This is a two part article and Part 2 of this article will run in next month’s issue of MPD.
by Brad Hall and Wyatt Taylor, Analog Devices, Inc.
Many aerospace and defense electronics systems in the SATCOM, radar, and EW/SIGINT fields have long required access to a portion, or all, of the X- and Ku- frequency bands. As these applications move to more portable platforms such as unmanned aerial vehicles (UAVs) and handheld radios it is critical to develop new small form factor, low power radio designs that operate in the X- and Ku- bands, while still maintaining very high levels of performance. This two-part article outlines a new high frequency IF architecture that drastically reduces the size, weight, power, and cost of both the receiver and transmitter without impacting the system specifications. The resulting platform is also more modular, flexible, and software defined than existing radio designs.
In recent years, there has been an ever increasing push to achieve wider bandwidths, higher performance and lower power in RF systems, all while increasing the frequency range and decreasing the size. This trend has been a driver for technology improvements which have allowed for greater integration of RF components than has been seen before. There are many drivers pushing this trend.
SATCOM systems are seeing desired data rates up to 4 Gbps to support transmitting and receiving terabytes of collected data per day. This requirement is pushing systems to operate in the Ku- and Ka- band due to the fact that wider bandwidths and higher data rates are easier to achieve at these frequencies. This demand means a higher density of channels and a wider bandwidth per channel.
Another area of increasing performance requirements is in EW and Signals Intelligence. Scan rates for such systems are increasing, driving the need for systems which have a quick tuning PLL and/or wide bandwidth coverage. The drive toward lower size, weight, and power (low-SWaP) and more integrated systems stems from the desire to operate handheld devices in the field as well as increase channel density in large fixed location systems.
The advancement of phased arrays is also enabled by further integration of RF systems in a single chip. As integration pushes transceivers smaller and smaller, it allows each antenna element its own transceiver, which in turn enables the progression from analog beamforming to digital beamforming. Digital beamforming provides the ability to track multiple beams at one time from a single array. Phased array systems have a myriad of applications, whether it is for weather radar or for EW applications or directed communications. In many of these applications, the drive to higher frequencies is inevitable as the signal environment at lower frequencies becomes more congested.
In this two-part article, these challenges are addressed using a highly integrated architecture based upon the AD9371 transceiver as an IF receiver and transmitter, allowing the removal of an entire IF stage and its associated components. Included is a comparison between traditional systems and this proposed architecture, as well as examples of how this architecture can be implemented through a typical design process. Specifically, the use of an integrated transceiver allows for some advanced frequency planning that is not available in a standard super-het style transceiver.
Overview of Super Heterodyne Architecture
The super-heterodyne architecture has been the architecture of choice for many years due to the high performance that can be achieved. A super-heterodyne RX architecture typically consists of one or two mixing stages, which are fed into an Analog to Digital Converter (ADC). A typical super-het transceiver architecture can be seen in Figure 1.
The first conversion stage up-converts or down-converts the input RF frequencies to an out-of-band spectrum. The frequency of the 1st IF (intermediate frequency) depends on the frequency and spur planning as well as the mixer performance and available filters for the RF front end. The 1st IF is then translated down to a lower frequency that the ADC can digitize. Although ADCs have been making impressive advances in their ability to process higher bandwidths, their upper limit today is around 2 GHz for optimal performance. At higher input frequencies, there are trade-offs in performance vs input frequency that must be considered as well as the fact that higher input rates require higher clock rates, which drive up power.
In addition to the mixers, there are filters, amplifiers and step attenuators. The filtering is used to reject unwanted out-of-band (OOB) signals. If unchecked, these signals can create spurious that falls on top of a desired signal, making it difficult or impossible to demodulate. The amplifiers set the noise figure and gain of the system, providing adequate sensitivity to receive small signals while not providing so much that the ADC oversaturates.
One additional thing to note is that this architecture frequently requires Surface Acoustic Wave (SAW) filters to meet tough filtering requirements for anti-aliasing in the ADC. With SAW filters comes sharp roll-off to meet these requirements, however significant delay as well as ripple is also introduced.
An example of a super-het receiver frequency plan for X-band is shown in Figure 2. In this receiver, it is desired to receive between 8 GHz and 12 GHz with a 200 MHz bandwidth. The desired spectrum mixes with a tunable local oscillator (LO) to generate an IF at 5.4 GHz. The 5.4 GHz IF then mixes with a 5 GHz LO to produce the final 400 MHz IF. The final IF ranges from 300-500 MHz, which is a frequency range where many ADCs can perform well.
Receiver Specs – What Matters
Aside from the well-known gain, noise figure and 3rd order intercept point specifications, some typical specifications that influence the frequency planning for any receiver architecture include image rejection, IF rejection, self-generated spurious, and LO radiation.
- Image spurs – RF outside of band of interest that mixes with LO to generate tone in IF
- IF spurs – RF at IF frequency that sneaks through filtering before the mixer and shows up as a tone in the IF
- LO radiation – RF from the LO leaking out to the input connector of the RX chain. LO radiation gives a means of being detected even when in a receive-only operation. (See Figure 3)
- Self-generated spurious – spur at IF that results from mixing of clocks or local oscillators within the receiver
Image rejection specs apply to both the 1st and 2nd mixing stage. In a typical application for X- and Ku- band, the 1st mixing stage may be centered around a high IF in the 5-10 GHz range. A high IF is desirable here due to the fact that the image falls at Ftune+2×IF as shown in Figure 4. So the higher the IF, the further away the image band will fall. This image band must be rejected before hitting the 1st mixer, otherwise out of band energy in this range will show up as spurious in the 1st IF. This is one of the primary reasons why two mixing stages are typically used. If there were a single mixing stage with the IF in the hundreds of MHz, the image frequency would be very difficult to reject in the front end of the receiver.
An image band also exists for the 2nd mixer when converting the 1st IF down to the 2nd IF. As the 2nd IF is lower in frequency (anywhere from a few hundred MHz up to 2 GHz), the filtering requirements of the 1st IF filter may vary quite a bit. For a typical application where the 2nd IF is a few hundred MHz, the filtering can be very difficult with a high frequency 1st IF, requiring large custom filters. This can frequently be the most difficult filter in the system to design due to the high frequency and typically narrow rejection requirements.
In addition to image rejection, the LO power levels coming back from the mixer to the receive input connector must be filtered aggressively. This ensures that the user cannot be detected due to radiated power. To accomplish that, the LO should be placed well outside of the RF passband to ensure adequate filtering can be realized.
Introducing the High-IF Architecture
The latest offering of integrated transceivers includes the AD9371, a 300 MHz to 6 GHz direct conversion transceiver with 2 receive and 2 transmit channels. The receive and transmit bandwidth is adjustable from 8 MHz up to 100 MHz and can be configured for Frequency-Division Duplex (FDD) or Time-Division Duplex (TDD) operation. The part is housed in a 12 mm2 package and consumes ~3 W of power in TDD mode or ~5 W in FDD mode. With the advancement of quadrature error correction (QEC) calibrations, image rejection of 75-80 dB is achieved.
The advancement of performance of the integrated transceiver ICs has opened up a new possibility. The AD9371 incorporates the 2nd mixer, 2nd IF filtering and amplification, variable attenuation, ADC, as well as digital filtering and decimation of the signal chain. In this architecture, the AD9371, which has a tuning range of 300 MHz to 6 GHz, can be tuned to a frequency between 3 GHz and 6 GHz and receive the 1st IF directly (see Figure 6). With a gain of 16 dB, NF of 19 dB and OIP3 of 40 dBm at 5.5 GHz, the AD9371 is ideally specified as an IF receiver.
With the use of the integrated transceiver as the IF receiver, there is no longer a concern of the image through the 2nd mixer as is the case with the super-heterodyne receiver. This can greatly reduce the filtering required in the 1st IF strip. However, there must still be some filtering to account for 2nd order effects in the transceiver. The 1st IF strip should now provide filtering at 2x the 1st IF frequency to negate these effects, a much easier task than filtering the 2nd image and 2nd LO away, which can be as close as several hundred MHz. These filtering requirements can typically be addressed with low cost, small off-the-shelf LTCC filters.
This design also provides a high level of flexibility in the system and can be easily reused for different applications. One way that flexibility is provided is in the IF frequency selection. A general rule of thumb for IF selection is to put it in a range that is 1 GHz to 2 GHz higher than the desired spectrum bandwidth through the front end filtering. For example, if the designer desires 4 GHz of spectrum bandwidth from 17 GHz to 21 GHz through the front end filter, the IF can be placed at a frequency of 5 GHz (1 GHz above the desired bandwidth of 4 GHz). This allows for realizable filtering in the front end. If only 2 GHz of bandwidth is desired, an IF of 3 GHz could be used. Furthermore, due to the software-definable nature of the AD9371, it is easy to change the IF on the fly for cognitive radio applications where blocking signals can be avoided as they are detected. The easily adjustable bandwidth of the AD9371 from 8 MHz to 100 MHz further allows for avoiding interference near the signal of interest.
With the high level of integration in the High-IF Architecture, we end up with a receiver signal chain that takes up about 50% of the space required for an equivalent super-het while decreasing the power consumption by 30%. In addition, the High-IF Architecture is a more flexible receiver than the super-het architecture. This architecture is an enabler for low-SWaP markets where small size is desired with no loss of performance.
In this first part, we have discussed some of the applications where low-SWaP receivers have become a driver to improve technology. An overview of the traditional approach with a super-het architecture was given and then an alternate solution using a High-IF Architecture was shown.
In the next part, we will be exploring some of the details regarding frequency planning with the high-IF architecture as well as providing several example systems to guide the designer.