A Broadband Solar Radio Dynamic Spectrometer Working in the Millimeter-wave Band

A Cassegrain antenna-feed system is designed to be the receiving unit of the observation system. The antenna is designed with left-hand circular polarization (LCP) to receive the one polarized component during the bursts. The primary reflector is developed with carbon fiber to reduce its weight, as a result of which the load of the track platform could also be reduced by a large extent. The diameter of the dish is designed to be 80 cm to improve the signal-to-noise ratio of the system. The half-power beamwidth (HPBW) of the main lobe is ∼0.8° in the frequency range of 35–40 GHz with a sidelobe suppression of >20 dB, and the gain of the antenna is about 45 dBi around 35.25 GHz (Figure 2). The feed is mounted just under the secondary reflective surface and acquires a solar radio signal after a two-step reflection. The HPBW of the feed is about 16° (Figure 2 (a)).

Zoom In Zoom Out Reset image size

The AFE lies behind the acquisition module. This module is used to process the analog signal received by the antenna-feed system to the working domain of the following digital receiver; the corresponding diagram is shown in Figure 3. The received signal is amplified, filtered, and downconverted to the IF of 937.5 ± 250 MHz. We note that a two-step downconversion is used to eliminate the deterioration resulting from intermodulation components with the first-level IF of 9 GHz ± 500 MHz and the second-level IF of 937.5 MHz ± 250 MHz. The filters follow the amplifier and mixer to suppress the image frequency and the intermodulation frequency, respectively.

Zoom In Zoom Out Reset image size

The digital receiver converts the analog signal to a digital one, and processes these signals to the final dynamic spectrum. This receiver is composed of a motherboard and a daughterboard, and the two boards are connected via field-programmable gate array (FPGA) Mezzanine Card high-pin connector (Figure 4).

Zoom In Zoom Out Reset image size

In this digital receiver, the IF analog signals from the AFE are first converted into digital ones by an analog-to-digital converter (ADC) on the daughterboard, and then transmitted to the motherboard by a serial protocol. We note that the ADC is the critical device for signal conversion, and it determines the performance of the digital part of the system. Therefore, AD9691 (14 bit 1.25 GSPS) is chosen, with the consideration of the demanded application in a wide frequency domain, and compatible with a wide frequency range (as shown in Figure 5). The input frequency range can be better adapted by selectively soldering resistors and capacitors of the different values given in Table 3 (Analog Devices 2015; Shi et al. 2020; Virkler et al. 2020).

Zoom In Zoom Out Reset image size

The motherboard is compatible with the FPGA XC7K410T and the DSP TMS320C6678 completes the signal processing and transmits the data to the host computer. A series of signal processing tasks are finished in the FPGA, e.g., receiving data, deframing, and 8k-point FFT. The FPGA is configured with the same JESD204B protocol, the same number of lanes (eight), and at the same data rates (6.25 Gbps per lane) as the ADC to achieve handshaking. Data deframing is done by calling the FPGA intellectual property core to separate the data of the two ADCs from the JESD204B serial data. As illustrated in Figure 6, the base-2 time extraction butterfly operation is used to increase the speed of the operation (Yan et al. 2021). Packing, timestamping, and uploading are accomplished in the DSP.

Zoom In Zoom Out Reset image size

We know that the lower the noise figure, the higher the sensitivity of the system, so the noise figure should be reduced if possible. At the same time, a reasonable system gain will allow a suitable dynamic range for solar activities.

The system noise figure is determined by the amplifier cascade principle

$$\begin{eqnarray}&&{F}_{\mathrm{sys}}={F}_{1}+\left[\displaystyle \frac{\left({F}_{2}-1\right)}{{G}_{1}}\right]+\left[\displaystyle \frac{\left({F}_{3}-1\right)}{{G}_{1}{G}_{2}}\right]+\cdots ,\end{eqnarray} \tag{ 1 }$$

where F_sys is the noise figure of the system, F1 (F2, F3) are the noise figures of the first-stage (second-stage, third-stage) amplifier contribution, respectively, and G1 (G2) is the gain of the first stage (second stage). As can be seen from Equation (1), the system noise figure is mainly determined by the first-stage amplifier, and the noise figure can be improved by choosing a first-stage device with a lower noise figure and large gain. In the developed system, we choose SBL-2734034025-KF as the first-stage low-noise amplifier (LNA), and the noise figure and gain are ∼300 K and ∼40 dB from 35–40 GHz at room temperature, respectively (see Figure 7).

Zoom In Zoom Out Reset image size

Reasonable design for system gain requires maintaining adequate system dynamic range while avoiding entering the nonlinear region. Based on this, we can estimate the received power in the AFE according to the formula

$$\begin{eqnarray}&&\displaystyle \frac{1}{2}{A}_{e}{S}_{\mathrm{Sun}}B={{kT}}_{a}B\Rightarrow {T}_{a}=\displaystyle \frac{{A}_{e}{S}_{\mathrm{Sun}}}{2k},\end{eqnarray} \tag{ 2 }$$

where A_e is the effective area of the antenna, S_Sun is the spectral flux density corresponding to the observation frequency, T_a is the noise temperature of the antenna, k is the Boltzmann constant, and B is the observation bandwidth. A_e is decided by

$$\begin{eqnarray}&&{A}_{e}=\displaystyle \frac{{G}_{a}{\lambda }^{2}}{4\pi },\end{eqnarray} \tag{ 3 }$$

where G_a is the gain of the antenna and λ is the wavelength of the observation frequency. So the total power P_sys in bandwidth B is given as

$$\begin{eqnarray}&&{P}_{\mathrm{sys}}=k\left({T}_{a}+{T}_{\mathrm{sys}}\right){{BG}}_{\mathrm{sys}}.\end{eqnarray} \tag{ 4 }$$

Taking 35.25 GHz as an example, the total power P_sys is about −7.2 dBm/500 MHz, considering the system gain G_sys of about 72 dB @ 35.25 GHz, the antenna gain (G_a ) of about 45 dBi, the flux density of about 2208 SFU (T_a is about 1458 K), and T_sys ∼300 K. P_sys is lower than the P1dB of the amplifier and mixer. The P1dB of the amplifier and mixer is usually around +10 dBm and +5 dBm at 35 GHz respectively. The differential full-scale input power of the ADC is +7.9 dBm. The gain at each level of the system shows that the dynamic range is limited by the ADC, so the dynamic range of about 15 dB is sufficient for the expected level of solar activity at 35–40 GHz.

As the 5 GHz bandwidth is channelized for 1 GHz, so there are five switching times (two IFs of level 2, instantaneous bandwidth 1 GHz, and 10 channels in total) to cover the entire bandwidth. The frequency-switching time of the AFE is critical to the time resolution of the system. The frequency-switching time is less than 140 ns, and an RS422 protocol is used to achieve remote control. Such a short frequency-switching time can ensure that the channel switching takes almost no time to achieve high time resolution.

While observing the 35–40 GHz dynamic spectrum, the system also observes the solar radio flux density at the center frequency of the 10 channels. Before the observation, the linearity of the 10 channels needs to be confirmed.

Accurate linearity can be derived by measuring an ideal blackbody radiation source filled with liquid nitrogen or a high-temperature blackbody radiation source to cover the antenna, and the temperature injected into the system by adjusting the attenuation of the attenuator behind the antenna (Wang et al. 2014). The temperature injected is given as

$$\begin{eqnarray}&&{T}_{{IN}}=\displaystyle \frac{{T}_{N}}{L}+\left(1-\displaystyle \frac{1}{L}\right)\times {T}_{0},\end{eqnarray} \tag{ 5 }$$

where T_IN is the injection noise temperature, T_N is the blackbody temperature, L is the attenuation coefficient of the attenuator, and T₀ is the attenuator temperature.

However, to ensure a steady input, we adopt a vector signal generator (SMW200A R&S) to generate broadband white noise in our test. We assume the flux density of the solar microwave emission in the frequency range of 35–40 GHz to be 1200 SFU (single polarization). The power received by a 80 cm dish (effective area of ∼0.182 m²) is about 2.18 × 10⁻¹⁷ mW Hz⁻¹ (−166.6 dBm Hz⁻¹). And, therefore, the lowest power in the test should cover −79.6 dBm/500 MHz. Considering that the flux density enhancement during solar bursts cannot usually exceed several thousand SFU, the dynamic range can be restricted in the range of 3–7 dB, and therefore our tests are carried out in the range of −87.8 dBm/500 MHz to −72.8 dBm/500 MHz. The tested results of the system linearity are shown in Table 4 and Figure 8. We can see that the linearity is better than 0.9999, which can linearly reflect the relationship between the external input temperature and the counts of the digital receiver. A linear system can easily and realistically reflect the input–output relationship in the subsequent solar radio flux calibration process.

Zoom In Zoom Out Reset image size

In a real-world observation, fluctuations in the observed data are mainly caused by the target (including the external ambient) and gain fluctuations, so the system stability is tested to distinguish those fluctuations in the observations that may be generated by the instrument rather than the target. Stability was tested in the laboratory at room temperature (without temperature control) and in a real-world environment. The test results show that fluctuations in gain have little effect on the system's counts compared to real-world variations.

First, the gain–temperature relationship was measured, and it can be concluded that the higher the temperature the smaller the gain, and that the temperature control of the system can improve the system stability. The instrument stability is then tested by connecting the first-stage LNA with a 50 Ω terminal instead of the antenna (Figure 9). As shown in Table 5, the coefficient of variation, standard deviation divided by the mean, is less than 0.61% within 9 hr (the range of variation is 2.2%–3.2%).

Zoom In Zoom Out Reset image size

The real-world observation is shown in Figure 10. Observations for several days in the period 2021 August 9–September 3 were selected and they were accompanied by various weather conditions such as cloud, rain, rain to clear, and sunny in one day. The variation of output values due to environmental factors is so large that even if the system is calibrated, it should be supplemented by other observation equipment at the same moment to exclude the influence of environmental factors when there is a rise in radiation intensity.

Zoom In Zoom Out Reset image size