Advantages of Tunable Filters for RF Sampling Architectures

okhxyyo

Advantages of Tunable Filters for RF Sampling Architectures [Copy link]

Advantages of Tunable Filters for RF Sampling Architectures

While RF amplifiers, VGAs, and digital step attenuators have evolved to support multi-gigahertz bandwidths in a single device, the total signal chain cumulative passband selectivity is typically fixed and tailored for one band of operation. This means that a new bill of materials is required for each specific band design. Tunable bandpass filters allow multi-band radios to be made from a small signal RF chain, with the ability to fine-tune the filter’s center frequency, passband, and out-of-band rejection.

Using a single tunable device is particularly important when designing radios for multiple 4G LTE or 5G NR applications, as reusing the same device across multiple platforms can provide significant cost advantages. Additionally, the cost of development time and testing efforts can be further reduced.

Figure 1: OTFL101 S21 response over the entire tuning range

To illustrate these advantages, this article focuses on a suitable application for tunable filters, namely the output reconstruction filter of a wideband direct RF digital-to-analog converter (DAC). The filter used in this example is the OTFL101, a member of the Otava family of bandpass tunable filters that meet filtering requirements between 2.5 GHz and 40 GHz. With power handling capabilities up to 1 W RMS and linearity of +45 dBm, these filters can be used almost anywhere in the RF signal chain.

The OTFL101 (shown in Figure 2) is a single-chip filter with a tuning range of 2.5 to 7 GHz, a size of 2.3 x 1.6 mm, and no external components. It is controlled via a three-wire serial interface using 1.8 VDC CMOS signals (CLK, data, device select). The filter is a fifth-order design in which each of the five capacitor banks or resonators can be individually configured with a 5-bit word for each capacitor bank or resonator. Figure 1 shows the superimposed S21 response at the five center frequencies and illustrates how the fractional bandwidth can be kept approximately constant over the tuning range by applying simple tuning.

Figure 2: OTFL101 evaluation board

RF Sampling Architecture Filtering Requirements

While there are many RF sampling digital-to-analog converters (DACs) available, the analysis performed here is based on a multi-channel AMD Xilinx Zynq UltraScale+ Gen3 RFSoC, which enables high-bandwidth RF sampling or direct RF conversion from DC to 6 GHz at sampling rates up to 9.85 G/samples/s using an on-chip digital upconverter. An upconverter is a digital modulator that precedes the DAC. Its high-resolution digital NCO enables precise modulation of the RF without the need for external analog mixers or modulators.

Figure 3: Direct-RF transmitter block diagram

The mixed terms that pass through the complex modulator are replicated by the DAC sampling process at the sampling rate of Fs, creating the image tones at nFs +/-Fo and the desired real signal at Fo. It is common practice to apply a reconstruction filter directly after the digital-to-analog conversion to remove these unwanted terms before amplification.

Figure 4: DAC image relative to output DAC frequency Fo (6.144Gsps)

The most dominant image tones at the DAC output are located at the Fs-Fo and 2Fs-Fo frequencies. Figure 4 shows their frequencies relative to the desired signal at Fo when the DAC sampling rate is 6.144 Gsamples/s. From this graph, the nearby tone frequencies of the desired output signal with a Fo of 4 GHz can be extracted:

2.144 GHz image (Fs-Fo image)

6.144 GHz sampling clock leakage (Fs-leakage)

Image at 8.288 GHz (2Fs-Fo image)

Image at 10.144 GHz (Fs+Fo image)

As the DAC frequency response rolls off beyond 6 GHz, the spurious and image tones above that frequency tend to be smaller in magnitude than the in-band Fs-Fo images and do not require as much filter selectivity. Additionally, for a given DAC sampling rate, the frequency separation between the desired signal and its image changes as the target RF output frequency changes. This means that the DAC reconstruction filter must be replaced or adjusted to track the desired RF signal and provide adequate selectivity when needed.

Table 1: Comparison of fixed filters and OTFL101

Table 1 briefly describes the comparison of the OTFL101 tunable filter with existing off-the-shelf ceramic or LTCC filters.

Comparison with Fixed Frequency Filters

Figure 3 shows a typical signal chain for implementing a direct RF transmitter (excluding the amplifier front end). This article will use this architecture to illustrate the performance of the signal chain when using either a traditional fixed filter or the OTFL101 as the DAC reconstruction filter. The key metrics of interest are the absolute output power level, SNR, and output noise floor density (Table 2). Since the SNR performance of the transmit chain is primarily determined by the digital upconverter, modulator, and small signal amplification chain, the power amplifier front end can be ignored in this discussion.

Table 2: Key signal chain metrics

The analysis also assumes operation at maximum gain setting (DSA at minimum attenuation) and considers two variants of the same broadband GaAs amplifier (Qorvo QPA9126 and QPA9127) with either 15 or 20 dB power gain. These two variants have no impact on power consumption. Despite the higher insertion loss, the tunable filter only results in a 1.5 dB degradation in SNR compared to a fixed filter implementation when used as a higher gain amplifier post filter.

This is primarily a result of the 1 dB reduction in output power combined with a 0.5 dB increase in the noise floor. For a 50 MHz modulated signal with a 9 dB peak-to-average ratio, the calculated SNR is sufficient to drive the power amplifier front end at nearly 68 dB. It is also worth noting that the tunable filter has no effect on the output linearity performance and performs just as linearly as the fixed filter.

Signal Chain Performance

The transmitter signal chain performance has been evaluated using a Xilinx development kit consisting of an AMD Xilinx ZCU208 card and an XM655 balun board for the differential to single-ended post-conversion DAC. The output of the balun was connected to an OTFL101 evaluation board, which was then connected to a Rhode & Schwarz FSW spectrum analyzer. All results were not de-embedded for transmission line and board interconnect losses.

Figure 5: DAC output response before filtering (Fo at 4GHz)

In the example test case, a baseband CW signal is modulated to a DAC RF output frequency of 4 GHz. Figure 5 shows the DAC output image along with the desired signal at 4 GHz within the 12 GHz span without DAC post filtering. The DAC sampling rate has been set to 6.144 GHz. The output spectrum after filtering through the OTFL101 with the center frequency set to approximately 3.9 GHz shows that the Fs-Fo image at -2.2 GHz and the 2Fs-Fo image at +3.88 GHz have over 50 dB suppression compared to the desired signal (Figure 6). The reduction in the out-of-band noise floor is not as noticeable due to the wide span setting of the analyzer.

Figure 6: After OTFL101 filtering, Fo = 4GHz

Now let’s look at a higher DAC output frequency, for this the filter is next tuned to 5 GHz as shown in Figure 7, using the same process except the DAC numerically controlled oscillator (NCO) frequency is now shifted to place the desired CW signal frequency at 5 GHz and the DAC sampling rate is set to 6.4Gsps. Figure 8 shows the clean output spectrum after filtering by the OTFL101.

Figure 7: OTFL101 response from 4 to 5 GHz Fc

Figure 8: OTFL101 after filtering at Fo = 5GHz

The OTFL101 has proven to be very effective in suppressing unwanted images, achieving over 50 dB suppression of the Fs-Fo image at 1.39 GHz and over 50 dB suppression of the 2Fs-Fo image at 7.78 GHz. Similar results are achieved over a wide range of RF output frequencies, making this RF family fully software programmable from 2.5 to 6 GHz.

Original link: https://www.eeworld.com.cn/zt/Qorvo/view/1567