Inner Circle Newsletter: Summer 2024

A Sizzling Summer Summary prepared just for you!

Read on for highlights from all our technical and regulatory open source digital radio work. ORI’s work directly benefits amateur radio, provides educational and professional development opportunities for people all over the world, and puts ethics and good governance first.

Opulent Voice Flying High

Opulent Voice is an open source high bitrate digital voice (and data) protocol. It’s what we are using for our native digital uplink protocol for ORI’s broadband microwave digital satellite transponder project. Opulent Voice has excellent voice quality, putting it in a completely different category than low bitrate digital communications products such as D-Star, Yaesu System Fusion, and DMR.

Opulent voice switches between high resolution voice and data without requiring the operator to switch to a separate packet mode. Opulent voice also handles keyboard chat and digital file transmission. Seamless integration of different data types, using modern digital communication techniques, differentiates Opulent Voice from any other amateur radio protocol.

Opulent Voice will fly on the University of Puerto Rico’s RockSat-X launch on 13 August 2024. It’s been a very positive experience working with the students and faculty at the University.

An implementation on FPGA for the PLUTO SDR is well underway, with a active international team delivering quality results. This implementation will not only turn your PLUTO SDR into an Opulent Voice transceiver, but it will have remote operation functionality.

Hear what Opulent Voice sounds like by following the links in an earlier update at https://www.openresearch.institute/2022/07/30/opulent-voice-digital-voice-and-data-protocol-update/

We’ve come quite a long way in less than two years! The FPGA implementation upgrades the modulation from 4-ary frequency shift keying to minimum shift keying, and increases forward error correction performance and flexibility.

HAMCON:ZION 2024 is This Week!

Please visit us at HAMCON:ZION 2024 this weekend, 12-13 July 2024 in St. George, Utah, USA.

The event website is https://www.hamconzion.com/

ORI will have a club booth at the event. We opened our space to QRZ.com (https://www.qrz.com/) and Deep Space Exploration Society (https://dses.science/). This combined exhibit is a one-stop shop for the best in community, technical, and citizen science amateur radio activity.

We have a volunteer presenting on Artificial Intelligence and Machine Learning in Amateur Radio. The talk opens with a brief summary of the history of our relationship with created intelligence and then explores case studies of the use of artificial intelligence and machine learning in amateur radio. Talk is 1pm on Friday in Entrada B.

Open Research Institute at DEFCON32

We will present an Open Source Showcase at DEFCON in the Radio Frequency Village 12-13 August 2024, with accessible exhibits and demonstrations. Here is the list of scheduled project demonstrations.

Regulatory Efforts: ORI works hard for open source digital radio work and moves technology from proprietary and controlled to open and free in intelligent and mutually beneficial ways. Our work on ITAR, EAR, Debris Mitigation, AI/ML, and Synthetic Aperture Radar will be presented and explained. Find out more at https://github.com/OpenResearchInstitute/documents/tree/master/Regulatory

Ribbit: this open source communications protocol uses the highest performance error correction and modern techniques to turn any analog radio into a digital text terminal. No wires, no extra equipment.. Learn how to use this communications system and get involved in building a truly innovative open source tactical radio service. Find out more at https://www.ribbitradio.org

Satellite: ORI has the world’s first and only open source HEO/GEO communications satellite project. All working parts of the transponder project will be demonstrated, from Opulent Voice to antenna designs.

The Dumbbell antenna: We have an HF antenna design based on a highly effective inductive loading technique first written about in 1958. Learn about this antenna and find out how to make your own. Repository can be found at https://github.com/OpenResearchInstitute/dumbbell

RFBitBanger: an HF QRP system and novel digital protocol called SCAMP. Kit information and updates will be available. Get your Batch 2 kit today at https://www.ebay.com/itm/364783754396

Radar: Our regulatory and technical work in synthetic aperture radar will be demonstrated. One of our volunteers will be giving a talk about open source synthetic aperture radar in the RF Village speakers track. Here is the abstract.

Synthetic Aperture Radar (SAR) is one of the most useful and interesting techniques in radar, providing high resolution radar satellite images from relatively small satellites. SAR is not limited by the time of day or by atmospheric conditions. It complements satellite photography and other remote sensing techniques, revealing activity on the Earth that would otherwise be hidden. How does the magic happen? This talk will explain the basics of SAR in an accessible and friendly way. That’s the good news.

The bad news? SAR is controlled by ITAR, the International Traffic in Arms Regulations, and is listed in the USML, the United States Munitions List. ITAR regulates the export of defense articles and services and is administered by the US State Department. This includes both products and services as well as technical data. Such as, catalogs of high resolution radar imagary. The articles and services regulated by ITAR are identified in the USML. If ITAR doesn’t get you, then EAR just might. The Export Administration Regulations (EAR) are administered by the US Commerce Department, and items are listed on a Commerce Control List (CCL). Commercial products and services and dual-use items that are not subject to ITAR could be regulated by EAR. Even if you are free of ITAR and EAR, you may still be regulated by yet another agency, such as the FCC.

Regulation of SAR chills commercial activity, creating costly and time-consuming burdens. But why does any of this matter to signals hackers? Because technology has overtaken the rules, and devices used by enthusiasts, researchers, students, and hackers are increasingly likely to have enough capability to fall into export-controlled categories. The penalties are harsh. Fear of running afoul of ITAR is enough to stop a promising open source project in its tracks.

Is there a way forward? Yes. ITAR has a public domain carve out. Information that is published and that is generally accessible or available to the public is excluded from control as ITAR technical data. That’s great in theory, but how can we increase our confidence that we are interpreting these rules correctly? How can we use and build upon these rules, so that our community can learn and practice modern radio techniques with reduced fear and risk? Can we contribute towards regulatory relief when it comes to SAR? We will describe the process, report on the progress, and enumerate the challenges and roadblocks.

RFBitBanger Batch 2 Kits Available

Kits are available at our eBay store at this link https://www.ebay.com/itm/364783754396

Be a part of the future with a prototype Batch 2 kit build of the RFBitBanger, a low-power high-frequency digital radio by Dr. Daniel Marks KW4TI. Presented by Open Research Institute, this kit is designed to produce 4 watts of power and opens up a new digital protocol called SCAMP.

SCAMP Is now available in FLDigi!

Source code and extensive instructions can be found at https://github.com/profdc9/fldigi

Your donation in exchange for this kit directly enables the further development of an innovative Class E amplifier based radio design. It has a display, button menu navigation, and keyboard connection for keyboard modes and keyboard-enabled navigation. This radio can be taken portable or used in a case. If you have a 3d printer, then Dr. Marks has a design ready for you to print in the repository linked below.

  • Built-in digital modes: CW, RTTY, SCAMP (FSK and OOK, multiple speeds)
  • Key jack supports straight keys and iambic paddles
  • Open Source hardware and firmware, Arduino UNO compatible https://github.com/profdc9/RFBitBanger
  • External sound-card FSK digital modes supported (including FT4/FT8)
  • Experimental SSB support
  • Serial port support (2400 baud) for send and receive in keyboard modes

SCAMP is a new protocol that allows keyboard-to-keyboard contacts with a digital protocol that has excellent connection performance. See Dr. Marks presentation about RFBitBanger at QSO Today Academy in September 2023 to learn more about SCAMP and the RFBitBanger project. Link below:

All surface mount parts on the main board are pre-installed at the factory. All the through-hole parts you need to complete the radio are provided for you to solder yourself.

Builder’s notes and photos of all the components to help you identify and install them can be found here:

https://github.com/OpenResearchInstitute/RFBitBanger-kit/tree/main/batch2

If you don’t know how to wind toroids or solder surface mount capacitors, this is an excellent kit to learn on. There are just six toroids on the main board, and two on each band pass filter board. You can build just one band pass filter board and operate on a single band, or you can build an assortment. We provide 12 filter boards, enough toroids to build any 9 filters, and a supply of capacitors that will let you build those 9 filters for 9 different HF ham bands. These capacitors are size 1206, which is the largest common size for SMT capacitors and the easiest to solder manually. All you’ll need is a pair of tweezers and your regular soldering iron and solder. We provide detailed instructions on winding the toroids and soldering the capacitors. You get spare filter boards to experiment with.

Friendly Support is provided through a dedicated Open Research Institute Slack channel.

Instructions on how to join this community are here:

https://www.facebook.com/openresearchinstitute https://www.instagram.com/open_research_institute/
https://x.com/OpenResearchIns

Opportunities This Week at ORI

Greetings all!

What do you need help with this week?

Here’s some opportunities at ORI.

1) Pierre and Ahmet are looking for people to help with mobile app design on Ribbit

The Ribbit Radio app is in both Android and Apple testing. The updates to Rattlegram are incorporated and the app is functional on both platforms. We have had excellent response for test teams and things are moving forward.

To make the app as great as it can be, we could use some additional human resources for UX/UI/code development. If this sounds like something you are interested in, please join #ribbit on our Slack or write to me directly and I’ll get you in touch with the team leads. 

2) DEFCON volunteers for the booth/exhibit. We’ve got just enough people to cover it. It’s a great event. We have solid support from RF Village and we advertise to Ham Radio Village. If you have been sitting on the sidelines waiting for a chance to do something for ORI, this is the best event of the year. 

https://defcon.org/ for details about DEFCON 10-13 August 2023
https://wirelessvillage.ninja/ for details about our Village, RF Hackers Sanctuary.

3) FPGA designs for Haifuraiya and Neptune. Want to use MATLAB/Simulink, VHDL, and Verilog to make open source digital communications designs for aerospace, terrestrial, and drones? These designs run on updated FPGA stations in ORI Remote Labs, and everything is on the microwave amateur radio bands. When you see microwave frequencies mentioned, then it’s good to also say that “we use these bands or lose them”. We’ve got plenty to do. Get in touch on #haifuraiya or #neptune on Slack or to any ORI director. 

4) Meander Dipole construction phase. Project Dumbbell explores an overlooked HF antenna design. There’s been strong interest in these designs from multiple people (some of which are on this list), clubs, and organizations. We need to build the designs that MATLAB says look pretty good. Time to make it work over the air and write up some construction and measured performance articles. 
As always, there’s plenty more going on, but these projects have some specific needs, today. 

Thank you to everyone that supports our work. I’d like to especially thank the IEEE and ARRL for being excellent partners to ORI. 

-Michelle Thompson

Project Neptune

Neptune is an open source implementation of an open protocol for high-throughput multimedia drone communications from Open Research Institute. This protocol directly competes against the proprietary designs from DJI.

Implementation prototypes use the Xilinx Ultrascale+ and Analog Devices 9002 RFSOC. These development boards are accessible at no cost through ORI Remote Labs West. Amateur radio bands will be used to the fullest extent. Full Vivado and MATLAB licenses are included. 

A review of the OFDM-based physical layer draft specification will happen shortly and invitations are going out now. Participants must be proficient with the design review process, understand OFDM, accept our developer and participant code of conduct, and support open source work.

To join as a participant please visit https://openresearch.institute/getting-started 

Join the #neptune channel on ORI Slack.

To keep informed about our work you should know about our newsletter at https://www.openresearch.institute/newsletter-subscription/

Thank you to everyone that has made this innovative and groundbreaking project possible. We deeply appreciate your support and we will be successful with this design.

Inner Circle Newsletter April 2023

Welcome to our newsletter for April 2023!

Inner Circle is your update on what’s happening at and adjacent to Open Research Institute. We’re a non-profit dedicated to open source digital radio work. We support technical and regulatory efforts. A major beneficiary of this work is the amateur radio services. Sign up at this link http://eepurl.com/h_hYzL

Contents:

  • Guest Editorial by Dr. Daniel Estévez Amaranth in Practice
  • Federal Communications Commission Technological Advisory Council resumes!
  • HDL Coder for Software Defined Radio Class May 2023
  • FPGA Workshop Cruise with ORI?
Markdown preview

Amaranth in practice: a case study with Maia SDR

Maia SDR is a new open-source FPGA-based SDR project focusing on the ADALM Pluto. The longer term goals of the project are to foster open-source development of SDR applications on FPGA and to promote the collaboration between the open-source SDR and FPGA communities. For the time being, focusing on developing a firmware image for the ADALM Pluto that uses the FPGA for most of the signal processing provides realistic goals and a product based on readily available hardware that people can already try and use during early stages of development.

The first version of Maia SDR was released on Februrary 2023, though its development started in September 2022. This version has a WebSDR-like web interface that displays a realtime waterfall with a sample rate of up to 61.44Msps and is able to make IQ recordings at that rate to the Pluto DDR (up to a maximum size of 400MiB per recording). These recordings can then be downloaded in SigMF format.

Exploring the RF world in the field with a portable device is one of the goals of Maia SDR, so its web UI is developed having in mind the usage from a smartphone and fully supports touch gestures to zoom and scroll the waterfall. A Pluto connected by USB Ethernet to a smartphone already give a quite capable and portable tool to discover and record signals.

The following figure shows a screenshot of the Maia SDR web user interface. More information about the project can be found in https://maia-sdr.net

Amaranth

Amaranth is an open-source HDL based in Python. The project is led by Catherine “whitequark”, who is one of the most active and prolific developers in the open-source FPGA community. Amaranth was previously called nMigen, as it was initially developed as an evolution of the Migen FHDL by M-Labs.

I cannot introduce Amaranth any better than Catherine, so I will just cite her words from the README and documentation.

The Amaranth project provides an open-source toolchain for developing hardware based on synchronous digital logic using the Python programming language, as well as evaluation board definitions, a System on Chip toolkit, and more. It aims to be easy to learn and use, reduce or eliminate common coding mistakes, and simplify the design of complex hardware with reusable components.

The Amaranth toolchain consists of the Amaranth hardware definition language, the standard library, the simulator, and the build system, covering all steps of a typical FPGA development workflow. At the same time, it does not restrict the designer’s choice of tools: existing industry-standard (System)Verilog or VHDL code can be integrated into an Amaranth-based design flow, or, conversely, Amaranth code can be integrated into an existing Verilog-based design flow.

The Amaranth documentation gives a tutorial for the language and includes as a first example the following counter with a fixed limit.

from amaranth import *

class UpCounter(Elaboratable):
    """
    A 16-bit up counter with a fixed limit.

    Parameters
    ----------
    limit : int
        The value at which the counter overflows.

    Attributes
    ----------

    en : Signal, in
        The counter is incremented if ``en`` is asserted, and retains
        its value otherwise.
    ovf : Signal, out
        ``ovf`` is asserted when the counter reaches its limit.
    """
    def __init__(self, limit):
        self.limit = limit

        # Ports
        self.en  = Signal()
        self.ovf = Signal()

        # State
        self.count = Signal(16)

    def elaborate(self, platform):
        m = Module()

        m.d.comb += self.ovf.eq(self.count == self.limit)

        with m.If(self.en):
            with m.If(self.ovf):
                m.d.sync += self.count.eq(0)
            with m.Else():
                m.d.sync += self.count.eq(self.count + 1)

        return m

Amaranth Elaboratable‘s are akin to Verilog module‘s (and in fact get synthesized to module‘s if we convert Amaranth to Verilog). IO ports for the module are created in the __init__() method. The elaborate() method can create additional logic elements besides those created in __init__() by instantiating more Signal‘s (this example does not do this). It also describes the logical relationships between all these Signals by means of a Module() instance usually called m. Essentially, at some point in time, the value of a Signal changes depending on the values of some Signal‘s and potentially on some conditions. Such point in time can be either continuously, which is described by the m.d.comb combinational domain, or at the next rising clock edge, which is described by the m.d.sync synchronous domain (which is, roughly speaking, the “default” or “main” clock domain of the module), or by another clock domain. Conditions are expressed using with statements, such as with m.If(self.en), in a way that feels quite similar to writing Python code.

For me, one of the fundamental concepts of Amaranth is the division between what gets run by Python at synthesis time, and what gets run by the hardware when our design eventually comes to life in an FPGA. In the elaborate() method we have a combination of “regular” Python code, which will get run in our machine when we convert the Amaranth design to Verilog or generate a bitstream directly from it, as well as code that describes what the hardware does. The latter is also Python code, but we should think that the effects of running it are only injecting that description into the list of things that Amaranth knows about our hardware design.

Code describing the hardware appears mainly in two cases: First, when we operate with the values of signals. For instance, self.count + 1 does not take the value of self.count and add one to it when the Python code is run. It merely describes that the hardware should somehow obtain the sum of the value of the register corresponding to self.count and the constant one. This expression is in effect describing a hardware adder, and it will cause an adder to appear in our FPGA design. This behaviour is reminiscent of how Dask and other packages based on lazy evaluation work (in Dask, operations with dataframes only describe computations; the actual work is only done eventually, when the compute() method is called). I want to stress that the expression self.count + 1 might as well appear in elaborate() only after a series of fairly complicated if and else statements using regular Python code. These statements will be evaluated at synthesis time, and our hardware design will end up having an adder or not depending on these conditions. Similarly, instead of the constant 1 in the + 1 operation, we could have a Python variable that is evaluated in synthesis time, perhaps as the result of running fairly complicated code. This will also affect what constant the hardware adder that we have in our design adds to the value of the self.count register.

Secondly, we have the control structures: m.If, m.Else, and a few more. These also describe hardware. Whether the condition is satisfied is not evaluated when the Python script runs. What these conditionals do is to modify the hardware description formed by the assignments to m.d.sync and m.d.comb that they enclose so that these assignments are only effective (or active) in the moments in which the condition is satisfied. In practice, these statements do two things in the resulting hardware: They multiplex between several intermediate results depending on some conditions, in a way that is usually more readable than using the Mux() operator that Amaranth also provides. They also control what logic function gets wired to the clock enable of flip-flops. Indeed, in some conditions a synchronous Signal() may have no active statements, in which case it should hold its current value. This behaviour can be implemented in hardware either by deasserting the clock enable of the flip-flops or by feeding back the output of the flip-flops to their input through a multiplexer. What is done depends mainly on choices done by the synthesis tool when mapping the RTL to FPGA elements. As before, we can have “regular” Python code that is run at synthesis time modifying how these m.If control structures look like, or even whether they appear in the design at all.

In a sense, the regular Python code that gets run at synthesis time is similar to Verilog and VHDL generate blocks. However, this is extremely more powerful, because we have all the expressiveness and power of Python at our disposal to influence how we build our design at synthesis time. Hopefully the following examples from Maia SDR can illustrate how useful this can be.

maia-hdl

maia-hdl is the FPGA design of Maia SDR. It is bundled as a Python package, with the intention to make easy to reuse the modules in third party designs. The top level of the design is an Amaranth Elaboratable that gets synthesized to Verilog and packaged as a Vivado IP core. As shown below, the IP core has re_in and im_in ports for the IQ data of the ADC, an AXI4-Lite subordinate interface to allow the ARM processor to control the core through memory-mapped registers, AXI3 manager interfaces for the DMAs of the spectrometer (waterfall) and IQ recorder, and ports for clocking and reset.

The IP core is instantiated in the block design of a Vivado project that gets created and implemented using a TCL script. This is based on the build system used by Analog Devices for the default Pluto FPGA bitstream. In this respect, Maia SDR gives a good example of how Amaranth can be integrated in a typical flow using the Xilinx tools.

There are two classes of unit tests in maia-hdl. The first are Amaranth simulations. These use the Amaranth simulator, which is a Python simulator than can only simulate Amaranth designs. These tests give a simple but efficient and powerful way of testing Amaranth-only modules. The second are cocotb simulations. Cocotb is an open-source cosimulation testbench environment for verifying VHDL and Verilog designs using Python. Briefly speaking, it drives an HDL simulator using Python to control the inputs and check the outputs of the device under test. Cocotb has rich environment that includes Python classes that implement AXI devices. In maia-hdl, cocotb is used together with Icarus Verilog for the simulation of designs that involve Verilog modules (which happens in the cases in which we are instantiating from Amaranth a Xilinx primitive that is simulated with the Unisim library), and for those simulations in which the cocotb library is specially useful (such as for example, when using the cocotb AXI4-Lite Manager class to test our AXI4-Lite registers).

One of the driving features of Maia SDR is to optimize the FPGA resource utilization. This is important, because the Pluto Zynq-7010 FPGA is not so large, specially compared with other Xilinx FPGAs. To this end, Amaranth gives a good control about how the FPGA design will look like in terms of LUTs, registers, etc. The example with the counter has perhaps already shown that Amaranth is a low-level language, in the same sense that Verilog and VHDL are, and nothing comparable to HLS (where regular C code is translated to an FPGA design).

FFT

The main protagonist of the Maia SDR FPGA design is a custom pipelined FFT core that focuses on low resource utilization. In the Maia SDR Pluto firmware it is used as a radix-2² single-delay-feedback decimation-in-frequency 4096-point FFT with a Blackman-harris window. It can run at up to 62.5 Msps and uses only around 2.2 kLUTs, 1.4 kregisters, 9.5 BRAMs, and 6 DSPs. One of the tricks that allows to save a lot of DSPs is to use a single DSP for each complex multiplication, by performing the three required real products sequentially with a 187.5 MHz clock. A description of the design of the FFT core is out of the scope of this article, but I want to show a few features that showcase the strengths of Amaranth.

The first is the FFTControl module. The job of this module is to generate the control signals for all the elements of the FFT pipeline. In each clock cycle, it selects which operation each butterfly should do, which twiddle factor should be used by each multiplier, as well as the read and write addresses to use for the delay lines that are implemented with BRAMs (these are used for the first stages of the pipeline, which require large delay lines). As one can imagine, these control outputs are greatly dependent on the synchronization of all the elements. For example, if we introduce an extra delay of one cycle in one of the elements, perhaps because we register the data to satisfy timing constraints, all the elements following this in the pipeline will need their control inputs to be offset in time by one cycle.

It is really difficult to implement something like this in Verilog or VHDL. Changing these aspects of the synchronization of the design usually requires rethinking and rewriting parts of the control logic. In Amaranth, our modules are Python classes. We can have them “talk to each other” at synthesis time and agree on how the control should be set up, in such a way that the result will still work if we change the synchronization parameters.

For example, all the classes that are FFT pipeline elements implement a delay Python @property that states what is the input to output delay of the module measured in clock cycles. For some simple modules this is always the same constant, but for a single-delay-feedback butterfly it depends on the length of the delay line of the butterfly, and for a twiddle factor multiplier it depends on whether the multiplier is implemented with one or three DSPs. These are choices that are done at synthesis time based on parameters that are passed to the __init__() method of these modules.

The FFTControl module can ask at synthesis time to all the elements that form the FFT pipeline what are their delays, and figure out the reset values of some counters and the lengths of some delay lines accordingly. This makes the control logic work correctly, regardless what these delays are. For instance, the following method of FFTControl computes the delay between the input of the FFT and the input of each butterfly by summing up the delays of all the preceding elements.

def delay_butterflies_input(self):
    """Gives the delay from the FFT input to the input of each of the
    butterflies"""
    return [
        self.delay_window
        + sum([butterfly.delay for butterfly in self.butterflies[:j]])
        + sum([twiddle.delay for twiddle in self.twiddles[:j]])
        for j in range(self.stages)
    ]

This is then used in the calculation of the length of some delay lines that supply the control signals to the butterflies. The code is slightly convoluted, but accounts for all possible cases. I don’t think it would be reasonable to do this kind of thing in Verilog or VHDL.

mux_bfly_delay = [
    [Signal(2 if isinstance(self.butterflies[j], R22SDF) else 1,
            name=f'mux_bfly{j}_delay{k}', reset_less=True)
     for k in range(0,
                    delay_butterflies_input[j]
                    - delay_twiddles_input[j-1]
                    + self.twiddles[j-1].twiddle_index_advance)]
     for j in range(1, self.stages)]

Another important aspect facilitated by Amaranth is the construction of a model. We need a bit-exact model of our FFT core in order to be able to test it in different situations and to validate simulations of the Amaranth design against the model. Each of the modules that form the pipeline has a model() method that uses NumPy to calculate the output of that module given some inputs expressed as NumPy arrays. Here is the model for a radix-2 decimation-in-frequency single-delay-feedback butterfly. Perhaps it looks somewhat reasonable if we remember that such a butterfly basically computes first x[n] + x[n+v//2], for n = 0, 1, ..., v//2-1, and then x[n] - x[n+v//2] for n = 0, 1, ..., v//2-1.

[class R2SDF(Elaboratable):]
[...]
def model(self, re_in, im_in):
    v = self.model_vlen
    re_in, im_in = (np.array(x, 'int').reshape(-1, 2, v // 2)
                    for x in [re_in, im_in])
    re_out, im_out = [
        clamp_nbits(
            np.concatenate(
                (x[:, 0] + x[:, 1], x[:, 0] - x[:, 1]),
                axis=-1).ravel() >> self.trunc,
            self.w_out)
        for x in [re_in, im_in]]
    return re_out, im_out

The interesting thing is that, since each of the FFT pipeline modules has its individual model, it is easy to verify the simulation of each module against its model separately. The model of the FFT module, which represents the whole FFT core, simply puts everything together by calling the model() methods of each of the elements in the pipeline in sequence. An important detail here is that the arrays self._butterflies and self._twiddles are the same ones that are used to instantiate and connect together the pipeline modules, in terms of the hardware design. By having these synergies between the model and the hardware design, we reduce the chances of them getting out of sync due to code changes.

[class FFT(Elaboratable):]
[...]
def model(self, re_in, im_in):
    v = self.model_vlen
    re = re_in
    im = im_in
    if self._window is not None:
        re, im = self._window.model(re, im)
    for j in range(self.nstages):
        re, im = self._butterflies[j].model(re, im)
        if j != self.nstages - 1:
            re, im = self._twiddles[j].model(re, im)
    return re, im

Instantiating Verilog modules and primitives

A question that often comes up is how to instantiate Verilog modules, VHDL entities or FPGA primitives in an Amaranth design. Kate Temkin has a short blog post about it. In maia-hdl this is used in in several cases, such as to implement clock domain crossing with the Xilinx FIFO18E1 primitive. The most interesting example is however the Cmult3x module, which implements complex multiplication with a single DSP48E1 that runs at three clock cycles per input sample (some simple algebra shows that a complex multiplication can be written with only three real multiplications).

When designing modules with DSPs, I prefer to write HDL code that will make Vivado infer the DSPs I want. This is possible in simple cases, but in more complicated situations it is not possible to make Vivado understand exactly what we want, so we need to instantiate the DSP48 primitives by hand.

The drawback of having an Amaranth design that contains instances of Verilog modules, VHDL entities or primitives is that we can no longer simulate our design with the Amaranth simulator. If our instances have a Verilog model (such as is the case with Xilinx primitives via the Unisim library), we can still convert the Amaranth design to Verilog and use a Verilog simulator. This is done in maia-hdl using Icarus Verilog and cocotb. However, this can be somewhat inconvenient.

There is another possibility, which is to write different implementations of the same Amaranth module. One of them can be pure Amaranth code, which we will use for simulation, and another can use Verilog modules or primitives. The two implementations need to be functionally equivalent, but we can check this through testing.

The way to acomplish this is through Amaranth’s concept of platform. The platform is a Python object that gets passed to the elaborate() methods of the modules in the design. The elaborate methods can then ask the platform for some objects that are usually dependent on the FPGA family, such as flip-flop synchronizers. This is a way of building designs that are more portable to different families. The platform objects are also instrumental in the process of building the bitstream completely within Amaranth, which is possible for some FPGA families that have an open-source toolchain.

In the case of the maia-hdl Cmult3x we simply check whether the platform we’ve been passed is an instance of XilinxPlatform and depending on this we have the elaborate() method either describe a pure Amaranth design that models the DSP48 functionality that we need, or instantiate a DSP48E1 primitive. Note that in the case of the pure Amaranth design we do not model the full functionality of the DSP48. Only that which is applicable to this use case.

[class Cmult3x(Elaboratable):]
[...]
def elaborate(self, platform):
    if isinstance(platform, XilinxPlatform):
        return self.elaborate_xilinx(platform)

    # Amaranth design. Vivado doesn't infer a single DSP48E1 as we want.
    [ ... here a pure amaranth design follows ... ]

def elaborate_xilinx(self, platform):
    # Design with an instantiated DSP48E1
    [...]
    m.submodules.dsp = dsp = Instance(
        'DSP48E1',
    [...]

Registers

Another aspect where the flexibility of Amaranth shines is in the creation of register banks. In maia-hdl, the module Register corresponds to a single 32-bit wide register and the module Registers forms a register bank by putting together several of these registers, each with their corresponding address. The registers support a simple bus for reads and writes, and an Axi4LiteRegisterBridge module is provided to translate between AXI4-Lite and this bus, allowing the ARM CPU to access the registers.

Registers and register banks are created with Python code that describes the fields of the registers. The basic ingredient is the Field named tuple:

Field = collections.namedtuple('RegisterField',
                               ['name', 'access', 'width', 'reset'])

We describe a register by giving it a name, an access mode (which can be read-only, write-only, read-write, or some other more specialized modes that we will describe below), a width, and a reset or default value.

The best way to understand how to work with these registers is to see how they are used in the Maia SDR top-level design.

self.control_registers = Registers(
    'control',
    {
        0b00: Register(
            'product_id', [
                Field('product_id', Access.R, 32, 0x6169616d)
            ]),
        0b01: Register('version', [
            Field('bugfix', Access.R, 8,
                  int(_version.split('.')[2])),
            Field('minor', Access.R, 8,
                  int(_version.split('.')[1])),
            Field('major', Access.R, 8,
                  int(_version.split('.')[0])),
            Field('platform', Access.R, 8, 0),
        ]),
        0b10: Register('control', [
            Field('sdr_reset', Access.RW, 1, 1),
        ]),
        0b11: Register('interrupts', [
            Field('spectrometer', Access.Rsticky, 1, 0),
            Field('recorder', Access.Rsticky, 1, 0),
        ], interrupt=True),
    },
    2)
self.recorder_registers = Registers(
    'recorder',
    {
        0b0: Register('recorder_control', [
            Field('start', Access.Wpulse, 1, 0),
            Field('stop', Access.Wpulse, 1, 0),
            Field('mode_8bit', Access.RW, 1, 0),
            Field('dropped_samples', Access.R, 1, 0),
        ]),
        0b1: Register('recorder_next_address', [
            Field('next_address', Access.R, 32, 0),
        ]),
    },
    1)

Here we show two register banks: one for the control of the IP core and another for the control of the IQ recorder. There is a similar third register bank for the control of the spectrometer (waterfall).

The parameters of the Registers constructor are a name, a dictionary that contains the registers in the bank (the keys of the dictionary are the addresses, and the values are the Register objects), and the width of the address bus. Note that these addresses correspond to the addressing of the native register bus. When we convert to AXI4-Lite, the addresses get shifted by two bits to the left because each register is 4 bytes wide.

The parameters of the Register constructor are a name and a list of Field‘s describing the fields of the register. Fields are allocated into the 32-bit register according to their order in the list, starting by the LSB. For instance, in the interrupts register, the spectrometer field occupies the LSB and the recorder field occupies the next bit.

If we look at the control registers, we can see that the registers for product_id and version have access type R, which means read-only. These registers are never wired in the design to other signals that would override their default values, so they are in fact constants that the CPU can read to check that the IP core is present and find its version number.

Next we have a control register, which has an sdr_reset field. This is wired internally to a bunch of reset signals in the IP core. It has a default value of 1, which means that most of the IP core starts in reset. The CPU can write a 0 to this field to take the IP core out of reset before using it. Accessing this sdr_reset field within the design is very simple, because the Registers and Register implement __getitem__(), allowing us to access them as if they were dictionaries. This example shows how it works. Here we are connecting the field sdr_reset to the reset input of something called rxiq_cdc (which implements clock domain crossing between the ADC sampling clock and the internal clock used in the IP core).

m.d.comb += rxiq_cdc.reset.eq(
    self.control_registers['control']['sdr_reset'])

If we look at the interrupts register, we can see an example of the Rsticky access mode. This means read-only sticky. A field of this type will be set to 1 when its input (which is wired internally in the IP core) has the value 1. It will keep the value 1 even if the input goes back to 0. The field is cleared and set to 0 when it is read. The intended use for this access mode is interrupts. A module can pulse the input of the field to notify an interrupt, and the field will hold a 1 until the CPU reads the register, clearing the interrupts. The interrupts register even has an interrupt=True option that provides an interrupt output that can be connected directly to the F2P interrupt port of the Zynq. This interrupt output will be high whenever any Rsticky field in the register is non-zero.

Finally, the recorder_control field gives some examples of the Wpulse access type. This is a write-only field with pulsed output. Writing a 1 to this field causes a one-cycle pulse at its output. This is ideal for controlling modules that require a pulse to indicate some event or command. For example, this is the case with the start and stop commands of the IQ recorder.

The Amaranth code that makes all of this work is not so complicated. You can take a look at the register.py file in maia-hdl to see for yourself.

Another interesting feature of this register system is that it can write an SVD file describing the register map. CMSIS-SVD is an XML format that is often used to describe the register maps of microcontrollers and SoCs. Maia SDR uses svd2rust to generate a nice Rust API for register access.

The Registers and Register classes have svd() methods that generate the SVD XML using Python’s xml.etree.ElementTree. This is relatively simple, because the classes already have all the information about these registers. It is, after all, the same information that they use to describe the hardware implementation of the registers. This is another example of how by using synergies between the code that describes the hardware design and code that does something related to that hardware design (in this case, writing SVD), we make it harder for changes in the code base to cause inconsistencies.

Conclusions

In this article we have gone from a “hello world” type counter in Amaranth to some rather intricate code from the inner workings of an FFT core. My intention with giving these code examples is not to expect the reader to understand all the code, but rather to give a feeling for how using Amaranth in complex projects can look like. Perhaps by now I have managed to convince you that Amaranth is a powerful and flexible alternative to HDLs such as Verilog and VHDL, or at least to get you interested in learning more about Amaranth and the world of open-source FPGA and silicon design.

Federal Communications Commission Technological Advisory Council resumes!

Want someone on your side at the FCC? We have good news. The FCC TAC is going to resume work. The previous term ended in December 2022 with an in-person meeting in Washington, DC. ORI was a member of the AI/ML Working Group and served as a co-chair of the “Safe Uses of AI/ML” sub-working group. The next term will be for two years. The appointment will require another round of nominations and vetting. An invitation to ORI has been made. ORI will speak up for open source digital radio work and the amateur radio services. Thank you to everyone that made the 2022 FCC TAC term productive and beneficial. Join the #ai channel on ORI Slack to get more involved. Not part of Slack? Visit https://openresearch.institute/getting-started to Get Started.

HDL Coder for Software Defined Radio Class May 2023

Sign-ups are live at https://us.commitchange.com/ca/san-diego/open-research-institute/campaigns/hdl-coder-for-software-defined-radio

Advanced training for Digital Communications, Software Defined Radio, and FPGAs will be held 1-5 May 2023. Do you know someone that can benefit from customized and focused training? Please forward this email to them. Designed to benefit open source digital radio, this course also benefits the amateur radio services.

Presented by ORI and taught by Mathworks, this class will cover the following topics.

COURSE OUTLINE

Day 1 – Generating HDL Code from Simulink & DSP for FPGAs

Preparing Simulink Models for HDL Code Generation (1.0 hrs)

Prepare a Simulink model for HDL code generation. Generate HDL code and testbench for simple models requiring no optimization.

  • Preparing Simulink models for HDL code generation
  • Generating HDL code
  • Generating a test bench
  • Verifying generated HDL code with an HDL simulator

Fixed-Point Precision Control (2.0 hrs)

Establish correspondence between generated HDL code and specific Simulink blocks in the model. Use Fixed-Point Tool to finalize fixed point architecture of the model.

  • Fixed-point scaling and inheritance
  • Fixed-Point Designer workflow
  • Fixed-Point Tool
  • Fundamental adders and multiplier arrays
  • Division and square root arrays
  • Wordlength issues and Fixed-point arithmetic
  • Saturate and wraparound.
  • Overflow and underflow

Optimizing Generated HDL Code (4 hrs)

Use pipelines to meet design timing requirements. Use specific hardware implementations and share resources for area optimization.

  • Generating HDL code with the HDL Workflow Advisor
  • Meeting timing requirements via pipelining
  • Choosing specific hardware implementations for compatible Simulink blocks
  • Sharing FPGA/ASIC resources in subsystems
  • Verifying that the optimized HDL code is bit-true cycle-accurate
  • Mapping Simulink blocks to dedicated hardware resources on FPGA

Day 2 – DSP for FPGAs

Signal Flow Graph (SFG) Techniques (SFG) Techniques and high-speed FIR design (2.0 hrs)

Review the representation of DSP algorithms using signal flow graph. Use the Cut Set method to improve timing performance. Implement parallel and serial FIR filters.

  • DSP/Digital Filter Signal Flow Graphs
  • Latency, delays and “anti-delays”!
  • Re-timing: Cut-set and delay scaling
  • The transpose FIR
  • Pipelining and multichannel architectures
  • SFG topologies for FPGAs
  • FIR filter structures for FPGAs

Multirate Signal Processing for FPGAs (4.0 hrs)

Develop polyphase structure for efficient implementation of multirate filters. Use CIC filter for interpolation and decimation.

  • Upsampling and interpolation filters
  • Downsampling and decimation filters
  • Efficient arithmetic for FIR implementation
  • Integrators and differentiators
  • Half-band, moving average and comb filters
  • Cascade Integrator Comb (CIC) Filters (Hogenauer)
  • Efficient arithmetic for IIR Filtering

CORDIC Techniques AND channelizers (2.0 hrs)

Introduce CORDIC algorithm for calculation of various trigonometric functions.

  • CORDIC rotation mode and vector mode
  • Compute cosine and sine function
  • Compute vector magnitude and angle
  • Architecture for FPGA implementation
  • Channelizer

Day 3 – Programming Xilinx Zynq SoCs with MATLAB and Simulink & Software-Defined Radio with Zynq using Simulink

IP Core Generation and Deployment (2.0 hrs)

Use HDL Workflow Advisor to configure a Simulink model, generate and build both HDL and C code, and deploy to Zynq platform.

  • Configuring a subsystem for programmable logic
  • Configuring the target interface and peripherals
  • Generating the IP core and integrating with SDK
  • Building and deploying the FPGA bitstream
  • Generating and deploying a software interface model
  • Tuning parameters with External Mode

Model Communications System using Simulink (1.5 hrs)

Model and simulate RF signal chain and communications algorithms.

  • Overview of software-defined radio concepts and workflows
  • Model and understand AD9361 RF Agile Transceiver using Simulink
  • Simulate a communications system that includes a transmitter, AD9361 Transceiver, channel and Receiver (RF test environment)

Implement Radio I/O with ADI RF SOM and Simulink (1.5 hrs)

Verify the operation of baseband transceiver algorithm using real data streamed from the AD9361 into MATLAB and Simulink.

  • Overview of System object and hardware platform
  • Set up ADI RF SOM as RF front-end for over-the-air signal capture or transmission
  • Perform baseband processing in MATLAB and Simulink on captured receive signal
  • Configure AD9361 registers and filters via System object
  • Verify algorithm performance for real data versus simulated data

Prototype Deployment with Real-Time Data via HW/SW Co-Design (2.0 hrs)

Generate HDL and C code targeting the programmable logic (PL) and processing system (PS) on the Zynq SoC to implement TX/RX.

  • Overview of Zynq HW/SW co-design workflow
  • Implement Transmitter and Receiver on PL/PS using HW/SW co-design workflow
  • Configure software interface model
  • Download generated code to the ARM processor and tune system parameters in real-time operation via Simulink
  • Deploy a stand-alone system

FPGA Workshop Cruise with ORI?

Want to learn more about open source FPGA development from experts in the field? Ready to capitalize on the HDL Coder for Software Defined Radio Class happening in May 2023? Want to get away? How about something that can give you both? We are looking at organizing an FPGA Workshop Adventure Cruise. Be part of the planning and write fpga@openresearch.institute

Are you interested in supporting work at ORI?

Consider being part of the board. We’d like to expand from 5 to 7 members in order to better serve our projects and community.

We’ve got lots going on with Opulent Voice, Haifuraiya, AmbaSat Respin, and regulatory work.

Thank you from everyone at ORI for your continued support and interest!

Want to be a part of the fun? Get in touch at ori@openresearch.institute

Class Announcement: HDL Coder for Software Defined Radio

Open Research Institute (https://openresearch.institute) has negotiated a customized class for advanced digital radio work. Content was chosen to directly benefit open source workers in the amateur radio service.

Registration link is below. A heavily subsidized link is offered. Class runs 1-5 May 2023 and is entirely virtual. Theoretical and practical content is offered. Class is taught by Mathworks instructors and uses their training platform. 

https://us.commitchange.com/ca/san-diego/open-research-institute/campaigns/hdl-coder-for-software-defined-radio

The goal of this class is to expand the technical volunteer corps in amateur radio and amateur satellite services. Radio work accomplished on FPGAs and complex RFICs is the primary focus.

Inner Circle Newsletter March 2023

March 2023 Inner Circle
Welcome to our newsletter for March 2023!

Inner Circle is a non-technical update on everything that is happening at ORI. Sign up at this link http://eepurl.com/h_hYzL

Contents:
FPGA Workshop Cruise with ORI?
ORI’s Birthday 6 March – Celebrate With Pins!
RFBitBanger Prototypes
Announcing the ORI App Stores
QSO Today Ham Expo Spotlight
Jay Francis in QEX
Pierre W4CKX Declares Candidacy for ORI Board of Directors

FPGA Workshop Cruise with ORI?
Want to learn more about open source FPGA development from experts in the field? Want to get away? How about something that can give you both? We are looking at organizing an FPGA Workshop Adventure Cruise. Be part of the planning and write fpga@openresearch.institute

ORI’s Birthday – Celebrate With Pins!
We celebrate our 4th birthday on 6 March 2023. Thank you to everyone that has helped ORI grow and succeed in so many different ways. To commemorate our anniversary, we have a limited edition acrylic logo pin. They will be available for a small donation at all upcoming in-person events. Where will be be? We’ll be at DEFCON 31 and IEEE IWRC in Little Rock, AR, USA 13-14 September 2023. Want to include us at your event before then? Let us know at hello@openresearch.institute

RFBitBanger Prototypes
Interested in high frequency amateur (HF) bands? Want to learn about Class E amplification? Excited about open HF digital protocols that aren’t just signal reports? Well, we have a kit for you. Here’s a walk-through by Paul KB5MU of all RFBitBanger modes. This project is lead by Dr. Daniel Marks, is enthusiastically supported by ORI, and will be demonstrated at DEFCON in August 2023. We are doing all we can to have kits available for sale by DEFCON, or sooner.

Announcing the ORI App Stores
Open Research Institute can be found in the Google Play Store and the Apple App Store. That’s right – we are in both app stores delivering open source mobile apps. Thank you to everyone that has helped make this possible. The Ribbit app will be available on both platforms as our initial release. Do you know of an open source application that needs a home? Get in touch at hello@openresearch.institute and let’s talk. We want to share our platform and support applications that help open source and amateur radio.

QSO Today Ham Expo Spotlight
We hope to see you again at QSO Today Ham Expo, 25-26 March 2023. If you haven’t gotten a ticket yet, please visit https://www.qsotodayhamexpo.com/
This is a wonderful event that showcases the best parts of amateur radio. The theme for this Ham Expo is “New License – Now What?” Recordings will be available on the Ham Expo platform for a month, and then will be available on YouTube for anyone to view. ORI will volunteer at the March 2023 QSO Ham Expo session and will have technical presentations, a booth, and poster sessions at the Autumn 2023 QSO Today Ham Expo.

Jay Francis in QEX
Please see page 20 of the March/April 2023 issue of QEX magazine for an article co-authored by Jay Francis, our AmbaSat Re-Spin team lead. Excellent job, Jay!

Pierre W4CKX has declared his candidacy for ORI Board of Directors
We welcome Pierre’s interest in being a member of the board. Pierre is the Ribbit project lead. He brings broad industry knowledge, experience in Agile project management, a commitment to ethical leadership, and innovative energy. Learn about all our directors at https://www.openresearch.institute/board-of-directors/

Are you interested in supporting work at ORI? Consider being part of the board. We’d like to expand from 5 to 7 members in order to better serve our projects and community.

We’ve got lots going on with Opulent Voice, Haifuraiya, AmbaSat Respin, and regulatory work. We support IEEE in many ways, one of which is logistics support with technical presentations such as “Advances in AI for Web Integrity, Ethics, and Well Being” by Srijan Kumar PhD. Video recording of his talk can be found here.

Thank you from everyone at ORI for your continued support and interest!

Whatever will be do for our April 1st newsletter?

Want to be a part of the fun? Get in touch at ori@openresearch.institute

QoS: Call for Input

We’re moving into the challenge of multiplexing on the transponder with a goal of delivering Quality of Service (QoS) metrics and policies.

This is “how do the uplink packets get properly prioritized on the downlink, to make the most of limited resources”.

These resources are spectrum, power, and time.

QoS doesn’t make any of our communications channels go faster. This is a common misconception about QoS. Here’s some descriptions from conversations this week. I would like to hear more opinions about QoS in general, and any specific requirements that people on this list might have.

Kenneth Finnegan writes,

“In #networking it’s common to forget that QoS is mainly about deciding which packets you’d rather drop first.

If you don’t like that idea, then you just need to pony up and throw more capacity at the problem.”

Adam Thompson continues,

“In the presence of a pizza that’s not big enough for all the hungry people, QoS inhibits less-important pizza eaters. This lets more-important eaters-of-pizza get more pizza than their fair share, at the expense of the less-important eaters.

“In the presence of a pizza that’s not big enough for all the hungry people, QoS inhibits less-important pizza eaters. This lets more-important eaters-of-pizza get more pizza than their fair share, at the expense of the less-important eaters.

QoS never (ever!) makes the pizza bigger – if you need more pizza, you must still bake or buy more, or someone’s going to go hungry!

Complex QoS systems might let you differentiate between e.g. crust and topping and permit cutting the pizza into bizarre topographies/topologies, but still can’t make the pizza any bigger.

Finally, if there is enough pizza for everyone, QoS doesn’t do anything useful.”

If this last part sounds familiar, then you’re not alone. QoS often doesn’t do anything useful… in a resource rich environment. This may be the main reason that we sometimes hear that QoS is a “failure”, that it’s “never used”, or “why bother for hams since hams don’t care about this subject at all”.

It is true that most amateur communications are made with acres and acres of spectrum, with a very generous power limit (although you are supposed to use the minimum required power) and no time limits on how often you can try to make a contact.

When we talk about microwave broadband digital communications, it’s a different situation. And, with space channels, there are constraints. We have less bandwidth to work with because we’re on a sub-band. We have latency, which is non-trivial for GEO or beyond. We have power concerns and pointing requirements.

“Adaptive” QoS that does nothing until congestion forces some decisions, at which time we sort with respect to SNR, has been our baseline.

What we want to do when constraints are hit is what we need to better define. Right now, we have a whiteboard design (summarized above) and a paper about Adaptive Coding and Modulation (ACM) that was published in AMSAT-DL and AMSAT-UK Journals.

We have the implementation guidelines from GSE as well, which address QoS and show how to set up queues.

With a controllable downlink going out over the air, and a defined uplink protocol, now is the time to work on exactly how to multiplex the traffic. Evariste asked about this exact thing less than a week ago at the FPGA meetup.

Decisions about QOS heavily affect the central part of the design, so let’s get this right.

Do you have experience implementing QoS policies? Do you have experience with bad QoS policies as a consumer? Do you have an idea about what you want to see this design do?

Well, you’re in the right place, and we’d love to hear what you have to say about it.

Participate at https://openresearch.institute/getting-started

ORI participation at OSCW 2021

Recording, transcript, and slides of Open Research Institute’s presentation at Open Source Cubesat Workshop 2021.


Hello everybody! I’m Michelle Thompson W5NYV and I’m here to tell you all about what Open Research Institute is and what we have been doing.

Open Research Institute (ORI) is a non-profit research and development organization which provides all of its work to the general public under the principles of Open Source and Open Access to Research. As we all know, these mean particular things, and those things have to be defined and they have to be defended.

Open Source is type of intellectual property management where everything you need to recreate or modify a design is freely available. As a baseline, we use GPL v3.0 for software and the CERN Open Hardware License version 2.0 for hardware. All we do is open source work, primarily for amateur radio space and terrestrial, but also some other fields, as you will see.

So who are we?

Here is our current board, and our immediate past CEO Bruce Perens. We have one opening on the board, as Ben Hilburn, one of our founders, very recently retired from being an active Director at ORI. He remains as one of our senior advisors. We are looking for someone to join ORI board that supports what we do and wants to help make it happen. It’s an active role in a flat management structure. Board members are are experienced in management, engineering, operations, and technology, and three out of the current number of four are from underrepresented groups in STEM.

As a board, it is our mission to serve our participants, developers, and community members. We now have at least 535 that participate in what we call the Open Source Triad: our mailing list, Slack, and GitHub. All work is organized in independent projects or initiatives.

We have some affiliations and we proudly ascribe to the Open Space Manifesto from Libre Space Foundation. We work with radio organizations, several universities, and have worked with a variety of for-profits.

What do we do?

Here’s a visual summary of top level projects and initiatives. The vertical axis is risk. Higher risk projects are at the top, lower risk projects are at the bottom. Maturity increases left to right. Maturity may indicate schedule, but the score is also influenced by complexity or difficulty. The color of the shape indicates how much stress that project is under or what the risk level is at this time. The size of the shape is the budget estimate. By far, the largest budget, riskiest, and least mature work is in the AquaPhage project, which is open source bacteriophage research and development. Bacteriophage are viruses that attack and destroy bacteria. This is biomedical and not amateur radio. This project was halted by COVID and has not yet resumed.

Our digital multiplexing payload project is called P4DX, and it’s in the middle in green. This is a multiple access microwave digital regenerating repeater for space and terrestrial deployment.

Channels divided in frequency are the uplink. The uplink is on 5 GHz. The processor on the payload digitizes and multiplexes these signals and uses DVB-S2/X as a single time-division downlink. The downlink is on 10 GHz. The system adapts to channel conditions and handles things like quality of service decisions. For example, low and high latency digital content. The uplink is divided up using a polyphase channelizer, based on the open source work done by Theseus Cores.

For the current prototype, we are only using MPEG transport stream, but generic data is the goal. The prototype beacon signal is 5 MHz wide and we are using one modulation and one error coding (yet). We are not yet rotating through all the allowed combinations in DVB-S2 (yet).

Our prototype work can also serve as a terrestrial multimedia beacon. Work was demonstrated to groups with mountaintop spaces in October 2021, and deployment will be as soon as possible.

M17 project is an open source VHF/UHF radio protocol. Think open source digital mode HTs and repeaters. This project is only slightly more stressed than P4DX, but it’s further along in maturity because it’s narrower in scope. We believe M17 Project will be very successful from current development to scaling up to commercial product launch. The M17 protocol is the native digital uplink protocol, with some modifications for 5GHz, for P4DX. We are working hard to get M17 on and through more satellites and more sounding rocket tests today.

Engineers General is our initiative to hire highly competent open source workers to reduce burnout and increase quality in open source work important to amateur radio. We have one contractor currently, eight resumes, and have applied for funding for two more. We are actively looking for funding for the remaining five.

The “birdbath” is a large dish antenna at the Huntsville Space and Rocket Center. This was used in the past, but has been parked for decades. It took two years of negotiation, but ORI has the support of the museum and permission to begin work renovating this dish for citizen science and amateur radio educational use. Work parties from earlier this year were rescheduled due to COVID.

Upper right there are two completed projects. One is ITAR/EAR Regulatory Work. It took over a year, but we received a determination from the State Department that open source satellite work is free of ITAR, from Commerce that it is free of EAR, and we obtained an advisory opinion that publishing on the internet counts as publishing under the regulations. This is a huge step forward for not just amateur radio, but anyone that wants to contribute to open source space work.

Debris Mitigation Regulatory Work took 10 months to complete. The process culminated in a highly successful meeting with the FCC Wireless Telecommunications Board, the Office of Engineering Technology, and the Satellite Bureau in late October 2021.

Lower right is Battery Matching, a project that matches NiCd cells for very durable batteries in the style that used to be done in amateur satellites, and puts the methods and documentation in the public domain.

AmbaSat Inspired Sensors used to be on the bottom right but now it’s bumped back a bit in maturity level is higher risk. This was supposed to be a project done by students at Vanderbilt university, but no students materialized, primarily due to COVID. We had one kick-butt professional volunteer who was working on a 10GHz beacon that went into the sensor connector on the main board, but the project was moving slowly, and ORI decided to provide additional operational support. Additional volunteers joined the team, we reviewed the finances, and then took some actions. We updated the main board to move it from the illegal ISM band it was in to the legal 70cm ham band. We improved power and ground and addressed some other design concerns. The boards are back as of last week and software and firmware development is underway. The 10 GHz sensor “beacon” work is proceeding quickly as well. AmbaSat is an excellent educational platform, but the ISM band decision isn’t the only problem with it. It’s very small.

We decided to look at combining the 70cm AmbaSat with another open source satellite board to make a combined spacecraft design. I reached out to Pierros Pappadeus at Libre Space, and we are moving forward with using the SatNOGS Comms project. We look forward to contributing to the FPGA codebase and flying both AmbaSat and SatNOGS Comms designs as early and as often as possible, starting with sounding rockets and ending up in space.

All of these projects are open source and all work is published as it is created.

When?

We have timelines! We were incorporated in February of 2018, got our 501c3 in March of 2019, and we hit the ground running and haven’t stopped since.

We’ll distribute a copy of the slides so you can see our wins and losses along the along the way. There’s a lot going on in here.

Here’s what’s been going on since March, and the future plans we know about.

We use Agile framework for management, and most of us have some sort of formal certification either completed, or in process. This is the Agile manifesto and it is the foundation of how our board decides things and how it supports project leads and volunteers. Note the second item, and put in the word hardware instead of software, and that’s one of the reasons we demonstrate early and often and incorporate the feedback quickly.

Where are we?

Here’s the locations of the concentrations of current major contributors and participants. When we say international, we mean it. Our participants have a wide range of ages, are generally educated in engineering, come from a variety of backgrounds, but do tend to be relatively young and male.

We have some physical locations that are important for carrying out the work we do. Remote Labs are lab benches connected to the internet that allow direct access to advanced lab equipment and two different large Xilinx development boards and DVB-S2/X gear. We have relocated our second Remote Lab equipment from Florida to Arkansas, and have added a three-dish interferometry site for amateur radio and public science use. Remote Labs are here for you all to use. If you need large FPGA resources and test equipment up to 6 GHz, then we have your back.

We bought Open Lunar Foundation’s satellite lab. It’s in storage waiting for the M17 project lab construction to conclude, and then the equipment will go there to pack that lab full of wonderful test equipment, materials, and supplies.

Why do this?

We believe that an open source approach to things like amateur digital communications, bacteriophage research, and sticking up for the non-commercial use of space will result in the best possible outcomes for the good of humanity.

We have a lightweight agile approach to doing things. We keep our overhead very low, we are radically participant-focused, and the work must be internationally accessible.

You can see that public demonstrations and regulatory work are given a high priority. Working code and working hardware are highly valued. Working means working over the air.

Thank you to everyone at Libre Space for the support and opportunity to present here today.

https://www.openresearch.institute/wp-content/uploads/2021/12/OSCW-2021-W5NYV-ORI-1.pdf

P4XT (Phase One) Workshop Design Review

Learn about our work on the digital microwave broadband transponder for amateur radio.

https://www.youtube.com/watch?v=QXiWCgo10lg

All work is donated to the general public as open source.

This digital multiplexing transponder is a regenerative design, taking full advantage of a wide variety of cutting edge technology, intended for amateur radio use in space and terrestrial deployments.

This review focuses on decisions made for the prototype board set that implements the transmitter side of the payload.

Comment and critique welcome and encouraged.

Floating Vivado License for FPGA Work Purchased – Will be Available for Community Use

Thanks to the generous support of Yasme Foundation, ARRL Foundation, and many individual Open Research Institute supporters, ORI has purchased a full floating Vivado license for FPGA work. This includes the System Generator for DSP.

We are testing a setup that will make team and community use of this license possible. This is a big step forward from our current situation and will greatly accelerate FGPA design and test.

The first step was setting up a license server at a donated data center. Many thanks to Nick KN6NK for offering the time, resources, and expertise to get this working.

The second step, being tested right now, is using GitHub as a directory service for adding users and keys.

The goal is for users of the license to be able to add themselves with minimal admin overhead while asserting some reasonable control over access.

GitHub provides a way for users to get public keys. The work required of us is to script user management and periodically sync key management.

Thank you to EJ Kreiner for helping test and refine this community asset. We anticipate being able to support as many amateur technical communities and projects as possible, to get the greatest possible use from the license.

Special thanks to ARRL and Yasme. We would not be able to afford this investment without their support.

FPGA iCEBreaker Workshop – digital communications for amateur satellites

Greetings all!

Field Programmable Gate Arrays (FPGAs) are one of three fundamental types of digital architectures used for communications R&D.

The others are general purpose processors and graphical processing units (GPUs).

This fall, in San Diego, California, there will be an FPGA course sponsored by Open Research Institute. There are 10 spots with amateur communications as the focus of the work.

FPGAs are a primary technology in satellite communications. They’re used in R&D and in deployment. It is difficult to get started with FPGA design for several reasons. The tools have traditionally been proprietary. The companies that make the tools price them for large corporations to buy. Coursework for FPGA design is rare.

This is where iCEBreaker makes a difference.

An iCEBreaker Workshop 10 pack has been made available. They are described at this link https://www.crowdsupply.com/1bitsquared/icebreaker-fpga

I will use this hardware to put on a course for anyone interested in amateur radio satellite and terrestrial development. All course materials will be published.

The first course will be in San Diego. If you’re in the area, please get in touch! MakerPlace and CoLab are the likely sites.

Later workshops could be at places like Symposium, Xenia, or Hamcation. The full course cannot be accomplished in a day, but a workshop could get the basics across and provide a substantial boost to motivated amateur satellite engineering volunteers. Let me know what you think.

more soon!
-Michelle W5NYV