New Paradigm Network Amateur Radios - Part 1
Take a look at my paper for the upcoming DCC: http://www.ka9q.net/KA9Q-radio.pdf. I argue that it's the protocols and interfaces between the various modules that really matter. Having not seen a whole lot of discussion of the topic, I argue for IP multicasting.
In your discussion of HF radios you mentioned the Radioberry but not the Hermes Lite 2 (http://www.hermeslite.com/). The HL2 is a totally open source/open hardware network 5W HF transceiver with gigabit ethernet and lots of software support. So instead of a $3000 Flex 6000 plus transverter, maybe a $250 HL2 with a transverter would be a better option for an NPNAR, especially if someone with RF hardware chops would design a transverter add-on card to attach to the HL2. Steve KF7O, the designer of the HL2 has mentioned interest in supporting such a card.
What ever happened to Bruce Perens' (K6BP) white box radio? IIRC he was coming up with a reference design for an open source digital voice radio. Seems like it should be more possible than ever now that FPGAs are fairly cheap and manufacturers have fairly stable architecture and programming. I think if we want radios such as you describe it will be up to the ham community to develop the spec and publish it. Then manufacturers could just crank out boxes without having to devote a lot of resources to engineering. Of course that was what JARL did with D-star and look how that turned out. Who knew by publishing the spec in Japanese it magically became closed and proprietary to Icom... 🙄
At any rate, I have a Lime SDR mini, LimeNET Micro and two Lime RFE amplifiers. The RFE was delayed for nearly two years due to COVID and the end of production of the main PA chip. They had to completely redesign the amp for a new chip. Seems like this a pretty common problem with RF devices. The LimeNET Micro is the most interesting of the set, since it is designed around a Raspberry Pi Compute Module 3. Basically it is a complete SDR computer, just plug in a monitor, keyboard, sound card and mouse. It also has USB, wifi and Ethernet onboard, opening up possibilities for a "network attached radio." But the learning curve for all that is huge and the delay in the RFE meant it got pushed to the back of the workbench.
I've played around with generating different modulation and hearing tones on a receiver, but the learning curve for GNU radio is enormous even using the GUI companion software. Scratch radio is better but it still needs to be built from source. A reference design for a carrier board around a Raspberry PI 4 compute module, SDR, 5 Watt PA and power/battery charging seems like it should be attainable, along with a prebuilt image for the Pi, preferably one that is maintained long term. All of the network interface stuff is already on the RPI, the GPIO is well understood, and GNU radio runs on a 3B, so it should really shine on a 4. A stripped down custom image with set up scripts to get a base configuration would complete the radio. Heck, if done right it could even make it through the FCC part 15 scanner requirement.
There are $5 chips available from companies like Maxim that feature full I/Q transceivers. Yes they need local oscillators, amplifiers and so forth but have all the basic capabilities for amateur wideband communications below 1GHz.
I think I have a path forward now for experimentation. From what I see the SoftRock Rx/Tx I built for 10-15metres back in 2012. It seems that it will accept an audio signal up to 100kHz and if true it would be great for experimentation. On 10 up to 20kHz bandwidth is legal, although into a dummy load I could go wider. And an RTL-SDR will do nicely as a receiver, I would think.
Coupled with gnu radio (for experimentation, not production) I'm sure I could learn a lot more about it.
Today is my club's "Winlink Wednesday" with everyone using 1200bps packet. It is very slow - not just the data transmission, but everything in the process is slow. From a sailboat or an earthquake zone I can see obvious benefits... if Starlink is not available that is. Otherwise I think Morse code would be quicker.