P20151: Satellite Localization
/public/

Build & Test Prep

Table of Contents

Team Vision for Build & Test Prep Phase

Our goal this phase was to update the project plan with specific dates for critical system tests, to finish ordering parts, to redesign the exterior box, and to play around with our hardware.

Received Parts:

Shipping:

In Process:

The parts for one station are ordered or processing with the MSD Office. The current remaining budget is $3,885.

We are still redesigning the box. The project plan now contains dates for all critical tests. We better understand the hackRF SDR and software that uses it.

Test Plan Summary

Oour test plan is available in 2 forms:

  1. Excel Visual document. This document contains critical system tests.
  2. Complete Word doc test plan. This document contains a complete list of all tests for the LASSO project.
  3. Please see the project plan for a detailed review of when tests will occur and a summary of what their key components are.

Status Updates

Storage and Orders

We received a storage container for containing our hardware. We already had access to the EE senior design lab. We have been persistant with the MSD office, making sure all our ordered parts were actually ordered. In one instance, the Raspberry Pi was marked ordered, but never actually ordered.

Box Redesign

 Photo of Product we ordered for our system.

Photo of Product we ordered for our system.

The new housing is made of steel to address concerns regarding ESD from the previous plastic design. We have found a product that suits our needs. Being able to buy the housing is a good thing to help keep the ground stations modular so other universities will be able to adopt our design. This housing is approximately 16x12x6 inches and will accommodate all necessary components. The Uninterrupted Power Supply will not fit inside the box, so mounting that is still a challenge we need to figure out. This new enclosure exceeds team LASSO’s weather proof requirements. Originally we had wanted a system that has an IP rating of 5-4, meaning that our system is splash proof (rain) and keeps out almost all dust particles. Now, it will have an IP rating of 6-6. This means it is completely dust proof, and has protection against direct high pressure water jets. Modifications regarding ventilation may need to be made. This will be determined after preliminary heat testing.
 Converted Product to CAD file.

Converted Product to CAD file.

After finding our desired product, we converted it to a PTC Creo file to incorporate into our final design. This product does have a key and lock. This is beneficial as the ground stations will most likely be left in remote locations. Security for the electronics is very important.

 Inside view of the ground Station

Inside view of the ground Station

 Inside view of the ground station

Inside view of the ground station

The new housing is equipped with bonding studs to protect components from electrical discharge. Components are to be mounted directly to the designated steel plate inside the protective housing. When components are connected to the same steal plate within the housing, any build up of static electricity from the components will be dissipated via chassis ground. There is ample room for wired connections to happen. Part of the redesign was having more space to layout the wires to make working with the components much easier. Current risks are ventilation and waterproofing. If we need to add ventilation for the fans, then we risk the integrity of the systems waterproofing. Testing will be done to determine whether any sort of cooling will need to be made.

 Wiring Diagram for electrical components inside the ground station.

Wiring Diagram for electrical components inside the ground station.

This is a simple wiring diagram for the electrical components that are placed inside the box. Currently, we are still unsure if fans will be needed. However, they are included in here just in case. There are some connections that are not pictured, and that is because they are outside the box. Connections that are not pictured are: AC Plug into UPS, Raspberry Pi to GPS, LNA to Lightning Arrestor, Lightning Arrestor to ground, and Lightning arrestor to Antenna.

TDoA

Since the detailed design, we focused on making TDoA faster and more robust. TDoA now works with 3 different solvers:
  1. The symbolic solver: slow, but most accurate
  2. Mininmum Distance to Hyperbolas: least squares, fast, accurate
  3. Most similar time difference: least squares, fastest, least accurate

A Monte Carlo for 10 triangles takes several hundred hours for the symbolic solver, but only ~2 hours for the either least square solver.

Least squares works by minimizing a cost function using Matlab's optimization toolbox.

  1. The first cost function tries to minimize the solution's distance to each hyperboloid. If the solution is on a hyperboloid, it incurs no cost from that equation. If the solution is on all three hyperboloids, its cost is 0.
  2. The second cost function computes what the time differences would be, given the solution. The cost function is zero if these time differences are equivalent to the measured time differences.

We can always solve these cost functions, both in scenarios where the TDoA hyperbolas on a plane enclose a triangle and when they do not.

 The symbolic solver and the two least square methods converge to different locations within the enclosed TDoA Triangle.

The symbolic solver and the two least square methods converge to different locations within the enclosed TDoA Triangle.

 Least squares solvers can get a solution even when TDoA does not form a triangle. This scenario is common at low elevations.

Least squares solvers can get a solution even when TDoA does not form a triangle. This scenario is common at low elevations.

Because the least squares methods can solve problems where the symbolic solver cannot, the least square methods have roughly 10% more coverage of the sky.

Since the previous review, we also fixed a bug with Azimuth error. The code was recording azimuth measurements of (1,359,357,3 degrees) as 180 degrees and not 3 degrees. This has decreased the Azimuth IQR by ~40%.

 Monte Carlo Results for the symbolic solver with the Azimuth error bug, Copied from the Detailed Design Review.

Monte Carlo Results for the symbolic solver with the Azimuth error bug, Copied from the Detailed Design Review.

 Using least squares, the larger triangles have ~10% better coverage and about 60% of the IQR in Azimuth.

Using least squares, the larger triangles have ~10% better coverage and about 60% of the IQR in Azimuth.

In instances when we do not have the correct answer, we need a method to estimate the uncertainty of the TDoA Solution. Using a Monte Carlo approach, we perturbate the inputs and create distributions of the outputs. We then assume a normal distribution and estimate the uncertainty as 2 standard deviations of that distribution.

We use 2 standard deviations because that uncertainty would explain 95% of the data. The normal assumption works well for medium and high elevations; however, it fails at low elevations. At low elevations, the normal assumption breaks down and we estimate uncertainties much higher than the actual error. This is the reason we see a 30% drop in uncerainty coverage across the triangles when estimating the uncertainty.

 The dark blue dots represent a distribution of possible reference points. The light blue lines represent a distribution of possible directions. Both are roughly Gaussian.

The dark blue dots represent a distribution of possible reference points. The light blue lines represent a distribution of possible directions. Both are roughly Gaussian.

 The estimated uncertainty is much higher than the absolute uncertainty calculated by the Monte Carlo.

The estimated uncertainty is much higher than the absolute uncertainty calculated by the Monte Carlo.

 Notice the estimated uncertainty (blue) overlaps well with the absolute error (red) at high and median elevations.

Notice the estimated uncertainty (blue) overlaps well with the absolute error (red) at high and median elevations.

 Notice the estimated uncertainty (blue) overlaps well with the absolute error (red) at high and median elevations.

Notice the estimated uncertainty (blue) overlaps well with the absolute error (red) at high and median elevations.

 This point cloud is nonnormal. Assuming a normal distribution is no longer valid.

This point cloud is nonnormal. Assuming a normal distribution is no longer valid.

For a comprehensive overview of the TDoA Algorithm, see the TDoA Presentation.

Other Software

We started working on the server program (in C++), the signal recorder (in Python), and the access calculator (in Orekit, Java). See the Subsystem Build & Test for more details.

Risk and Problem Tracking

Our updated risks can be found in the Risk Management document.

Cabling resdesign has manifested itself as a problem. Details on the current status of the cabling redesign can be found in the problem tracking document.

Design Review Materials

Progress report presentation

Notes from the review:

  1. We should identify backup vendors in case our chosen vendors cannot process our orders.
  2. We should ensure a 2nd set of hardware even at this stage in case something breaks.
  3. Make sure we do not design out satellite hunting.
  4. Fans could be controlled by a $4 controller + temperature probe. Fans may need a weather hood for "weatherproofing"
  5. We should have a key to the box.
  6. Box should have easy disconnects with external wires
  7. Power management may be a blindspot. Can a solar panel / battery be used, if desired? Do we need a op amp power supply?

Plans for next phase

Our vision for the subsystem review is to complete all tests associated with one station and a complete one test associated with the hardware of 2 stations.

Three Week Plans

  1. Andrew
  2. Anthony
  3. Zach
  4. Luca
  5. Zoe
  6. Matthew
  7. Connor

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

https://creativecommons.org/licenses/by-nc-sa/4.0/


Home | Planning & Execution | Imagine RIT

Problem Definition | Systems Design | Preliminary Detailed Design | Detailed Design

Build & Test Prep | Subsystem Build & Test | Integrated System Build & Test | Customer Handoff & Final Project Documentation