Receiver Signal Chain Overview and Theory of Operation [Analog Devices Wiki] (2024)

This version (01 Jul 2024 17:00) was approved by Connor Bryant.The Previously approved version (01 Jul 2024 16:54) is available.

Table of Contents

Link to higher level page: Satcom Reference Design

The overall RX block diagram is shown below. The wiki sections below will walk through the design elements of each sub-circuit.

The 4T4R beamforming network consists of a modular array of four BFIC tiles, allowing for sixteen elements. The multi-beam analog beamforming begins with an ADL8142 LNA. The signal then passes through a Ka-band bandpass filter (27.5-30GHz), followed by the ADAR3001S BFIC. After the ADAR3001S, each of the four components of every channel are combined using an ADAR5000 Wilkinson combiner.

By utilizing high performance silicon switches such as the ADRF5024 and ADRF5144, the down-converter design allows for three separate signal paths. High frequency X, Ku and Ka-band signals are applied to the LNA input of the ADMFM2000. After LNA amplification and optional off-chip filtering, the input signal is applied to the ADMFM2000's internal mixer where it is down converted to an IF by mixing with an external LO. For more information about LO generation for this particular design, see this link. Midrange C-band signals bypass the internal LNA and mixer but still use the ADMFM2000's internal filtering and VGA. S-band and lower signals bypass the ADMFM2000 SIP and are sampled by the AD9082 ADC after further IF conditioning.

The following table shows the frequency conversion paths for this design.

The image below shows a graphical representation of the frequency plan. The x-axis represents the operational frequency of the reference design while the y-axis represents the different frequency bands ranging from L to Ka. Markers are shown for RF frequency, IF frequency, sampling frequency (Fs), image frequency, and alias frequency.

The horizontal lines represent the bandwidth of RF, IF, and aliasing frequency for each frequency band. The solid vertical lines show Fs and Fs/2, which are used to determine at what frequencies aliasing will occur. For frequencies higher than the 8GHz usable analog bandwidth of the AD9082, down conversion from an RF to IF is required. For the lower frequencies (C-band and below), an IF is not required so only the RF is plotted.

Shown below, the sampling frequency is lowered to 5GHz from 6GHz to avoid the signal folding onto itself over the boundary separating the second and third Nyquist zone.

The IF section is the final segment before the AD9082 ADC. Similar to the down-converter section, there are three pathways for signal conditioning ahead of the ADC. An HMC8413 wideband GaAs LNA provides moderate gain before the signal is filtered. The high frequencies such as Ka, Ku, X-band are conditioned in switch path 1 of an ADRF5040 silicon switch. Signals from S-band and lower are conditioned in switch path 2 while signals from C-band are filtered in switch path 3.

The AD9082 (“MxFE”) is an ideal digitizer for this wideband RF front end due to its high level of integration, performance, and RF sampling rate. The MxFE’s high instantaneous and analog input bandwidth enable RF sampling up to 8GHz, providing architectural and frequency planning flexibility at the system level. It also has highly configurable, on-chip DSP capabilities (DDC/DUC channelizers, NCOs, and programmable FIR filters) as well as four DAC channels and two ADCs for use in this multi-channel system.

The ADF4371 integrated PLL/VCO is an ideal choice for generating the MxFE sample clock. The ADF4371 can generate a clean, low-jitter clock from 62.5MHz up to 32GHz, and its differential outputs are capable of directly driving the MxFE clock pins. The ADF4371 is also used to generate clocking for the LO signal. You can read more about the LO signal path here.

The LTC6953 clock distributor is used to ensure proper clock synchronization between all of the ADF4371 devices in the reference design. An external source generates a clock signal, which is then buffered and replicated before being sent to the FPGA, the AD9082, and both the TX and RX LO signal paths.

Although only a single LTC6953 is required for 4T4R, With EZSync™ or ParallelSync™, multichip synchronization is realizable, allowing for expandability while maintaining the proper clock signals.

The ADI Frequency Planning Utility tool was used to determine the optimal AD9082 ADC sample rate, IF sampling frequency, and IF bandwidth. A 6GSPS ADC rate allows for maximum instantaneous bandwidth (IBW) and complete avoidance of HD2 aliasing in-band for a 1GHz signal bandwidth centered at 4.5GHz (2nd Nyquist). The figure below illustrates this.

The receiver and transmitter front ends described in this reference design use a frequency plan designed using the AD9082, specifically with 6GSPS and 12GSPS ADC and DAC sample rates, respectively. However, for the case of C-band operation, both the ADC and DAC sample rates are lowered to avoid the signal overlapping the edge of the 2nd and 3rd Nyquist boundary. The RF components themselves are wideband enough to allow for flexibility in the frequency plan and filtering.

Inside of the ADAR3001S, there are 16 digital step attenuators and 16 time delay units (TDU) combined into Variable Amplitude and Phase (VAP) blocks. Each digital step attenuator provides up to 31.5dB of attenuation with a resolution of 0.5dB. Each TDU can provide up to 35.3 ps of delay in steps of 0.56 ps. Each of the VAPs can tuned in a minimum of 10ns.

The ADAR3001S contains internal RAM which can store up to 64 states for each beam for a total of 256 beam states. For faster access, there is also optional FIFO memory, which stores up to 16 beam states per channel for a total of 64 beam states. The ADAR3001S is conveniently controlled by 4-wire SPI, allowing for control of up to 16 devices on the same serial line.

The downconversion and IF section starts with an ADRF5730 6-bit digital step attenuator capable of up to 31.5dB of attenuation in 0.5dB steps. These DSAs can be controlled either by parallel logic level voltages determined by truth tables or by 3-wire SPI. The internal DSAs of the ADMFM2000 work by parallel voltage control inputs. All of the RF switches, including the internal switches of the ADMFM2000, are controlled by parallel digital voltage supplies. In order to achieve full functionality of the switches, they must be kept at either a digital “high” or “low” state, determined by their respective datasheets.

The following simulations were created using Keysight SystemVue/Genesys. All components in the RX signal chain were modeled with frequency-dependent sys-parameter or s-parameter datasets to ensure accurate simulation results. The signal chain below takes into account the following components:

• Manufacturer-provided, frequency-dependent s-parameter or sys-parameter datasets for every active and passive RF component

• A standalone model of the ADMFM2000 microwave downconverter containing signal chains each with their own frequency-dependent s-parameter or sys-parameter datasets

• Frequency-dependent component interconnect modeling i.e. PCB trace loss approximation for DC-50GHz CPWG traces, 30mil Rogers 4350 PCB stackup

The Genesys workspace incorporates various system level cascade analyses that takes advantage of MATLAB equation-based frequency planning. Three different analyses are shown below:

• Noise power Spectral Simulation

• Two-Tone Test Spectral Simulation

• System Cascade highlighting gain (CGAIN), noise figure (CNF), third input third order intercept (EIIP3), input 1dB compression point (EIP1DB), minimum detectable signal (MDS), and carrier to noise distortion ratio (CNDR)

The noise power spectral simulation shown below uses 76 individual carrier tones with a total combined input power of -11 dBm spanning from 30GHz to 31GHz for a total bandwidth of 1000MHz. Harmonics calculated up to the third order and thermal noise are included. Tones below -150dBm are ignored. After passing through the necessary system path, the IF frequency, shown on the second plot, becomes 4500MHz which can be directly sampled by the AD9082.

The two tone spectral simulation shown below uses 2 individual carrier tones with a total combined input power of -11 dBm centered around 30.5GHz to 31GHz and a delta of 50MHz. Harmonics calculated up to the third order and thermal noise are included. Tones below -150dBm are ignored.

As mentioned above, the system cascade simulation measures gain (CGAIN), noise figure (CNF), third input third order intercept (EIIP3), input 1dB compression point (EIP1DB), minimum detectable signal (MDS), and carrier to noise distortion ratio (CNDR). The x-axis shows all of the individual components cascaded together and the nodes in between. The variables are shown on two separate y axes for scaling purposes. The markers are placed on “Node 10” which comes directly after the digital equalizer.

The following simulations were created using Keysight SystemVue/Genesys. All components in the RX signal chain were modeled with frequency-dependent sys-parameter or s-parameter datasets to ensure accurate simulation results. The signal chain below takes into account the following components:

• Manufacturer-provided, frequency-dependent s-parameter or sys-parameter datasets for every active and passive RF component

• Frequency-dependent component interconnect modeling i.e. PCB trace loss approximation for DC-50GHz CPWG traces, 30mil Rogers 4350 PCB stackup

The Genesys workspace incorporates various system level cascade analyses that takes advantage of MATLAB equation-based frequency planning. Three different analyses are shown below:

• Noise power Spectral Simulation

• Two-Tone Test Spectral Simulation

• System Cascade highlighting gain (CGAIN), noise figure (CNF), third input third order intercept (EIIP3), input 1dB compression point (EIP1DB), minimum detectable signal (MDS), and carrier to noise distortion ratio (CNDR)

Simulations below use the ADAR3001 definition for channel gain. Channel gain is the power ratio between a single RF output and single RF input, with all other inputs terminated with 50Ω, DSA code = 0, and TDU code = 0. In a real scenario, coherent gain would be used and is the power ratio between a single RF output and single RF input, with the other three inputs driven. All inputs would have similar amplitude and would be phased such that they combine coherently. For ADAR3001 coherent gain (dB) = channel gain (dB) + 12 (dB). For ADAR5000 coherent gain (dB) = channel gain (dB) + 6 (dB).

The noise power spectral simulation shown below uses 76 individual carrier tones with a total combined input power of -40 dBm spanning from 30GHz to 31GHz for a total bandwidth of 1000MHz. Harmonics calculated up to the third order and thermal noise are included. Tones below -200dBm are ignored. After passing through the analog beamforming front end, the signal is ready for further conditioning by the wideband receiver section.

The two tone spectral simulation shown below uses 2 individual carrier tones each with an input power of -40 dBm centered around 30.5GHz to 31GHz and a delta of 50MHz. Harmonics calculated up to the third order and thermal noise are included. Tones below -200dBm are ignored.

As mentioned above, the system cascade simulation measures gain (CGAIN), noise figure (CNF), third input third order intercept (EIIP3), input 1dB compression point (EIP1DB), minimum detectable signal (MDS), and carrier to noise distortion ratio (CNDR). The x-axis shows all of the individual components cascaded together and the nodes in between. The variables are shown on two separate y axes for scaling purposes. The markers are placed on “Node 10” which comes directly after the digital equalizer.

The following simulations were created using Keysight SystemVue/Genesys. All components in the RX signal chain were modeled with frequency-dependent sys-parameter or s-parameter datasets to ensure accurate simulation results. The signal chain below takes into account the following components:

• Manufacturer-provided, frequency-dependent s-parameter or sys-parameter datasets for every active and passive RF component

• A standalone model of the ADMFM2000 microwave downconverter containing signal chains each with their own frequency-dependent s-parameter or sys-parameter datasets

• Frequency-dependent component interconnect modeling i.e. PCB trace loss approximation for DC-50GHz CPWG traces, 30mil Rogers 4350 PCB stackup

The Genesys workspace incorporates various system level cascade analyses that takes advantage of MATLAB equation-based frequency planning. Three different analyses are shown below:

• Noise power Spectral Simulation

• Two-Tone Test Spectral Simulation

• System Cascade highlighting gain (CGAIN), noise figure (CNF), third input third order intercept (EIIP3), input 1dB compression point (EIP1DB), minimum detectable signal (MDS), and carrier to noise distortion ratio (CNDR)

Simulations below use the ADAR3001 definition for channel gain. Channel gain is the power ratio between a single RF output and single RF input, with all other inputs terminated with 50Ω, DSA code = 0, and TDU code = 0. In a real scenario, coherent gain would be used and is the power ratio between a single RF output and single RF input, with the other three inputs driven. All inputs would have similar amplitude and would be phased such that they combine coherently. For ADAR3001 coherent gain (dB) = channel gain (dB) + 12 (dB). For ADAR5000 coherent gain (dB) = channel gain (dB) + 6 (dB).

The noise power spectral simulation shown below uses 76 individual carrier tones with a total combined input power of -40 dBm spanning from 30GHz to 31GHz for a total bandwidth of 1000MHz. Harmonics calculated up to the third order and thermal noise are included. Tones below -200dBm are ignored. After passing through the necessary system path, the IF frequency, shown on the second plot, becomes 4500MHz which can be directly sampled by the AD9082.

The two tone spectral simulation shown below uses 2 individual carrier tones with a total combined input power of -11 dBm centered around 30.5GHz to 31GHz and a delta of 50MHz. Harmonics calculated up to the third order and thermal noise are included. Tones below -150dBm are ignored.

As mentioned above, the system cascade simulation measures gain (CGAIN), noise figure (CNF), third input third order intercept (EIIP3), input 1dB compression point (EIP1DB), minimum detectable signal (MDS), and carrier to noise distortion ratio (CNDR). The x-axis shows all of the individual components cascaded together and the nodes in between. The variables are shown on two separate y axes for scaling purposes. The markers are placed on “Node 10” which comes directly after the digital equalizer.

Home