Interference has been the bane of wireless communications since the earliest days of radio, and scientists and engineers of every era no doubt found their challenges to be the worst. Then, as now, RF and microwave filters are some of the foremost contributors to keeping interference in check, and most of the attention to filter design is focused on end-user devices that use tiny acoustic-wave filters. However, the need for traditional connectorized filters has not abated because in many applications there is no suitable substitute. So, it seems prudent to review some basic rules about key performance specifications to help designers correctly specify these filters so they meet the needs of their particular applications.
A reasonable question might be why inexpensive, compact SAW, BAW, and FBAR filters cannot solve every problem, relegating the hogged-out “box” to the annals of microwave history. The first reason is that these filters serve very different platforms, even though their mission is the same. Acoustic-wave filters are a fraction of the cost and size of filters made using discrete components and are fabricated in very high volumes, while their counterparts are made by hand in much smaller volumes.
Without acoustic-wave filters, phones, tablets, laptops, and any other devices in which RF and microwave transceivers are integrated would not exist in their current form or perhaps at all. However, they are limited in frequency to about 10 GHz depending on the type and handle RF power levels up to about 30 dBm. In contrast, filters fabricated using discrete components can be made to handle much more power and operate at frequencies far into the millimeter-wave region. Waveguide filters can handle even more power and can operate at even higher frequencies. Both are inherently ruggedized as they are housed in metal enclosures and can easily be made even more rugged.
Conventional connectorized filters are used in large numbers not just in wireless systems but in nearly every defense system, medical equipment, and many other systems as well. Finally, as these filters are typically made and tuned by hand, customizing their characteristics to accommodate the unique requirements of a specific system can be accomplished and they can be sold in very small numbers. This makes them well suited for applications in which the total requirement throughout a product run might be only a few or a few hundred systems.
Getting to Know the Response Curves
All filter types have the same basic bandpass, lowpass, highpass, and bandstop response curves, which are shown in Figure 1. Specific characteristics obviously vary with the filter’s exact specifications as well as the different ways they are achieved.
Bandpass filters like the one shown in Figure 2 are the most common type of RF and microwave filter used in communications and other systems as they allow signals between two specific frequencies (the passband) to pass without attenuation while dramatically reducing the strength of signals outside this region. When used in the transmit path, they contain the signal or signals within a specific bandwidth to ensure they do not interfere with adjacent services. In the receive path they optimize optimum receiver performance, specifically signal-to-noise ratio, by reducing and potentially eliminating interfering signals.
The common topologies for bandpass filters are cavity, waveguide, ceramic, lumped element, helical, and crystal, each of which has characteristics that make it better suited for some applications than others. For example, cavity and waveguide bandpass filters can handle much more power than other types, while the others are much smaller and better suited for other applications.
Bandstop and Notch Filters
A bandstop or band-reject filter rejects (attenuates) frequencies within a specific range while leaving frequencies lower and higher than its stopband largely unaffected, and is the opposite of a bandpass filter. A bandstop filter requires a higher Q factor to produce the same amount of rejection as a bandpass filter as well as more sections to produce an equivalent passband-to-reject band transition.
It is a second-order (two-pole) type with two cut-off frequencies, the -3 dB or half-powerpoints, that creates a wide stopband bandwidth between them. Specifically, a stopbandfilter passes all frequencies from DC to the lower cutoff frequency as well as those frequencies above the second cutoff frequency. Frequencies between these two cutoff frequencies are the filter’s bandwidth.
If a bandstop filter’s stopband is very narrow, perhaps only a few Hertz, it is called a notch filter as a deep notch is created with high selectivity, as shown in the Figure 1. This type of band-stop filter is highly selective and has a high Q factor, which makes it extremely useful for “notching out” an interfering signal or signals in order to allow the desired signals to pass.
Lowpass filters allow signals from DC to a specified frequency to pass unimpeded with no attenuation while rejecting signals higher than this frequency. They are essentially the opposite of highpass filters that allow higher frequencies to pass and those below the cutoff frequency to be rejected.
While in an ideal lowpass filter the cut-off would be a dramatic slope, in practice it declines gradually from the transition region to the stopband region as shown in the figure. Specifically, the cutoff frequency is the point at which frequency drops by -3 dB or 70.7% from the passband. The transition region is the area where this falloff occurs. The simplest lowpass filters consist of a resistor and capacitor, but more sophisticated lowpass filters combine series inductors and parallel capacitors.
A highpass filter passes signals higher than a specified cutoff frequency and rejects signals below the cutoff frequency; it performs the exact opposite function of a lowpass filter. The cutoff frequency is the frequency where the filter begins to make its transition from little attenuation to maximum or high attenuation. The frequency range above the cutoff frequency is the stopband region and the frequencies below it are the passband.
Highpass filters are used in both transmitters and receivers. In a transmitter, they will restrict the bandwidth of the output signal to the required specification. In a receiver, they preserve signals within a range of frequencies while also keeping signals at unwanted lower frequencies from passing, improving receiver signal-to-noise ratio.
Duplexers and Diplexers
A duplexer, which is composed of frequency-selective filters, allows bidirectional (duplex) communication over a single path and isolates the receiver from the transmitter, while allowing them to share a common antenna. Duplexers are widely used in base stations for virtually all types of communication systems as they provide the isolation required to prevent the transmitter’s output from overloading the receiver’s input. They also have the advantage of saving considerable space, coaxial cable, and connectors.
There are two main types of duplexers. Notch duplexers have significant signal attenuation at undesired frequencies and pass only the desired frequencies. Bandpass duplexers have wide passbands and high out-of-band rejection. At co-located antenna sites, bandpass duplexers are by far the most common as their ability to eliminate interference can dramatically improve receiver selectivity.
In contrast to a duplexer, a diplexer provides multiplexing in the frequency domain, multiplexing two ports onto one port, although more than two ports can be multiplexed, creating a triplexer (three ports to one port) or a quadruplexer (four ports into one port). This effectively allows inputs to use the same output port without interfering with each other. A diplexer differs from a combiner or splitter as the ports on a diplexer are frequency selective, and while a combiner captures all power delivered to one port and divides it between two other ports, a diplexer does not.
Key Specs to Remember
Cut-off frequency: This is the transition point from the passband to the start of the stopband in a lowpass or highpass filter and is normally specified at the 3 dB point.
Rejection frequency: The frequency or frequencies where the signal is attenuated at one or more specified values defines the rejection frequency. For bandpass and bandstop filters, the specified frequency is the center frequency. For lowpass and highpass filters, the specified frequency is the cut-off frequency.
Stopband: The lower and upper limiting frequencies (also called the lower and upper stopband corner frequencies) are where the stopband and the transition bands meet.
Isolation: In a diplexer, the filter’s transmit/receive isolation is the ability to reject the transmit frequency while accommodating the receive frequency and vice versa. The greater the isolation, the better the filter can isolate receive from transmit frequencies.
Insertion loss: This is a measure of power loss as the signal passes from the input to the output of the filter.
Return loss: The return loss is the amount of the signal that is returned or reflected by the filter, expressed in decibels. It is expressed in another way as VSWR.
Group delay: The group delay in a filter is the delay in time of the signal passing through the device as a function of frequency. Ideally, phase would be linear so group delay would be constant, but in practice group delay distortion in a filter is not ideal, as signals at different frequencies take different amounts of time to pass through it.
Shape factor: The shape factor of a filter is the ratio of the stopband bandwidth to the 3dB bandwidth and reflects the sharpness filter rejection. For example, if the 40dB bandwidth is 40 MHz, and the 3dB bandwidth is 10 MHz the shape factor will be 40/10 = 4.
Impedance: The value specified in ohms of the filter source (input) impedance and the terminating (output) impedance, which are typically are the same.
Relative attenuation: The attenuation difference measured from the minimum attenuation point to the desired rejection point. Relative attenuation is usually specified in dBc.
Ripple: The flatness of the passband in a filter is its ripple and is normally expressed in decibels; its amount will affect the return loss. That is, the greater the ripple, the lower the return loss, and vice versa.
In an industry where smaller is always considered to be better, the conventional RF and microwave filter is an outlier. Not only is it far larger than acoustic wave filters, it weighs more, and is more expensive. However, as explained earlier, there are a wide variety of applications in which small size is not the most important parameter and which require performance characteristics that acoustic-wave filters cannot achieve. There may well be a time when this is not the case, but that day is not coming soon.