
In 1892, the German “Law of Telegraph” became the first law in the world that dealt with electromagnetic interference.4 Similar actions followed and the Comité International Spécial des Perturbations Radioélectriques (CISPR) was founded in 1934 as part of the International Electrotechnical Commission (IEC).5 That same year, the U.S. Communications Act was passed, establishing the Federal Communications Commission (FCC) in the United States, which took over the radio regulation functions of the previous Federal Radio Commission.6 One of its stated purposes was “for the purpose of promoting safety of life and property through the use of wire and radio communications.”

Early studies of interference tended to be called “noise,” primarily because their presence was identified as audio noise. Many attempts were made to quantify and measure this noise so that measurement techniques and limits could be established. But getting agreement with different measurements proved difficult, in part due to the concern being limited to an “annoyance factor,” that is, how the noise or interference “annoyed” the intended transmission or product use. The consensus was that high-repetition noise was more annoying than low-repetition noise. This ultimately led to the development of radio noise, objective sound meters, and quasi-peak detectors.7

Early studies of interference tended to be called “noise,” primarily because their presence was identified as audio noise. Many attempts were made to quantify and measure this noise so that measurement techniques and limits could be established.
Development of EMI Receivers
The First Tuned Receivers
The initial EMI receivers had the ability to tune and measure interference versus frequency. However, these receivers required manual tuning and also required the operator to read an analog meter for the amplitude and the frequency dial for the frequency. Due to the early technology, several instruments were typically needed to cover the complete frequency range required. In addition, the amplitude response of the receivers did not have a flat frequency response, so some sort of substitution technique was required for calibrated measurements.
Invention of the Superheterodyne Receiver
Manual tuning and inaccurate amplitude measurements made initial interference measurements tedious and time-consuming. That changed with the invention of the superheterodyne receiver by Edwin Armstrong in 1918 (his prototype is shown in Figure 1). His receiver allowed for tuning and receiving signals at even higher frequencies than before and has served as the foundation for receiver designs even today.
Prototype superheterodyne receiver
Figure 1: Prototype superheterodyne receiver built at Armstrong’s Signal Corp laboratory in Paris, 191812
With the invention of the superheterodyne receiver, wireless telegraphy was suddenly not just the only opportunity. In 1919, the Radio Corporation of America (more commonly known as RCA) was incorporated, combining the patents of Marconi, de Forrest, Armstrong, and General Electric to focus on the business of radio, or voice broadcasting. The big breakthrough for RCA was the broadcast of the heavyweight fight between Jack Dempsey and Georges Carpentier. In a first of its kind, the fight was broadcast to over 300,000 people across the U.S. That created the demand for “radio” for personal use, and a whole new industry was born.8
EMI Receiver or Spectrum Analyzer?
With the initial issue of dealing with radio “noise,” the development of broadcast radio, and the introduction of radar during World War II, the need to analyze high-frequency signals for either content or noise was imperative to ensure the systems worked as expected and, even more importantly, met the emission standards established by the FCC and CISPR. This drove the need for not just a noise receiver or EMI receiver, but also a spectrum analyzer.

Figure 2 shows the typical architecture for a traditional swept spectrum analyzer. Developments of stable local oscillators that could be swept allowed for fast, continuous tuning and measurements across the defined frequency range. Note that there is typically a pre‑selector or low-pass filter before the first mixer. This allows for a lower noise floor for the measurement and prevents broadband signals from overdriving the mixer.

diagram of a traditional swept spectrum analyzer architecture
Figure 2: Traditional swept spectrum analyzer architecture
The EMI receiver has evolved with a very similar architecture, but there are subtle differences due to the nature of the signals to be analyzed. Figure 3 shows the RF front end for a typical EMI receiver.
diagram of a traditional EMI receiver front end
Figure 3: Traditional EMI receiver front end
One major difference between the spectrum analyzer and EMI receiver is the dual signal paths. The low band path in Figure 3 is for the lower frequencies (< 3.6 GHz in this example), which has unique pre-selectors that are a combination of bandpass filters for the lower frequencies. This allows for wider bandwidth than that available in the high band path, but also prevents broadband impulsive noise from overloading the first mixer. It also allows for less input attenuation to provide the dynamic range to measure the CISPR pulse and meets CISPR 16 requirements.

The high band path has a Yttrium Iron Garnet (YIG) swept preselector to protect the first mixer and resembles the traditional swept spectrum analyzer architecture. The bandwidth of the YIG preselector is much narrower than the RF preselector in the low band path, which ensures the dynamic range required at the higher frequencies.

Analog vs Digital IF Sections
Figure 2 shows the traditional analog IF section in a spectrum analyzer and is very similar to what was in EMI receivers as well. Perhaps the biggest change in either the spectrum analyzer or the EMI receiver was the introduction of digital IF sections. I remember in the 1980s that analog-to-digital converters (ADCs) were becoming prominent, with the idea of getting the signal of interest to digital as close to the input as possible. This has a great impact on measurement speed, accuracy, and the ability to measure complex signals using advanced digital signal processing (DSP) techniques.

Figure 4 shows the analog IF section replaced with a digital section. After the down conversion, the signal is converted to a digital value, which is a digital amplitude value. The term “digital IF” describes the digital processing that replaces analog IF processing found in traditional receivers and spectrum analyzers.9

diagram of a digital IF section architecture
Figure 4: Digital IF section architecture
At this stage, any additional processing is done with DSPs and mathematical functions, mainly some version of fast Fourier transform (FFT). Resolution bandwidths (RBWs), video filtering, averaging, and detection are all done with the desired mathematical function. Digitally implemented RBWs offer both improvements and improved filter performance, as well as tighter filter shape factors. Most noticeably, digital RBWs allow for much faster sweep times, as there is no charge time for the filter. Digital IF gain can provide extremely accurate reference levels. Digital logarithmic correction factors reduce measurement uncertainty associated with analog log amplifiers.

Table 1 shows a typical comparison of the differences in amplitude uncertainties between digital and analog IF sections. The data represented here was collected by surveying receiver and spectrum analyzer specification guides.10

table comparing the amplitude uncertainties with digital and analog IF architectures
Table 1: Comparison of amplitude uncertainties with digital and analog IF architectures
To resolve the issue between frequency range and resolution, the use of short-time fast Fourier transform (STFFT) engines can be used. This involves overlapping the FFTs by overlapping the captured time domain signal in the different FFT frames.
FFT Implementations
With the implementation of ADCs in the IF section and careful selection of RF preselection in the front-end, it is possible to measure broadband signals instantaneously. Current ADCs support multi-GHz sample rates and at least 14 bits, making for dynamic ranges of >80 dB. Combined with the RF preselection, instantaneous bandwidths of hundreds of MHz, possibly even 1 GHz, are possible.
ADC with FFT
While having an ADC with a fast sampling rate is required to capture a broadband signal, it comes with a trade-off between the frequency range of the captured signal and the frequency resolution of the acquired signal. In order to support the RBW requirements for CISPR standards, the number of samples is very large and may not be commercially feasible to implement.
Overlapping FFTs
To resolve the issue between frequency range and resolution, the use of short-time fast Fourier transform (STFFT) engines can be used. This involves overlapping the FFTs by overlapping the captured time domain signal in the different FFT frames. For example, if the frame length is set to 2048 points, the time samples 1-2048 are collected in the first frame, time samples 1024-3072 are collected in the second frame, etc. This example shows an overlap of 50%.

Figure 5 shows an example of measuring an impulse signal with 50% overlap. In this scenario, the second FFT frame (bottom row) gives a full response because the impulsive envelope occurred at the time of maximum weighting.11 With 90% overlap, the worst-case error from the windowing occurs when the envelope peak is displaced from the weighting peak by 5% of the FFT duration.

graph showing how measurements made with FFTs at 50% overlap in the time domain
Figure 5: Measurements made with FFTs at 50% overlap in the time domain
diagram showing the comparison of resolution bandwidth and FFT acquisition bandwidths
Figure 6: Comparison of resolution bandwidth and FFT acquisition bandwidths
An additional advantage is that the overlapping FFTs allow you to capture all data in the acquisition bandwidth in one dwell time, compared to having to stop at each frequency point for the dwell time for the traditional swept or stepped frequency domain scan.
Should You Use Overlapping FFTs?
While overlapping FFTs allows you to analyze broadband data instantaneously, it does come with a higher cost than traditional swept frequency receivers. Given that, here are several scenarios where the use of overlapping FFTs is beneficial:
  1. Faster measurement times: Because the overlapping FFTs allow you to capture broadband signals in one dwell time, the measurement time is much faster than the traditional swept frequency receivers. Table 2 shows the dramatic difference in measurement times for a typical automated broadband noise test.
  2. Impulsive noise detection: If your equipment under test (EUT) has impulse noise contributors, or if you need to investigate if it has those characteristics, then the FFT capability may be the only way to capture that phenomenon.

    Figure 7 shows an example on the left of an impulsive signal (in this case, a pulsed comb generator) where the traditional stepped scan is only able to capture one of the frequencies, vs. the overlapped FFT on the right, where all frequencies are captured on one acquisition.

  1. Unique EUT characteristics: If your EUT has unique characteristics (for example, it is not able to be left in a powered on condition for a complete test, or it has motors or switches that operate normally during the use of the product such as pumps or motor drives in a washing machine), then you would benefit from FFTs being able to capture the emissions when the EUT is exhibiting one of those emissions.
  2. Exhaustive pre-compliance validation: If you wish to exhaustively test your EUT for pre-compliance for any impulsive surprises, then the overlapping FFTs will allow you to detect those before you send it to the test lab for final compliance tests.
table showing measurement time comparisons between stepped scan and wideband FFT EMI receivers
Table 2: Measurement time comparisons between stepped scan and wideband FFT EMI receivers
stepped scan vs FFT scan for impulse signal
Figure 7: Stepped scan vs FFT scan for impulse signal
Summary
FFT implementations in EMI receivers offer many benefits, namely much faster measurement times and the ability to capture intermittent/impulsive emissions. Their use for design validation and pre‑compliance testing is very useful in identifying unique and intermittent emissions that may be missed using conventional EMI receivers. Their applicability for CISPR compliance testing, while defined in CISPR 16-1-1:2010-06 edition, may not mean you can use them for full compliance testing. Because of the limitations on sampling rates and memory depth, the ability to measure low PRF impulses is a challenge.

Some CISPR standards for specific device types may not allow for the use of FFTs for compliance measurements. You will need to review the CISPR standard applicable to your EUT to determine if that edition is referenced in your standard to decide whether FFT measurements are allowed for full compliance testing.

Endnotes
  1. Thomas S. W. Lewis, Empire of the Air, 1991, Harper Collins Publishers.
  2. Ibid., page 40.
  3. Gubisch, R., Holz, B., “The Engineer’s Guide to Global EMC Requirements: 2007 Edition,” pp. 2, Intertek, 2007.
  4. Mohd Fahmi, Abdul Rahim, Agilent Technologies, Inc., “Evolution and Trends of EMI Receiver,” 2013 IEEE International RF and Microwave Conference, Penang, Malaysia.
  5. CISPR, Wikipedia, https://en.wikipedia.org/wiki/CISPR.
  6. Federal Communications Commission, Wikipedia, https://en.wikipedia.org/wiki/Federal_Communications_Commission.
  7. Burrill, C.M., “Progress in the Development of Instruments for Measuring Radio Noise”, Proceedings of the IRE, vol. 29, issue 8 pp. 433‑442, August 1941.
  8. Houck, H. W., “The Armstrong Super-Autodyne Amplifier, part 1,” Radio Amateur News, Experimenter Publishing Co., New York, vol. 1, no. 8, February 1920, p. 403.
  9. Thomas S. W. Lewis, Empire of the Air, 1991, Harper Collins Publishers, Chapter 6.
  10. Keysight Technologies, “Enhance EMC Testing with Digital IF – Application Note,” June 18, 2015.
  11. Ibid.
  12. Keysight Technologies, “Boost EMC Test Throughput with Accelerated Time Domain Scan,” July 14, 2023.
Share this story:
William Koerner
William (Bill) Koerner is a Senior Application Engineer with Keysight Technologies, where he focuses on supporting the company’s microwave compliance testing solutions, including wireless regulatory (Wi-Fi, Bluetooth®, Zigbee), EMI receivers, and co-existence testing for medical devices. Koerner is also Keysight’s representative to the FCC’s Telecommunications Certification Body (TCB), the ETSI Broadband Radio Access Networks (BRAN) working committee, and several Wi-Fi Alliance TG working groups. Koerner got his start at the original Hewlett-Packard as a Microwave Systems Engineer in the mid-1980s. He can be reached at bill.koerner@keysight.com.
Back to Issue