Today's multi-channel wideband multi-octave tuned RF receivers often need to eliminate unwanted blocking signals in order to maintain the fidelity of the relevant signal. Filters play an important role in reducing these unwanted signals, especially in the receiver RF front-end and local oscillator (LO) sections of these systems. This article will explore filters in RF signal chains, discuss the concept of blocking signals, review traditional filtering techniques, and then introduce new product solutions for optimizing signal chain performance.

To continually reduce size, weight, power and cost while increasing or maintaining performance, it is necessary for RF system designers to evaluate every component in the signal chain and look for opportunities for innovation. Since filters typically take up a lot of board space, this is an area of ​​focus when considering size reduction.

At the same time, receiver architectures have continued to evolve, with analog-to-digital converters (ADCs) capable of sampling at higher input frequencies. As the ADC input frequency increases, the constraints on the filter in the signal chain also change. In general, this trend means a relaxation of filter rejection requirements, which provides an opportunity to further optimize size and tuning performance.

Before we begin our exploration, we will first provide an overview of the RF signal chain and definitions to illustrate where and why filters are needed. In addition, reviewing traditional technologies can also provide insight into the current situation. Then, by comparing these legacy technologies with the latest product solutions, it becomes clear how easily system designers can achieve their goals.

RF Signal Chain Overview

Figure 1 shows a typical wideband signal chain covering 2 GHz to 18 GHz. The basic working principle of this signal chain is as follows. The antenna receives a wide range of frequencies. A series of amplification, filtering, and attenuation controls (RF front-end) are required before the frequency is converted to an IF signal that the ADC can digitize. The filtering functions in this block diagram can be divided into four categories:

• Preselector sub-octave filtering

• Image/IF signal rejection

• LO harmonics

• Anti-aliasing

Figure 1.2 Ghz to 18 GHz receiver block diagram.

Figure 2. (a) Sub-octave preselection mitigates IMD2 problems; (b) filter band broadens with frequency.

Figure 3. (a) Image band and (b) IF band that must be suppressed before the mixer.

Preselector sub-octave filtering needs to be close to the start of the signal chain to address second-order intermodulation distortion (IMD2) spurs, which can arise in the presence of interfering signals (also known as blocking signals). This occurs when two out-of-band (OOB) spurs add or subtract and form an in-band spur that can mask the signal of interest. Sub-octave filters can remove these interfering signals before they reach the nonlinear elements of the signal chain, such as amplifiers or mixers. Typically, the absolute bandwidth requirement of a sub-octave filter becomes narrower as the center frequency decreases. For example, the first frequency band of a 2 GHz to 18 GHz signal chain may only cover 2 GHz to 3 GHz and requires good rejection on the low side (F_high/2) at 1.5 GHz and the high side (F_low × 2) at 4 GHz , while the highest frequency band of the signal chain may cover 12 GHz to 18 GHz with good rejection on the low voltage side at 9 GHz and the high voltage side at 24 GHz. These differences mean that more filters are needed to cover the low frequencies than the high frequencies. An example of the spectrum filtered by the preselector is shown in Figure 2.

Image/IF rejection filtering is usually downstream in the signal chain, between the LNA and the mixer. It is used to suppress image frequencies and unwanted IF frequencies. An image is a frequency band that, when present at the mixer input, produces a signal of the same amplitude as the desired signal at the mixer output. Image rejection can be achieved through several components in the signal chain, such as preselection filters, dedicated image rejection filters, and image rejection capabilities from single-sideband (SSB) mixers. IF signal rejection requires reducing the spectrum of IF frequencies before the mixer, preventing them from leaking directly onto the mixer and appearing as unwanted spurs. Figure 3 shows an example spectrum of the unwanted image and IF bands.

Depending on the LO generation circuit, the filtering requirements at this point in the signal chain may vary. The target signal input to the LO port of the mixer is a clean sine or square wave. Typically, LO circuits generate sub-harmonics and harmonics of the desired LO signal. These unwanted signals (see Figure 4) need to be suppressed before reaching the mixer to avoid unwanted MxN spurious products. If the LO signal is at a single frequency, then a fixed bandpass filter is sufficient and can be optimized to pass only the signal of interest. In a wideband signal chain, a tunable LO signal is usually implemented, so a bank of switched filters or a tunable filter is required.

Figure 4. LO harmonic filtering.

Figure 5. Without adequate suppression, aliasing in an ADC can cause interfering signals to appear in a certain frequency band.

When sampling with an ADC, the system designer needs to choose the Nyquist zone to be digitized. The first Nyquist zone ranges from DC to fS/2 (where fS is the sampling rate of the ADC). The second Nyquist zone is from fS/2 to fS, and so on. Anti-aliasing filters are used to suppress interfering signals in the Nyquist zone adjacent to the target Nyquist zone. Interfering signals at this point in the signal chain can come from different sources, such as MxN spurs generated in the mixer, downconverted signals adjacent to the signal of interest, or from harmonics generated in the IF signal chain. When digitizing, any interfering signal input to the ADC will alias into the first Nyquist zone. An example of the spectrum of the unwanted aliased signal is shown in Figure 5.

blocking signal

In RF communication systems, a blocker is a received interfering input signal that reduces the gain and signal-to-noise ratio (SINAD) of the signal of interest. The blocking signal may directly mask the target signal, or it may produce spurious products that mask the target signal. These unwanted signals can be the result of unintentional or intentional interference. In the former case, it is from another radio frequency communication system operating in the adjacent spectrum. In the latter case, it comes from a malicious electronic warfare system designed to intentionally interfere with RF communications or radar systems. Figure 6 shows an example of the spectrum of the blocking signal and the target signal.

Figure 6. Target and blocking signals.

Many RF components exhibit weakly nonlinear memoryless behavior. This means that they can be approximated by low-order polynomials. For example, a wideband frequency amplifier can be modeled by an odd-order polynomial consisting of only first- and third-order terms:

When there are two incident signals at the input of the amplifier in the operating frequency range, as in the case of the target signal ω1 and the blocking signal ω2, the input signal can be described as:

Substituting the input equation into an odd-order polynomial yields the following output:

When the amplitude of the target signal is much smaller than the blocker signal, A<

According to the simplified Equation 4, the target signal amplitude is now closely related to the blocking signal amplitude B. Since most of the RF components of interest are compressed, the alpha coefficients must be of opposite sign, such that α1α3 < 0. The result of the above two statements is inevitable because the gain of the target signal tends to zero for larger blocker signal amplitudes.

filter definition

To address the problem of interfering signals in RF communication systems, engineers rely on filters to reduce these signals and preserve the desired signal. Simply put, a filter is a component that allows frequencies to be passed in the passband and rejected in the stopband.

Typically, the insertion loss (dB) of a filter can be described as low-pass, high-pass, band-pass, or band-stop (notch). This term refers to the plotted allowable passband frequency response versus increased frequency. Filters can be further classified based on their frequency response waveforms, such as passband ripple, stopband ripple, and how quickly they roll off with respect to frequency. For illustration purposes, Figure 7 shows the four main filter types.

Figure 7. Filter waveforms by type.

Besides insertion loss, another important characteristic of filters is group delay. Group delay refers to the rate of change of transmission phase with respect to frequency. The unit of group delay is time (seconds), so this metric can be thought of as the transit time of a particular signal through the filter. The propagation time of a single frequency by itself usually has little effect, but when a wideband modulated signal is passed through a filter, the flatness of the group delay becomes important because it can introduce different time delays into the received signal, distorting the signal. Equation 5 gives the equation for the group delay, where θ is the phase and ƒ is the frequency:

Typical filter types with significant insertion loss and group delay characteristics are Butterworth, Chebyshev, Elliptic and Bessel. Each type is usually defined by an order, which describes how many reactive elements are in the filter. The higher the order, the faster the frequency roll-off.

When considering filters of similar order, the Butterworth filter provides the flattest possible passband response at the expense of frequency roll-off, while the Chebyshev filter has a good frequency roll-off but some passband ripple. Elliptic filters (sometimes called Cauer-Chebyshev) have more frequency roll-off than Chebyshev filters, but therefore also create ripple in the passband and stopband. The Bessel filter has the flattest frequency and group delay response, but the worst frequency roll-off. For illustration purposes, Figure 8 shows the ideal insertion loss and group delay of a fifth-order low-pass filter with a 3 dB frequency (f3 dB) of 2 Ghz, an allowable passband ripple of 1 dB, and a stopband ripple of 50dB.

For systems where it is important to maintain a constant phase over the entire frequency range, such as radar systems, the group delay flatness of the relevant frequency band is critical to avoid unexpected phase deviations in the received pulses. Assuming that the received signal range can cover 1 GHz or more, the group delay flatness over the wide band should be minimized. As a rule of thumb, group delay flatness should be kept to < 1 ns, but this depends on the system's tolerance for phase deviation. Figure 9 shows examples of filters with group delay flatness of 2.24 ns and 0.8 ns, respectively. Looking at these waveforms shows that for a flatter group delay, the phase change is more consistent across the frequency range.

Finally, the quality factor (Q-factor) of the reactive components used to design the filter is an important property that affects performance. The figure of merit is defined as the ratio of the reactive impedance of a particular circuit element to the series loss resistance. It is closely related to the technical process and the physical area used for implementation. The higher the figure of merit, the faster the frequency response and the lower the insertion loss.

Figure 8. Insertion loss and group delay of a fifth-order low-pass filter.

Figure 9. Deviation of the effect of group delay flatness from linear phase: (a) shows the group delay flatness of 2.24 ns (b) shows the flatness of 0.8 ns, comparing the two, it can be seen that the relationship between phase change and frequency is more consistent .

Traditional filtering techniques for RF communications

When designing filters for RF communication systems, there are several techniques for implementing classical filters. Traditionally, RF engineers have relied on discrete lumped-element implementations with surface-mount components, or distributed-element filters that include transmission lines printed on PCB material. However, in recent years, filters have been designed based on semiconductor processes, allowing the use of precise temperature-stabilized reactive components, and the figure of merit has improved. Additionally, semiconductor processes support the use of switches and tunable reactive elements, which can be more challenging in discrete lumped-element implementations. There are other technologies such as Bulk Acoustic Wave (BAW), Surface Acoustic Wave (SAW), Low Temperature Cofired Ceramics (LTCC), Cavity Filters or Ceramic Resonators.

Each approach and technique has trade-offs:

• Lumped LC filters are implemented with surface mount inductors and capacitors on the PCB. The advantage of this is that it is easy to assemble, and then adjust the value to change the performance of the filter.

• Distributed filters are designed as resonant slices of transmission lines implemented on a dielectric (either integrated into a PCB or stand-alone on a separate dielectric) and oriented to act as quasi-inductors or quasi-capacitors in certain frequency ranges . They exhibit periodic characteristics. In some cases lumped elements are added to improve/miniaturize distributed filters.

• Ceramic resonator filters use multiple ceramic resonators (this is a distributed element) coupled by lumped elements. The coupling element is usually a capacitor, but an inductor is sometimes used. This type of filter is a hybrid of distributed and lumped elements.

• Cavity filters are implemented by distributed elements (rods) enclosed in conductive boxes. They are known for being able to handle high power with little to no losses, but at the expense of size and cost.

• BAW and SAW technologies can provide excellent performance, but they often have frequency selection requirements and are not suitable for broadband applications.

• LTCC filters are implemented by combining multiple layers of distributed transmission lines in a ceramic package, which is similar to a distributed filter and can be used in a variety of applications, but it is fixed. Since they are 3D stacked, they end up taking up very little space on the PCB.

• With recent improvements in semiconductor performance, filters integrated into semiconductors support a wider frequency range. The ease with which digital control components can be integrated into these components facilitates the adoption of software-defined transceivers. Overall, the trade-off between performance and integration provides useful value to designers of broadband systems.

Table 1. Filter Type Comparison

The latest filter solutions

Analog Devices has developed a new family of digitally tuned filters that utilize enhanced semiconductor processes and industry-friendly packaging techniques. This technology enables small, high rejection filters that can alleviate blocking problems in receivers. These filters are highly configurable via standard serial-to-parallel interface (SPI) communication and feature fast RF switching speeds. In addition, ADI has added a 128-state look-up table in each chip to quickly change the filter state for fast frequency hopping applications. The combination of high rejection fast tuning and wide frequency coverage enables next-generation receiver applications to operate in adverse spectrum environments.

Figure 10. ADMV8818 functional block diagram.

Figure 11. Block diagram of a 2 GHz to 18 GHz receiver using the ADMV8818 as a preselector and mirror filter.

The latest products introduced using this technology are the ADMV8818 and ADMV8913. The former has four high-pass filters and four low-pass filters and operates from 2 GHz to 18 GHz; the latter has one high-pass and low-pass filters and operates from 8 GHz to 12 GHz.

The ADMV8818 is a highly flexible filter packaged in a 9 mm × 9 mm package with tunable bandpass, highpass, lowpass, or bypass responses between 2 GHz and 18 GHz. The chip consists of two parts: an input part and an output part. The input section has four high-pass filters and an optional bypass, selectable via two RFIN switches. Likewise, the output section has four low-pass filters and an optional bypass selectable via two RFOUT switches. Each high-pass and low-pass filter can be tuned with 16 states (4 control bits) to adjust 3 dB frequency (f3 dB). Figure 10 shows the functional block diagram of the ADMV8818.

With a flexible structure that can be quickly reconfigured and a small form factor, the ADMV8818 provides full coverage in the 2 GHz to 18 GHz frequency band without any dead zones. The ADMV8818 can be configured as a sub-octave preselection filter, mirror image, or IF filter. When configured in the signal chain shown in Figure 11, the receiver can maintain high sensitivity and can instead use the ADMV8818 as a preselector in the presence of large OOB signals.

For example, if a target signal is received near the 9 GHz band, but there is a strong OOB blocker signal in the 4.5 GHz band, the blocker will cause harmonics to appear near the 9 GHz target signal, preventing operation. Configuring the ADMV8818 as a 6 GHz to 9 GHz bandpass filter allows broadband signals to pass through while appropriately reducing the level of blocking signals before they cause harmonic problems in the nonlinear elements of the signal chain. The S-parameter sweep of the ADMV8818 configured for this situation covers the blocking signal, as shown in Figure 12.

Figure 12. ADMV8818 configured as a 6 GHz to 9 GHz bandpass filter. The filter rejects F2–F1, F1+F2, F/2, and F×2 spurious products.

A size comparison of a typical 2 GHz to 18 GHz preselection filter module is shown in Figure 13. Among them, the switch fixed filter preselector group is realized by distributed filtering technology on the ceramic substrate. Dimensions are estimated based on filter products on the market. Eight-throw switches are included in the estimate to compare equivalent functionality. The tunable BPF shown in the figure is the ADMV8818, which covers the same frequency range and offers more tuning flexibility than a switched filter bank. The ADMV8818 has a footprint savings of over 75% compared to switched filter banks. The preselector function in the receiver signal chain typically accounts for a significant proportion of the overall size of the system, so this footprint savings is critical in size-constrained electronic warfare systems that can flexibly vary in size versus performance. trade-offs between them.

The ADMV8913 is a combination high-pass and low-pass filter in a 6 mm × 3 mm package that is specifically designed to operate in the frequency range (X-band) from 8 GHz to 12 GHz with insertion loss as low as 5 dB. Both the high-pass and low-pass filters can be tuned with 16 states (4 control bits) to adjust the 3 dB frequency (f3 dB). In addition, the ADMV8913 integrates a parallel logic interface to set the filter state without requiring SPI communication. This parallel logic interface is useful for systems that require fast filter response times because it eliminates the time required for SPI processing. Figure 14 shows the functional block diagram of the ADMV8913.

Modern X-band radar systems, whether using mechanically steered antennas or high-channel digitally phased array beams, typically rely on filtering solutions that are compact in size, low insertion loss, and easy to configure. The ADMV8913 is ideal for this application due to its low insertion loss, small size, and flexible digital interface options (SPI or parallel control). These features allow it to be located close to the front end of these systems, ensuring excellent performance while reducing integration complexity.

Figure 13. Fixed-switched 2 GHz to 18 GHz BPF (left) versus digitally tunable 2 GHz to 18 GHz BPF (right). Footprint savings of over 75%.

Figure 14. ADMV8913 Functional Block Diagram

in conclusion

There are many factors to consider when designing an RF front-end for a wideband receiver. The front end must be designed to handle unpredictable blocking conditions while still detecting low-level signals. The ability to dynamically adjust front-end filtering performance to handle these blocking signals is a key feature of RF front-ends. ADI's new digitally controlled tunable filter IC products offer outstanding performance and enhanced digital functionality to meet the needs of many front-end applications. These two new products are just the first of many new developments in the digitally tunable filter portfolio.

Leave a Reply

Your email address will not be published.