Important Parameters for SDR Systems

Sep 24, 2021

A software-defined radio (SDR) is a radio communication system that uses software for all transmit and receive functionality. In an SDR, signal processing is mainly implemented by means of software in a general-purpose computer or reconfigurable embedded system. Hence in principle, based merely on the software used, an SDR can transmit and receive practically any radio protocols or signal waveforms. An SDR consists of a radio front end and digital back end.

The SDR’s radio front-end contains the receive (Rx) and transmit (Tx) functions to receive and transmit signals over a wide bandwidth; this is also what the antenna is connected to. State-of-the-art modern SDR systems cover an ultra-wide frequency spectrum. For example, the Per Vices SDR covers a very wide frequency band ranging from 0 to 18 GHz (and upgradeable to even higher upper bounds).

Fundamental components of an SDR are found on the following five modular circuit boards: Transmit Board (Tx chain), Receive Board (Rx chain), Power Distribution Board, Digital Board, and Time Board. The Tx chain consists of components such as digital-to-analog converters (DACs), anti-imaging filters, upconverters/mixers, and RF power amplifiers (RF PA). The Rx chain contains low noise amplifiers (LNA), low-pass (LP)/high-pass (HP) filters, downconverters/mixers, anti-aliasing filters, and analog-to-digital converters (ADCs). The power distribution board is responsible for providing a regulated power supply to various boards (Tx, Rx, digital board). Digital boards have digital electronics including an FPGA and hard processor system (HPS), general purpose processor (GPU) and outputs to SFP+ ports. The time board houses an oven controlled crystal oscillator (OCXO), which sends extremely accurate clocking signals to the ADC/DAC and FPGA. A depiction of the fundamental SDR architecture is shown in Figure 1.

Figure 1: Fundamental SDR architecture is shown with its various circuit boards.

SDR Components and Their Respective Performance Parameters

ADCs and DACs

The ADC is a device that samples a continuous signal and generates codewords (digital bits) with a resolution that is equal to number of bits of the ADC. Sampling is done at the clock frequency. The DAC transforms codewords to analog signals, and is essentially opposite to what an ADC does. Some of the main performance parameters of commercial ADCs are resolution (bits), maximum sample rate, signal-to-noise ratio (SNR), spurious-free dynamic range (SFDR), serialization time, and current consumption. Ideally, an ADC’s SNR is six times its number of bits, that is, 8-bit, 10-bit and 12-bit ADCs would have SNRs of 48, 60 and 72 dB respectively. Similarly, the performance of a DAC can be quantified by its output voltage range, deserialization time, and current consumption. An ADC is a very critical SDR component, as it will have a significant effect on the dynamic range of the overall SDR system. The highest performance SDRs have 16-bit ADCs/DACs to ensure high SNR and SFDR.

Analog and Digital Filters

A filter is an important component in the radio front end of an SDR to separate the low, mid and high band chains of the circuit board. A filter is a device that removes noise and unwanted signal and/or frequency components. Filters can be both analog and digital. Analog filters can remove anything above or below a given frequency, which is also known as cutoff frequency; however, a digital filter can be programmed to be more precise and have a steeper cutoff. A filter that removes signals below a specific frequency is called a high pass filter because it “passes” higher frequencies, while a filter that removes signals above a specific frequency is known as a low pass filter. The specific frequency that lets all signals pass above or below is known as cutoff frequency of that high pass or low pass filter. A high pass and a low pass filter can be cascaded to form a band pass filter that will only pass a signal having frequency that lies in between the cutoff frequencies of both cascaded filters. Analog filters can be realized with analog electronic components such as resistors, capacitors/inductors, and op amps which become more complicated if you desire steeper roll-offs (or more accurate in their attenuation abilities).

Digital filters can be way more precise in their filtering functions, but the input signal must be digital. There are two main categories of digital filters, namely a digital finite impulse response (FIR) filter and a digital infinite impulse (IIR) filter. IIR filters take less digital memory and can be easily derived from analog filters, while, on the other hand, FIR filters take a lot of memory and are generally more complex than their analog or IIR counterparts, and require a very careful design. The main advantage of FIR over IIR filters is their inherent stability. Important filter parameters are cutoff frequency, stopband, side lobe level (the difference in dBs between pass band and stop band response), active/passive, linear or nonlinear, as well as others. In an SDR, digital filters are implemented in FPGAs and allow for more fine tuning of signals.

Figure 2: The frequency responses of basic filter types are shown: (a) low pass filter, (b) high pass filter, (c) band-pass filter.

Crystal Clocks

The clock generator is what sends out a continuous and repeating pulse to the various boards in the SDR. Clock generators are often implemented using high precision crystal oscillators and are oven controlled to ensure temperature stability. These devices are critical to the SDR since all the functionalities of digital circuits (i.e., synchronization of ADCs/DACs, clocking for the FPGA), and signals carrying modulated data, are provided with a temporal reference using clock signals. Thus clock waveforms should be very clean and sharp and take into account important parameters such as jitter, accuracy/stability, drift, etc. All these parameters are important measures to ensure clock signals are understood by the different components at the correct frequency and voltage reference.

Amplifiers

Whenever the amplitude of an RF signal is too low to be used elsewhere in the circuit, or the signal’s amplitude must be matched to the input range of any component such as an ADC/DAC or a mixer, or it needs to be “amplified” (boosted) so the overall SNR (signal-to-noise ratio) does not degrade as the signal is routed through the rest of the circuit, an RF gain is needed. This is provided by an RF amplifier. There are various types of RF amplifiers, such as power amplifiers, variable gain amplifiers (VGA), low noise amplifiers (LNA), etc.

Often RF signals received from an antenna are in the microvolt (μV) range, whereas signal-processing circuits work far more effectively with signals typically ranging between 1 and 10 V (depending on the design). As the name suggests, LNAs have extremely low noise of their own and are usually placed right next to the antenna in the Rx chain. RF Power amplifiers have relatively high noise and are used to boost the RF signals at the transmitter side, increasing the output power before feeding it to the antenna element. VGAs are generally placed before the ADC to provide it with a stable input. Some important parameters for RF amplifiers are output power, distortion, gain, frequency response, sensitivity, SNR, crosstalk, and linearity.

Power Supply

The power supply is essentially a transformer with power regulation and distribution circuits. It distributes power to the various boards/components in an SDR. Some important parameters for an SDR power supply are input voltage range, output voltage range, output ripple voltage, accuracy, load/current, efficiency, temperature, size, and required protections. Typically, the input voltage is derived from utility AC supply, which is 220 VAC or 110 VAC, and the output voltage are mostly 28/12 VDC as required by most electronic circuits. Output ripple voltage represents the residual periodic voltage of a DC supply which has been derived from an AC source. Efficiency governs the electrical and thermal losses in a power supply which eventually determines the amount of cooling required, operating temperatures and reliability in various environmental conditions. Overcurrent protections are required to protect the system from damage in case of failures such as a short circuit. Of course, without the power supply, none of the boards or anything would work at all, leaving the SDR unfunctional.

Network Interface

Network interfaces on high performance SDRs are required for the high data throughput; these tend to be ethernet fibre optic cables using SFP+ ports on the device. Examples of important network interface parameters are ethernet framing, 10GBASE-R/40GBASE-R/100GBASE-R data rates, latency, etc. The network interface is important since this is how the IQ samples are streamed to/from the host machine for the various types of processing that may occur.

The Importance of These Components to the SDR Application

Modern SDRs are used for a wide range of applications such as spectrum monitoring, mmWave (5G) networks, satellite communications (SatCom), automotive sensing and ranging, and low latency applications. Depending upon the particular application, some SDR components are more critical than others. In this section we’ll discuss some major SDR applications and briefly present critical performance parameters and the components associated with those parameters.

SDRs in Spectrum Monitoring and Recording

Generally, during spectrum monitoring applications, very wide bandwidth SDRs are required to scan/sweep the entire spectrum of interest. This requires a very fast and high-resolution ADC and a very high throughput from the ADC to the SFP+ ports. The highest bandwidth SDRs have amplifiers that are very linear to gain very weak signals over a large bandwidth and are the obvious choice for such applications. Another consideration are the filters, which ensure that the SDR can narrow into bands of interest and filter out unwanted signals when electromagnetic interference or a jamming device is detected during wide band spectrum monitoring. In order to process the very large amounts of the RF spectrum being captured, network interfaces with 40GBASE-R are required.

SDRs in mmWave (5G) Networks

The millimeter-wave (mmWave) frequency spectrum ranges from 30 to 300 GHz, and 5G technology will typically utilize the lower regions (28/38 GHz) of this mmWave spectrum. For such extremely high frequencies (30-300 GHz), the ADCs and DACs must handle very high sample rates/frequencies. For the accuracy and high bandwidth of mmWave networks, very good clocks are required so that no jitter shows up as distortion on the signal, artifacts on the carrier, or spurious frequencies (spurs) on either side of the nominal carrier of the sampled data. Clocks are also very important to support the phase coherency required in beamforming/beamsteering of directional antennas for 5G cellular network nodes.

SDRs for SATCOM

For satellite communication (SATCOM) applications, power supplies are especially important, not only because you’ll likely need one that is able to have an input voltage from a solar cell type battery, but also because these devices will also require shielding from solar radiation. Another important parameter would be the dynamic range. SDRs are used in both SatCom ground stations and satellites in orbit, both of which require very sensitive radio front ends that have a good dynamic range and a low noise floor. Dynamic range is determined mainly by the ADCs as the ratio of maximum input power level to lowest input power level that an ADC can process. In addition to this, filters are also crucial for optimal communications in GNSS bands to realize interference resistant navigation systems (eg: GPS L1/L2 band, Galileo E1/E5a-E5b/E6 bands, GLONASS G1/G2/G3 bands).

Low Latency

Applications of low-latency links include high frequency trading, which requires extremely accurate and fast clocks to execute trades. Additionally, time stamping of financial transactions/trades may also be required by law in some jurisdictions at very specific times. FPGA code can be developed to support flow controls depending on how computationally intensive the processing is, or how large the encoded message is that’s being sent over the link, etc. DACs/ADCs with very low converter (de)serialization delays can reduce overall latency in the system. Thus, lowest latency SDR would be highly desirable for such latency critical applications.

Figure 3: High performance SDRs from Per Vices can be customized to support many applications.

Click here to learn more about Per Vices on everything RF.

Click here to view Software Defined Radios from Per Vices.

 

Contributed by :