开云体育

While my intent is for this to be a forum for discussing QEX articles, I hope that more will come of it. So I thought I'd make a post about my personal major interest.

An almost 40 year obsession of mine dating from when I started work in the oil industry and first learned DSP is the idea of creating a reliable communication system capable of achieving antipodal range at mW ERP.

FT8 provides similar real time capability, but requires higher power levels to keep the QSO time tractable. My goal is reliable asynchronous transmission of ~ 2 kB text in 24 hours to another station at <100 mW ERP. Email by ham radio. There are a slew of ways to implement the concept.

At the moment I am trying to master basic RF amplifier and oscillator design, so it will be a while before I get back into familiar territory doing DSP. My present goal is to be able to take a small signal transistor and based on either the data sheet or if needed measurements consistently design a circuit that does precisely what I want it to do.

I *think* I've finally deciphered Chris Boswick's and Thomas Lee's books and know how but have not yet built an amplifier to test my calculations against a VNA.

I should note that I am severely ADD and am prone to wander off and do something completely different such as play jazz guitar for weeks or months or do precision metal work. No matter what I do, I feel guilty about all the things I'm not doing. A few too many interests.

Have Fun!
Reg

I expect it would have to be mulithop for that distance. However, there was a WSPR report of UK to Australia by Jenny List of Hackaday using a single frequency at 100 mW IIRC. So clearly possible. I have in mind using multiple bands and changing frequency as the propagation changes. However, I am merely stating a goal, not an expectation. I'd be happy with reliable radio mail over shorter distances. A variant of UUCP/packet radio might be needed to meet the "reliable" requirement over antipodal distances.

The underlying concept is that the time-bandwidth-power product controls the SNR at the receiving station. I propose to trade time for power.

A general outline is as follows:

Based on station locations, historical data and atmospheric modeling, the transmission times and frequencies would be chosen for best propagation. During the transmission as the propagation changes the stations would shift frequency. If there is propagation on more than one band both might be used. A major issue with DX on HF is multipath propagation, however, I suspect that could be measured accurately enough to phase shift multiple arrivals at the same antenna and sum them coherently. Seismic processing is full of such shenanigans sometimes to attenuate noise and other times to boost the signal. A Raspberry Pi 4 can perform a mind boggling number of FLOPS.

For efficiency messages would need to be packetized and the sender would repeat the same packet until it received an ACK from the receiver in the form of a correct CRC.

An operator to sum multiple arrivals in phase is easy to implement once you know the statistics of the arrival delays. So once a packet is correctly received the actual signal can be analyzed to model the multipath and correct for it. That filter can then be used to speed up acquisition of the next packet and update the operator.

As an example, suppose the sender and recipient each had a pseudorandom sequence generated from their respective call signs. A low data rate signal a la QRSS is spread out over a wider BW by the PRN of the intended recipient. The sender is also listening for transmissions using its PRN so it can receive the packet ACK. Yet another version of the Aloha protocol from which ethernet and packet radio grew.

By *design* the GPS signal at the receiver is 56 dB (IIRC) below the noise floor of the receiver. 64 dB of processing gain makes it work. That is a *lot* of gain. There is a detailed discussion of the design in one of my spread spectrum books, but I don't recall which one and they are pretty thick.

This would need to be developed at the 2-3 W level over shorter distances before lowering the power. I'm in central Arkansas in the USA, so I've got 1500 miles (2500 km) E-W to the coasts.

However, I am still a long way from starting to build such a system. First I need to master basic RF amplifier and oscillator design. I mentioned the project to make clear that the group is intended to discuss anything from very exotic DSP techniques to basic small signal RF amplifier design.

Have Fun!
Reg

An additional comment that came up in a PM reply:

Send a packet multiple times until the sender gets an ACK in the form of a correct CRC. Once the packet is decoded place the entire recorded signal in the A matrix, the decoded signal in the y vector and solve Ax=y for the deconvolution operator that best transforms the signal into the corrected packet. Use that operator to deconvolve the next packet. Do that with each packet so that the propagation inverse operator is constantly updated. Effectively it should be diversity reception from a single antenna.

Not sure how well that would work at HF, but it's a minor variation of standard practice in seismic processing to eliminate water column reverberation.

Have Fun!
Reg

Reg,
I am probably stating the obvious and oversimplifying but WSPR is a great starting point to really see low power propagation.? I am in Northern Illinois, USA and have been running a RPi Zero W (previously used a 3) with the TAPR WSPR shield outputting 100mw into a wire antenna inside one of my out buildings in a fairly low noise area.? The antipode of my QTH is somewhere in the Indian ocean but when propagation is good I am routinely received by VK and ZL stations.? On my rx side I mostly use a ?BITX with an EFHW antenna and can rx VK and ZL when propagation cooperates.? I could probably do better with better antennas.
Doug
AC9RZ

Reg,
? An easy to use and fairly inexpensive RF source is the RFzero? ? ? ? ? about $64 + shipping, designed by?Bo, OZ2M.? Frequency range?2289 Hz to 200+ MHz, with frequency resolution on the order of milliHertz.? Output power +13 dBm from 400kHz to 200MHz.? It has a built-in GPS receiver, so it can frequency-lock to GPS.? It's controlled by an arduino microcontroller, and the source code is available for several working applications, including a WSPR beacon.? (You will need to add a low-pass filter to suppress transmission of harmonics.)? ? Its modulation capabilities are limited; certainly low BW FSK is possible.
David WA8YWQ

My RFzero boards and accessories arrived yesterday. Quality is excellent. I've not powered anything up yet as I am still collecting documentation.

After more thought, my plan is to connect an RFzero and step attenuator and a diode noise source and attenuator to a hybrid combiner whose output is the input of an RFzero. Then start measuring achievable processing gain for different modulation systems. At ~13 dBm output, I can safely connect most instruments to it without worry.

The ARM M0 just doesn't have the resources to do anything exotic, but a Raspberry Pi 400 certainly can and is an ideal host leaving all the resources of the RFzero to implement convenient control protocols.

Unfortunately, I shall have little time to play as I have a short weather window in which I need to get some outside work done. But I should be able to do some numerical simulations of various protocols.

Thanks, Bo!

Have Fun!
Reg

Hello Reg and the group

Glad you like your new toys. I am sure your endeavors will keep you busy for the winter. The RFzero are assembled by an EMS so the quality is fully pro:

Indeed the MCU is not powerful enough for sophisticated tasks. Joe, HS2JFW, has a remote RPi + RFzero setup running:

Bo
www.rudius.net/oz2m :: www.rfzero.net