This is an update that focuses on the SDR hardware used in receiving Starlink downlink signals. For the original post, please see HERE. Update # 1 can be found HERE.

Starlink receiver. Closed case.

Introduction

When I originally started on this project, I swiftly came to the conclusion that commercial, off-the-shelf SDR platforms capable of > 240 MHz receive were incredibly cost prohibitive. Although many of these platforms support cool features like MIMO, most do not extend beyond 160 MHz of instantaneous bandwidth. I own a couple of AD936x based USRPs (B210, B200mini), but these fall short at a maximum bandwidth of 56 MHz. (Great for their target market of LTE small cells, but not gonna cut it here.) The USRP X300 + UBX combo achieves dual 160 MHz TX/RX, but costs > 10k USD once you include the chassis, daughtercards and desktop. The new and shiny RFSoC based USRP X410 meets the bandwidth requirement and does so while costing as much as a new Toyota Corolla. To be clear, I believe the Ettus USRPs are a phenomenal value, and I am a very big fan and happy customer.

The conclusion was clear: I’d have to build my own wide bandwidth SDR platform. Furthermore, I will make decisions to keep the cost “down”, and for any parts I already own that I would not otherwise have selected, I will suggest cheaper alternatives. Although I intend to use this receiver for receiving Starlink downlink, I will future-proof the design with much greater dynamic range and frequency flexibility than what is required of this use-case.

Design

I selected a direct-conversion architecture for the receiver. This is a sensible option for a frequency flexible, wideband SDR: easier RF filtering and impairments are manageable. To facilitate high bandwidth sample streaming into my host, I opted to use PCIe between the FPGA and host processor. I didn’t want to fiddle with expensive external PCIe cable solutions or Thunderbolt docks, so I decided to build a mini-ITX host into the rear of the enclosure - simila

I’ve started work on a “Starlink observatory” with the primary objective being: to track overhead Starlink satellites, collect physical layer statistics and enable further signal analysis. Focus will be on the Satellite-to-User downlink.

So far, I have built and automated a setup to collect signal captures of the downlink 240 MHz traffic in the 10.7 - 12.7 GHz KU band. My setup consists of a KU band dish + LNB, a software defined radio capable of sustained 400 Msps (complex) receive, an LNB controller + power supply, a Linux host to perform signal analysis and all the miscellaneous RF plumbing required (filters, bias-tee, attenuators, etc). Presently, this setup is running day and night in my backyard+garage listening for Starlink downlink traffic and storing a subset to disk for further offline analysis. These files are really quite big - I don’t keep them around for long.

This is all still pretty early, but a couple observations:

(1) I have received Downlink Satellite-to-User traffic in the 250 MHz channels at: 11075, 11325 and 11575 MHz. I have never observed traffic at the lowest channel, 10825 MHz. Although my setup supports it, I have never looked at the upper (4) 250 MHz channels. Occupied signal bandwidth per channel is 240 MHz.

Frequency domain. Full 240 MHz Downlink signal capture on STAR-2471 @ 11575 MHz. Capture is 10 milliseconds.

(2) The central 1 MHz of each 250 MHz channel is occupied by ~9 tones spaced at 43.9495 kHz. 1 tone on the channel center, 4 tones on each sideband. I’m fairly sure I know what these are (quite exciting), but I’ll wait until I’ve done more due diligence.

Frequency domain. \"Zoomed in\" to the central 6 MHz span.

(3) When looking at the captured signal in the time-domain, a frame structure is visible. I have made multiple captures with the same frame “raster” and timings, but with different frame allocations. I’ve observed frame structures with full and partial frame occupation. This is likely a function of the required downlink capacity at any given time. For lack of a better name at this time, a “frame” is 1.3288 milliseconds. Each “frame” is spaced by a “guard” interval of 4.55 microseconds.

[Time domain. Zo

This is a follow-up to my original post on this topic. Please see HERE.

This last weekend, I embarked on an effort that I initially thought was just “due diligence”. I wasn’t expecting any surprises or revelations.

Along with each capture of a Starlink downlink signal, I also wanted to deduce exactly which Starlink satellite I was receiving at that precise moment in time. I was able to accomplish this using NORAD TLE datasets from CelesTrak. These datasets contain orbital model parameters that can be used to project the position of a satellite at a given time. In conjunction with an observer’s latitude and longitude, the azimuth and altitude can be calculated.

I did not know exactly the altitude/azimuth of my dish. Nonetheless, given its 24° offset angle, I had a “reasonable” guess of 60° / 25°. This “guess” and my latitude/longitude were inputs into my code. Upon detecting a signal, my code would estimate the altitude/azimuth of all Starlink satellites at the exact moment of start of reception. Then, I would convert to a 3-D unit vector and compute the error with respect to my “guess” observer unit vector. I record the 5 “most probable” satellites.

I added this new feature to my receive processing pipeline, and I let it run Sunday-Monday.

This was quite successful! My dish has a 3° beam width, and the timestamp used to project the Starlink orbits wasn’t exactly at the peak of the capture. This data would imply that the dish is pointing at ~ 56° / 33°.

26 identified satellites

However, not ALL signals I received could be explained by the Starlink NORAD dataset. 26 captures were clearly identifiable as specific Starlink satellites. 11 were unexplained.

11 unidentified satellites

As can be seen from the table above, the (5) "most probably" satellites aren't anywhere near where my dish is pointing.

A couple observations:

1. The “identified” satellites were only transmitting at 11.325, 11.575 GHz. The “unidentified” satellites were ALSO transmitting at 11.075 GHz.
2. All the “identified” signals

Hi folks,

I've been browsing for the last bit and have been noticing an increase in posts talking about water-leak monitors and temperature devices. (Maybe related to the radio usage for energy monitoring?) I thought I would add in some of my own insight on devices I've recently setup since I think some of the recent suggestions are a bit expensive.

By using a software defined radio, we can get some pretty great devices at reasonable costs. Some of them include:

• $45: 5x (five!) Govee H5054 Water-Leak detectors
• $13: AcuRite 06002M outdoor temp/humidity sensor
• $37: AcuRite 00986 refrigerator thermometer display with 2 sensors

My preferred radio is:

• $40: Nooelec NESDR Smart v4 Bundle (can probably get something cheaper but I wanted a decent antenna)

All of these devices work with the rtl_433 project and communicate to HA over mqtt. They all have excellent range (at least with the antenna I'm using).

Since I'm a big fan of devices that _don't_ require internet to continue functioning, these are all highly recommended by me.

For those interested in the execution and rationale, I wrote a post which includes sensor templates, automations, and blueprints.

https://www.kyleniewiada.org/blog/2021/10/affordable-water-leak-and-temp-monitoring/

Amazon and AliExpress are selling few cheap software defined radio (SDR) USB modules below $30 USD. Some SDR apps are available on Android smartphone. So I guess that many gaming handheld with Android OS and USB port can use such RTL-SDR radio tuner (receiver) for receiving wide range of wireless signals like AM/FM HF SW SSB DAB ATC as universal portable radio. Or use as simple spectrum analyzer.

Any thought or interest?

Here is the link: https://drive.google.com/file/d/1b1RhNQlqxd9Rp8kHMg52qyc-KAu7Ff2d/view?usp=sharing

I’m going to build this direct conversion SDR this week, but before committing money to it I would like to hear your criticisms/advices.

Thanks in advance for all your comments, I know that what I’m asking is not a small, easy thing, I really appreciate it.

I'm honestly not sure, since the SDR I'm buying (from AliExpress) is capable of transmitting radio waves, however I'm only using it for receiving radio and not transmitting anything

I'm worried about customs clearance

Hi, I wonder if anyone thought about extending Piratebox beyond wifi mesh networks to SDR/GSM ? I am not an expert on the SDR technology but I understood it can be used to build wireless mesh networks as well.

Anyone know if and of the esp32 variants can do software defined radio? I'm mostly interested in HF (shortwave), but it's generally interesting

“Spectrum analyzer.” Sure.

I’m just a nerdy hack complete amateur with an old MacBook, POS antenna, shareware, and an SDR+ dongle. Even I know how to zoom in on a freq and record transmissions.

I can’t imagine why they aren’t recording those transmissions and sending them to actually smart people for analysis.

Unless this is all complete nonsense.