Wireless Testing of Packaged FMCW Radars using The Wireless Connector®

Complexities in Testing FMCW Radars Over the Air

Frequency-modulated continuous-wave (FMCW) radars can be found in multiple different sensing applications, such as automotive radar, monitoring of vital signs, and bird or drone detection. Measuring low-level metrics of FMCW radar modules that have the antennas integrated in a PCB or in an IC package is not a trivial task. These radar modules generate fast time-varying signals (chirps), which complicates the measurement procedure and leads to additional requirements of the instrumentation and measurement environment. Especially over the course of such a wideband chirp, the performance of individual components changes. By measuring the FMCW radar only at one angle, which is standard practice in an anechoic chamber (AC), it is not immediately clear where possible discrepancies between simulation and measurement come from.

One of the issues is the angle dependency of the transmitter antenna. If the effective isotropic radiated power (EIRP) measured at one angle turns out to be lower compared to simulated values, it can be attributed to a multitude of reasons (RFIC, antenna, their interaction, and/or alignment issues), typically leaving a design team in the dark. A way to overcome this is to first measure the total radiated power (TRP) to verify that the RFIC is operating as expected and eliminate any misalignments. In an AC, this requires a spherical scan to be performed, which is a very time-consuming operation. An alternative approach is to use a reverberation chamber (RC).

In this application note, it is explained how an RC can be used to verify the working of an integrated FMCW radar module. R&D-type metrics, such as the power spectral density (PSD) and the spectrogram, are discussed, as well as production-type metrics, such as the TRP of one chirp and its ramp rate.

The Wireless Connector®

When a transmitting antenna is placed in an RC, the RC's operation can be viewed as taking the integral over the antenna's radiation pattern. Hence, The Wireless Connector® acts as an over-the-air connector. This operation doesn’t require any mechanical movement of the device-under-test (DUT) or reference antenna, leading to fewer moving parts and making the overall setup easier. In Figure 1, The Wireless Connector® model RC-01-0202 is shown. This table-top RC is operational between 5 and 170 GHz and supports smooth integration with state-of-the-art instrumentation from vendors like Keysight, Rohde & Schwarz and Tektronix.

Figure 1: The Wireless Connector®, model RC-01-0202, operational from 5 to 170 GHz.

Measurement Setup

In Figure 2, a schematic of the measurement setup is shown. The DUT is an FMCW radar evaluation board of Texas Instruments (IWR6843ISK-ODS REV C) and is configured such that it should create a linear chirp from 60 to 64 GHz. The chirp duration is 40 μs, meaning that the ramp rate is 100 MHz/μs. The evaluation board is set to repeat the chirp every 50 ms. The DUT has three transmitting antennas that are tightly integrated with the radar RFIC, meaning that the RF performance of the RFIC needs to be measured over the air.

Figure 2: Schematic of the measurement setup.

The RC used to conduct the measurements is shown in Figure 3. In the photo, two mechanical mode stirrers can be identified. These stirrers are used to create different electromagnetic environments for each measurement acquisition done by the oscilloscope. The working principle of the RC is such that, on average, the power transfer of the DUT to the receiving (Rx) antenna is constant. This allows the user to obtain repeatable measurement results, even when the DUT is placed at a different location in the chamber. Therefore, troublesome alignment procedures can be omitted, and don’t require the DUT or Rx antenna to make any mechanical movements.

Figure 3: The DUT inside The Wireless Connector®, model RC-01-0202.

The RF subsystem consists of a mixer, signal generator (SG), power amplifier (PA), and a low-pass filter (LPF). The SG provides a 59.6 GHz LO signal to the mixer, such that the 60 to 64 GHz chirp is mixed down to a chirp from 0.4 to 4.4 GHz. The PA amplifies the signal such that its amplitude is large enough to trigger reliably on the incoming signal. The filter's bandpass is from DC to 4.4 GHz and is used to mitigate higher-order mixing products. A vector network analyzer (VNA) has been used to measure the conversion gain from RF in the chamber to IF at the oscilloscope’s port. This conversion gain is used to calibrate the system such that we can present absolute measurements in this application note. 

The dynamic time-domain signal generated by the DUT is recorded by a fast-sampling oscilloscope. The device used in this setup is operational up to 6 GHz and has a maximum sampling speed of 16 GS/s, such that it can reliably sample the down-mixed chirp. The oscilloscope triggers on a positive rise voltage of 10 mV, which means that no trigger or synchronization signals are required between the DUT and oscilloscope.

Measurement Results

In the current measurement campaign, the DUT is repositioned at different locations in the RC. Between locations, the DUT is moved more than 5 cm (or 10 wavelengths at 60 GHz) and minimally rotated by 90°. At each DUT position 150 oscilloscope acquisitions are captured, and for each acquisition, the stirrers have a different configuration.

In Figure 4, the power spectral density of the radiated chirp is shown. The VNA-based frequency-domain calibration of the RC and its subcomponents is taken into account in this result. A few aspects can be observed from this figure. First of all, a ripple of a maximum of 1.5 dB (peak-to-peak) can be observed. Additionally, it can be seen in the figure that the power from 60.1 GHz to 60.9 GHz is on average, about 1 dB lower compared to the rest of the chirp. The ripple and the drop in power could be caused by the frequency-dependency of the RFIC, of the antenna, or of the combination of the two due to PA-loading effects.

It is clear to see that the chirp starts at 60.1 GHz, and stops at 64.0 GHz, meaning that the chirp BW is 3.9 GHz. Based on the acquired data, a TRP during one chirp of 9.8 dBm was calculated. In the datasheet of the evaluation board, a maximum output power of 10 dBm is listed, which is very well in line with the measured TRP of one chirp.

Figure 4: Power spectral density during the acquisition interval.

In Figure 5, the spectrogram of the FMCW chirp is shown. In this figure one can clearly see that the chirp is increasing in frequency over the span of several tens of microseconds. The down-mixed chirp begins at a frequency of 0.5 GHz and ends at 4.4 GHz, which corresponds to the 60.1 to 64.0 GHz also observed in Figure 4. Furthermore, the chirp duration is 39 μs, meaning that the ramp rate is on the order of 100 MHz/μs.

Figure 5: Spectrogram of the DUT’s chirp. Only the spectrogram of DUT position 1 is shown.

Based on the PSD and the spectrogram, the TRP of one chirp and the exact ramp rate can be determined. In Table 1, these results are summarized for the 3 different positions of the DUT in the chamber for 3 different truncations of the dataset. It can be seen that with as little as 10 acquisitions, a repeatable output power and ramp rate can be found that is consistent for different locations of the DUT. For reference, the total acquisition time is listed in the table and includes both the measurement itself and the post-processing time. In Figure 6, the resulting PSD for the 3 positions is shown, and the 2-σ uncertainty between the resulting PSDs is determined to be 0.4 dB. Note that in this repeatability measurement, the possible drift of the DUT and/or measurement equipment is included as well. This implies that alignment issues will, by default, be rather low in an RC, in contrast to ACs where uncertainties to alignment can result in significant uncertainties.

Table 1: Measured total radiated power of one single chirp and ramp rate versus the number of acquisitions for different positions in the RC.

Figure 6: PSD of the DUT for three different positions in the RC.

Conclusion

The Wireless Connector® acts as a field integrator of a transmitting antenna, therefore directly measuring the power radiated in all directions. This property proves to be useful for fast time-varying signals, such as FMCW radar modules. It is shown that a spectrogram or PSD can be readily measured, and parameters like TRP of the chirp, chirp BW, chirp duration and ramp rate can be easily derived from the available data. In general, the setup time is short since no alignment is required in an RC. Both the receiving and transmitting antenna remain stationary, meaning that there are not many moving parts needed in an RC. These properties make it a very suitable alternative to ACs for both R&D and production-level testing of FMCW radar modules.

The Wireless Connector® is a registered trademark of ANTENNEX Holding B.V. in the EU.