Optimising hardware design with near-field scanners

Article By : Arturo Mediano

A scanner is essentially a series of near-field probes placed in a grid to produce an image of a board's emissions that's more consistent and repetitive than by manually scanning a board with probes.

I love near field probes because they let me "see" magnetic and electric fields with an oscilloscope or with a spectrum analyser. They let locate sources of emissions in board, cables, and systems. Near-field scanners also let you see emissions, particularly all over a board. That's hard to do with a single probe.

There are several EMI/EMC scanners on the market today such as those from EMSCAN, DETECTUS, and Api, and others. A scanner is essentially a series of near-field probes placed in a grid. Thus, it can produce an image of a board's emissions that's more consistent and repetitive than you can get by manually scanning a board with probes.

The EMxpert scanner from EMSCAN is one such scanner. I use it in my lab. Here's an example of how I used it to evaluate a decoupling network for a course. Figure 1 shows a scanner with a scan area 21.8cm x 31.6cm) scanning a PCB under test.

[EMI scanner 01]
__Figure 1:__ *A PCB under test on top of the near field scanner. (Photo by A. Mediano)*

The scanner consists of thousands of loops spaced so that it provides resolution of less than 1mm. Frequency range goes from 50kHz to 8GHz, depending on the model. The loop antennas are sensitive down to -135dBm and a high-speed electronic switching system provides real-time analysis in less than 1s.

EMI scanners let you quickly analyse and compare design iterations and optimise hardware design. I use them for troubleshooting and for teaching. Here, I'll use it to demonstrate how a decoupling network can reduce emissions from a board.

Consider, for example, a typical circuit with a 24MHz clock (Figure 2). The board containing this circuit also has a decoupling circuit. The +5V power comes from a USB connector. The board includes an SMD fuse, a small LED for visual feedback, and a couple of decoupling capacitors. Load for the clock is a 50Ω resistor.

[EMI scanner 02]
__Figure 2:__ *Basic schematic for the decoupling network of the IC clock.*

A transient current (is) is required from the power supply to operate the IC. Usually, the high frequency content of that current (harmonics) is the source of many conducted and radiated EMI problems.

A decoupling network (usually surface-mount capacitors and ferrites) is used to minimise the high-frequency components going through the power-supply. If the decoupling circuit is working as expected, current iPSU will be reduced to DC because transients will take the path through the decoupling capacitor (iC) to power return. With two jumpers, we can enable/disable the decoupling network and evaluate its effectiveness. In Figure 3, the VCC trace is on the top layer. GND trace (no ground plane) is on the bottom layer.

[EMI scanner 03]
__Figure 3:__ *General view of PCB for our decoupling example.*

A spectral scan lets us identify signals from the board, which often come from oscillators and clocks. Signals may be parasitic oscillations or ringing, which are difficult to prevent. With the spectral scan, we can measure any signal from the board.

The spectral scan in Figure 4 lets you identify the harmonics of the 24MHz clock's transient currents and some EMI from the environment, including FM broadcasting signals (88MHz to 108MHz).

[EMI scanner 04]
__Figure 4:__ *A spectral scan clearly shows emissions from the PCB.*

With a spectral scan and spatial scan, you can identify the current path for that signal, critical information if you want to minimise EMI/EMC and SI problems. Figure 5 shows the emissions from the board when the decoupling network is not part of the circuit. That is, the switch across the ferrite bead is closed and the switch in series with the capacitor is open. The result: a big loop that produces emissions over much of the board. The big loop can create distortion for the clock signal, radiated emissions of the high-frequency harmonics, crosstalk with other boards or cables, and injected noise in the power supply or cables.

[EMI scanner 05]
__Figure 5:__ *Spectral and spatial scan without decoupling network. The path for current is clearly identified in a big loop (maximum levels are read as dBµV in red colour).*

With the decoupling network enabled, the loop size is much smaller (transients take the path of the decoupling capacitor) and EMI currents are contained in the area closer to the clock IC (Figure 6). Note maximum levels in red colour are now more than 16dB below the previous measurement (with the same scale, this new plot would be basically blue).

[EMI scanner 06]
__Figure 6:__ *With the decoupling circuit enabled, emissions are greatly reduced.*

A typical question when doing the review of a product is: "Did you use a decoupling capacitor?" Usually, the response is something like: “Of course, I have a 100nF capacitor.”

That's not enough.

Sometimes, you have a capacitor (or decoupling circuit) in your system but there is no effective decoupling because terminal impedances don't match the topology you have chosen, the capacitor technology/value isn't correct, or parasitic effects in the layout and package are dominant. With a near field scan, you can detect how your decoupling system is really working.

Leave a comment