RFI-5G Design

RFI-5G payload designs. Includes electrical, mechanical and software.

RFI-5G Payload block diagrams

FE Block Diagram

Power Supply Block Diagram

Signal Conditioning Block Diagram

 

4. General block diagram

 

RFI-5G Payload design

IGARSS_2023_Kband_RFI_Paper.pdf

Payload Structure

The payload is structured as a vertical stack of PCBs, like the rest of the satellite. This is done in order to be self-contained and to comply with the IEEE Open PocketQube Kit standard. 

This stack is designed so that the interferences from upper layers cannot penetrate down to the lower side and so the rest of the PocketQube. This isolation is mainly achieved thanks to the Structural piece, the aluminum shield, which connects the grounding from the PCBs in contact with it on both sides, so that it creates a grounded case, creating a Faraday Cage that is able to enclose the interferences radiated from within while also isolating those from the radiation coming from the exterior.

Another relevant structural component is the FR-4 Structural PCB. The only purpose of this PCB is to support the K-Band Antenna, which is very delicate and thin.

In both cases the Antenna and Top PCB with the FR-4 Structural and the Bottom PCB with the Interface they are attached with epoxy glue and soldered pins when possible. This ensures a good rigidity and a proper performance when faced the important vibrations produced during launch.

Payload Pinout

Schematics

Interface PCB

The Interface PCB is in charge of doing 3 main things:

This is why the interface schematic if divided into three parts.

low_freq_board.png

LDOs

This stage is in charge of regulating the 3.3V voltage line from the PQ to the various required voltage needed in the payload.

Voltage
 Target component
2.5V
 IF1Amp
2.7V
 LNA2
2.75V
3V
  

To be finished

The schematic below shows the design of the voltage regulation circuit for the payload. The purpose of this circuit is to provide stable, regulated power at various voltage levels required by different components within the payload, which is essential for reliable and accurate signal processing and data collection.

Now an in-depth look at the decisions made in the design will be done:

1. Voltage Regulators (U4 and U5 - MIC2215 Series)
2. Decoupling Capacitors
3. Voltage Dividers and Feedback Networks
4. Separate Power Lines for Isolation

low_freq_board-LDOs.png

Voltage Boosters

This schematic provides a look at a voltage-boosting circuit. Here, three independent boost converter stages are built around the LT3048IDC-TRMPBF component (U1, U2, and U3). Each of these stages is responsible for generating a specific voltage required by various components in the payload, focusing on boosting an input voltage to a higher, stabilized output for sensitive RF systems.

Here is an expanded explanation of the design decisions and key aspects:

1. Boost Converter ICs (U1, U2, and U3 - LT3048IDC-TRMPBF)
2. Input Filtering Components (L1, L2, L3, C24, C25, C27)
3. Output Voltage Feedback and Resistor Networks
4. Output Decoupling and Stabilization (C3, C4, C7, C8, C11, C12)
5. Bypass Capacitors for Noise Reduction
6. Application-Specific Voltage Rails
7. Isolation Between Voltage Rails

low_freq_board-Boost1.png

Instrumentation Amplifier

This schematic shows an instrumentation amplifier stage used to process the voltage output from the RSSI (Received Signal Strength Indicator) block, which holds the scientific data from the payload. The amplifier conditions this signal so that it can be accurately digitized by the DAC input of the satellite's microcontroller.

Below is an expanded explanation of the main components and design choices within this stage:

1. Instrumentation Amplifier (U6 - AD8224ARZ-R7)
2. Gain Resistors (R15, R16, R17, and R18)
3. Input and Output Filtering (C21, C22, C26)
4. Voltage Divider for Signal Conditioning (R19, R20, R21, R22, R23)
5. RSSI Input and Output Connections

low_freq_board-Inst_amps.png

Bottom PCB

1. IF Input
2. First Amplification and Bandpass Filtering
3. Mixing

This is the most important part in the payload design

4. Second Amplification and Filtering
5. Additional Bandpass Filtering (SAW Filter)
6. RSSI Measurement

RFI-5G_RF_Bottom.png

Top PCB

RFI-5G_RF_Top.png

PCBs

This payload is structured in 4 main block as stated in the Payload block diagram page. We will start by explaining the Interface PCB.

Interface PCB

This board is charge of electrically connecting the payload to the rest of the satellite, redistributing the grounding and converting it to a Multiple Point Ground (MPG), more specifically into two points, a single pin on the bottom and a multiple consecutive pin array on the top. 

Bottom PCB

This board takes the IF signal from the Top PCB and downconverts it to be captured by the RSSI and output a voltage proportional to the signal power.

Top PCB

This board takes the signal from the antenna at the K Band and downconverts it to a Intermediate Frequency (IF).

Antenna Design

INTRODUCTION

The goal of PoCat payload 3 is to identify Radio Frequency Interference (RFI) in the lower Ka-band (24-25 GHz). A patch antenna is chosen as the resonant element for its lightweight, 2D design, and reduced size due to the inverse frequency-wavelength relationship. A basic Microstrip Patch antenna consists of a conductive patch on one side of a dielectric substrate, with a ground plane on the other, emitting radiation through fringing fields at the patch edge.

Requirements and specifications image 1.png

The design process is split in two main tasks. First, the single element design, includes choosing the appropriate substrate material, determine the patch antenna dimensions and the feedline that will match the patch with the rest of the array, and secondly, the design of the feeding network and the patches’ distribution.

SINGLE ELEMENT: MICROSTRIP PATCH ANTENNA

SUBSTRATE

To optimize antenna performance, the choice of substrate is crucial. A low dielectric constant (𝜀r) is preferred for minimal losses, improved efficiency, broader bandwidth, and better radiation. Increasing substrate height or reducing permittivity can also increase the bandwidth, but it may introduce undesired radiation and coupling to other components due to surface waves in the substrate. The loss tangent (tan(𝛿) or Df) is another main parameter affecting losses, with higher values indicating more dielectric absorption and losses. Rogers-5880, with a Df value of 0.0009 at 10GHz, is a suitable choice due to its low loss characteristics.

SINGLE ELEMENT DIMENSIONS

The initial patch dimensions are given by the following equation, which gives a patch side of ~4 mm using the Rogers-5880 substrate at a 24.5 GHz frequency.

image 2.png

The length of the patches may be changed to shift the resonances of the centre fundamental frequency of the individual patch elements. The resonant input resistance of a single patch can be decreased by increasing the width of the patch. This is acceptable if the relation between the patch width and patch length (W/L) does not exceed 2 since the aperture efficiency of a single patch begins to drop, as W/L increases beyond this value.

Patch antenna main parameters:

Imagen3.png

PATCH ANTENNA MATCHING

In this project, the chosen method is to use a microstrip line attached at the edge of the patch, combined with adding an inset into the patch itself as shown in the following figure.

Imagen4.png

The longitude of the inset R can be calculated with the following equation where Zin(R) is the matching impedance and Zin(0) is the impedance at the edge of the patch (if the patch was fed in the end).

Imagen12.png

Hence, by feeding the patch antenna as shown, the input impedance can be decreased to tune it at the desired value. The spacing S is a more challenging parameter to estimate, and it requires the simulation software to define it.

2X2 PATCH ANTENNA ARRAY

The last step in the design process of a 2x2 patch antenna is to design the entire feeding network and the patches distribution. Knowing this field has been exhaustively studied, some made designs were considered at the beginning of this project, and researching about this technology, a design of a 24 GHz lineally polarized patch antenna, with the inset impedance matching technique, was found in an online repository from Zhengyu Peng, Ph.D. (Texas Tech University). In this manner, a reverse engineering study has been carried out to understand this design and then adapt it further to our requirements.

Array model and electrical lengths of the feeding network:

Imagen5.png

By adding an extra 180º length to access the patches from the bottom of the structure, the waves from the two resonant elements are adding up to each other. In addition, the corner mitering technique has been used as well.

Mitred bend:

Imagen6.png

A 90º bend in a transmission line adds a small amount of capacitance to the transmission line, which causes a mismatch. A mitred bend as shown in the previous figure reduces some of that capacitance, restoring the line back to its original characteristic impedance. On the other hand, the whole structure impedance matching is done by two quarter-wave impedance transformers, adjusting the trace’s width to obtain the desired characteristic impedance that sets 50 Ω at the output of the network.

λ/4 impedance transformers:

Imagen7.png

The λ/4 impedance transformer design equation is included below, where Z0 is the characteristic impedance of the line, Zin the input impedance and ZL the impedance of the load.

Imagen13.png

Each line has been designed to match this equation according to the required input and output impedance:

Imagen14.png

CONNECTOR

The antenna will be positioned on the upper side of the satellite. Therefore, the dimensions of the substrate layer are determined by the PoCat's own dimensions, which measure 40x40 mm². To enable the assembly of the entire structure, the payload's board includes holes for pins and screws. Additionally, a HIROSE C.FL connector is attached to the top section of the structure, where it connects to the 50 Ω feeding line. This allows easy access for the coaxial cable to connect to the upper layer from below, requiring the incorporation of a hole into the substrate and ground plane, placed at the farthest point compatible with elements from payload 2.

Connector mounting pattern following HIROSE’s recommendation:

Imagen9.png

FINAL 2X2 PATCH ANTENNA ARRAY MODEL

Imagen10.png Imagen11.png

Data Acquisition

Parameters:

struct rfi5gParameters{
  char* name;
  uint32_t timestamp;
  uint32_t temperature;
  uint32_t startFrequency;
}

Constants

General Constants

#define N_SAMPLES 1000 
#define NUM_FREQ 62 
#define R 10

N_SAMPLES: Number of samples to capture in one frequency bin.

NUM_FREQ: Number of frequency bins.

R: Number of repetitions for the calibration step.

Statistical Constants

#define KU_UPPER_THRES 3.3 
#define KU_LOWER_THRES 2.7 
#define SK_UPPER_THRES 200 
#define SK_LOWER_THRES -200

Upper and lower thresholds for kurtosis and skewness.

Notice that skewness is multiplied by 1000.

Time Domain Constants

For the time domain the threshold is defined as th = βσ + m

#define AGGRESSIVENESS_TIME 1.5 
#define PERCENTAGE_TH 0.01 
#define K 1.3339

AGGRESSIVENESS_TIME: Aggressiveness Time (aka β) is an empirical value obtained by simulating different INR values.

PERCENTAGE_TH: Percentage Threshold is the minimum percentage to consider the existence of an interferent signal, it is defined as the minimum number of detection out of N_SAMPLES.

Frequency Domain Constants

#define AGGRESSIVENESS_FREQ 1.5

AGGRESSIVENESS_FREQ: Aggressiveness Frequency is an empirical value obtained by simulating different INR values.

Structures Definition

typedef struct{
	uint16_t mean;
	uint16_t stdv;
	uint16_t ku;
	int16_t sk;
	uint16_t Ndet;
	uint8_t decision;
} tRFIbinCapture;

mean: Mean is a statistical parameter that shows the central tendency of the distribution.

stdv: Standard Deviation is the square root of the second standardized moment, it measures the dispersion of the distribution.

sk: Skewness, also known as the third standardized moment, is a metric that characterizes the asymmetry of the distribution.

ku: Kurtosis is the fourth standardized moment, that measures how big are the tails of the distribution.

Ndet: Number of detections counts the number of samples that had surpassed the threshold in time domain algorithm.

decision: Decision is an 8 bits variable, which flags with a pair of bits if an alogorithm has detected an interference.

From the least significant bit to the more significant bit, the flags are grouped in pairs:

-The first two bits are turned on if the statistical domain algorithm detects an interference.

-The second pair of bits are activated if the time domain algorithm detects an interference.

-The third pair of bits are activated if the frequency domain algorithm detects an interference.

-The fourth couple of bits is the combination of the previous decisions, if any algorithm has detected an interference these two bits are turned on.

Due to the possibility of ions or electromagnetic radiation to strike into the electronic circuits, causing an unwanted bit change, commonly refered as single event upset (SEU), each decision bit has a redundancy of one.

typedef struct{
	uint16_t id;
	tRFIbinCapture captures[NUM_FREQ];
} tRFIsweepCapture;

tRFIsweepCapture: This struct consists in the frequency axis data, since the system does not do a continuous frequency sweep, the axis is discretitzed in NUM_FREQ bins.

id: It is the identifier of a capture.

Comms considerations

As of now this section is in development

Non mandatory data:

RFI-Dedicated Thread

RFI Thread

Introduction

Like many other subsystems, the Payload is a self-contained sub-system and so it has its own FreeRTOS Thread. This thread is concieved to be able to control each of the two RFI monitoring payloads.

As an overview, the thread is structured as secuential general operations. And each of them is a key step in the process of making the payload work.

Block Diagram

To begin, the general behavior of the code is defined, since it is basic to have it structured from the most generic functions to the most specific ones. This general structure is basic since it is the one that will be in direct ”contact” with the OBC code, and its structure and functioning requirements.

This figure, as stated above, defines how the payload code is structured and its sequential steps. Note that the discontinuous arrow indicates that once the loop has finished, the payload thread enters a sleep mode, waiting for the OBC to awaken him again to restart the loop. So, the first to take a deeper look into is going to be the Check start conditions block:

As it can be seen above, the first thing to do in order to get the payload running properly is to check if the conditions required by the Ground Station user are met, this includes three things:

  1. The satellite is pointing at the desired region of the Earth steadily.
  2. The order to make an RFI measurement has been truly received.
  3. It is the scheduled moment to take the photo.

If none of these requirements are met, the thread awaits until it is signaled by the OBC and then rechecking everything again. Once the check has been succeed, it is time to Start the electronics and the timer:

Obviously, the first step is to power on the payload and consequently all its elements. Then the ADC and DAC are initialized and started. The same happens to the timer. Once they have been started it is time to make sure that they work as expected:

Now the ADC and DAC are checked and calibrated, the timer is only checked. This is done to assure that when using them in the measurements they don’t give false reading due to a malfunction. Having everything powered on, initialized and checked, the measure loop can begin:

This loop is run as many times as needed in order to scan all the frequency bins forming the L or Ka bands. This scans are slow, so not to block the whole satellite system, every time a frequency bin is scanned the thread checks for notifications coming from the rest of the PQ to pause itself if necessary. Then, once the satellite does not need to have the RFI thread stop anymore, the RFI thread continues from where it was.

In order to bring the explanation to a lower level, the list below explains each block step by step:

Now that the measurements have been done, it is time to power off this subsystem:

So, first both ADC and DAC are stopped, then the payload is powered off and then the timer is stopped.
To finish the thread execution, the end of the payload work is signaled to the OBC and then the thread enters sleep mode.

Detection algorithms

Temporal algorithm

The aim of this sub-algorithm is to detect pulses through Envelope Detection, a straightforward technique used for RFI monitoring. It involves sampling the input signal and comparing its power with a predetermined threshold. The detector continuously observes the power of the received signal in each temporal sample and establishes a threshold based on the typical noise floor level. When the power level of the samples exceeds this threshold, the detector interprets it as a presence of RFI.

Although the Envelope Detection algorithm has been chosen, other time domain detection algorithms exist such as the FFT (Fast Fourier Transform) based ones. FFT is the most important calculation when talking about time sampled signal processing because it allows to not only analyse the signal in the temporal domain but also in the frequency one. And by combining these two ones, advanced algorithms can be elaborated.

In addition to that, despite the very limited processing capabilities of the microcontroller used in the satellite, some advances are being done. Some preliminar tests have already proven the possibility to compute FFTs in that microcontroller. In the following weeks the integration of those algorithms with the rest of the code will be tried and tested.

Now the mathematical algorithm will be explained. First, the temporal samples vector, expressed as: 𝑥[𝑛];being N the number of samples taken Is iterated in order to calculate its mean:

image.png

Finally, if the 𝑥 is higher or lower than an arbitrary threshold the decision of whether there is in fact a signal or not is taken. This threshold is the most important part of the algorithm and its value highly affects the accuracy of the decision.

Statistical Algorithm

The purpose of this sub-algorithm is to estimate if there are signals based on their third and fourth statistical moments. These moments are extracted from the temporal samples taken in each frequency bin.

However, there are many other statistical parameters that can be taken into account, such as the standard deviation, the covariance or the correlation. Unlike the reason above, these parameters will not be used because they are not considered necessary with the given skewness and kurtosis reliability.

Now the kurtosis formula will be explained:

image.png

Where: 𝑥𝑖 is the i-th sample.
In the numerator of the fraction there is the calculation of the fourth statistical moment, while in the denominator there is the standard deviation elevated to the fourth power. Now the skewness formula will be explained:

image.png

Where: 𝑥𝑖 is the i-th sample.
In the numerator of the fraction there is the calculation of the third statistical moment, while in the denominator there is the standard deviation elevated to the third power.

Again, as explained in the time domain algorithm, the decision of the statistical algorithm is based on the comparison between the statistical parameters and predefined thresholds that, if exceeded or not, a decision is then made.

Frequency Algorithm

The aim of this sub-algorithm is to detect signals in the frequency domain and eventually mitigate them by frequency blanking. This technique operates in a similar manner to Envelope Detection, but in the frequency domain. If a significant increase in power is observed at a particular frequency bin compared to its neighbouring bins, it may indicate the presence of interference. These power peaks are more easily detected when they have high power in a narrow bandwidth.

This analysis is done by comparing the received power from adjacent frequencies and applying a threshold to distinguish these values. So, if the algorithm detects a signal that is above that threshold, it decides that it is in fact an Interference.