You need to test, we're here to help.

You need to test, we're here to help.

01 February 2021

Situational Awareness: RF Noise in the Lab

Fig. 1. Time domain (top) and spectral (bottom) views of signal shown in SPECTRUM-1 on a WaveSurfer 4000HD.
Fig. 1. Time domain (top) and spectral (bottom) views of
s
ignal shown in SPECTRUM-1 on a WaveSurfer 4000HD. 
Laboratories have multiple sources of RF that can affect measurements, like computers, cell phones, routers, local radio and TV stations, even a nearby airport. Knowing the RF background of your lab is another part of Situational Awareness. Only by knowing the background can you know what is actually due to the device under test.

One way to read RF is through spectral analysis of Fourier transforms (DFT and FFT). FFTs take a time domain view of a signal (e.g, amplitude versus time trace) and change it into a spectrum of amplitude plotted as a function of frequency. Frequency spectrums are great for observing signals than are asynchronous with the process being measured. They have a lower noise floor and offer better dynamic range than do time domain plots.  Consider the views of the same signal shown in the time domain and frequency domain in Fig. 1.

The time domain plot shows a noisy repetitive waveform, but the spectral view shows a structure. Note that the dynamic range of the spectrum is over 60 dB. The spectrum shows the harmonic structure, with primary elements being the harmonics of the 20.063 kHz fundamental. There is also a non-harmonic, low frequency component at 20 Hz. The specificity of the view is a primary advantage of spectral analysis.

A good deal of preliminary noise detection can be done by measuring near-field emissions about the circuit board. Near-field means that we're measuring closer than a wavelength, or a fraction of a wavelength, to the device. Commonly used tools for RF “sniffing” are the precision loop, the Hall-Effect current clamp or current probes, and the common 10x passive voltage probe, all shown in Fig. 2. 

Fig. 2. Three near-field probing options: the 10x probe loop, the precision loop, and the Hall-Effect current clamp.
Fig. 2. Three near-field probing options:
the 10x probe loop, the precision loop, and the Hall-Effect current clamp.

The precision loop is just a wire loop with insulation around the outside to protect it. It's calibrated so the voltage that we measure is a direct measure of the di/dt, which is a direct measure of the magnetic field flux. It can be calibrated back to magnetic field strength. If you don't have one of these, then any loop will do. You can use a commonplace 10x passive voltage probe loop as a handy way of doing quick, relative, measurements of near-field emissions. The current probe or current clamp measures the common currents going through the probe cable, which is, of course, another source of radiated emissions.

Let’s look at an example of an RF measurement using a 10x passive voltage probe. The homemade “RF probe” is simply a current pickup loop formed by clipping the standard probe ground wire to the center conductor of the probe, as shown in Fig. 2.

Fig. 3. Time domain and spectral components due to the oscilloscope itself (C2 in magenta). All the spectral components are below -98 dBm.
Fig. 3. Time domain and spectral components due to the
oscilloscope itself (C2 in magenta). All the spectral
components are below -98 dBm.
Begin by configuring a spectrum using the oscilloscope’s standard FFT math function, or ideally, the SPECTRUM-1 or SPECTRUM-PRO-2R Spectrum Analyzer software. 

With the probe disconnected from the oscilloscope input, we can view the background noise due to the oscilloscope itself, the “noise floor” (Fig 3). The oscilloscope-only spectrum has a peak response of under -98 dBm.

Fig. 4. Time domain signal and frequency spectra  for the lab environment “pickup.”
Fig. 4. Time domain signal and frequency spectra 
for the lab environment “pickup.”
Now, plug in the probe, and what you will see is the RF pickup in your lab. In the RF spectrum shown in Fig. 4, there is a broad range of narrow band spikes above 400 kHz, with the peak response at about -58 dBm. 

Some of the frequency components you see will be associated with particular RF sources in the environment operating at that frequency. Watching the peaks increase as you move nearer to a particular emitter will help identify it as the source of a particular frequency component. For example, moving the probe near a laptop computer, we see an increase in the radiated RF signal levels (Fig. 5). The spectral peaks above 400 kHz are now exceeding the-58 dBm peak probed before.

Fig. 5. “Sniffing” near a laptop shows a significant increase in RF noise. The spectral lines above 400 kHz are now peaking above 58 dBm.
Fig. 5. “Sniffing” near a laptop shows a significant increase in
RF noise. The spectral lines above 400 kHz are now
peaking above 58 dBm.
You may see some of these peaks again and again in measurements. Understanding their origin will help you know what is “real” vs. background when you probe the DUT.

With the SPECTRUM-PRO-2R software, you can save the background RF profile and use the Background Noise Removal feature to eliminate it from spectral measurements of your DUT.

Watch Dr. Eric Bogatin demonstrate this concept in the on-demand webinar, Real-time Spectral Analysis of Power Rails.

Also see:

Four Measurement Best Practices

Situational Awareness: Testing Oscilloscope Outer Limits

Situational Awareness: The Impact of the Interconnect


No comments: