vk6flab

Recently I received a lovely email from Michele IU4TBF asking some pertinent questions about the Bald Yak project. If you're unfamiliar, the Bald Yak project aims to create a modular, bidirectional and distributed signal processing and control system that leverages GNU Radio.

The short answer to how I'm doing getting GNU Radio to play nice with my computer is that I have bruises on my forehead from banging my head against the wall. When I get to success I'll document it. To be clear, I'm not sure what the root cause is. I suspect it lies between the GNU Radio developers, the people making packages and the manufacturer of my computer. I'm the lucky one stuck in the middle.

A more interesting question that Michele asked was, for Bald Yak, what is the A/D and D/A requirement for making GNU Radio talk to an antenna?

This is a much deeper question that meets the eye and I think it serves as a way to discuss what I think that this project looks like.

Ultimately in the digital realm, to receive, an analogue antenna signal needs to be converted to digital using an Analogue to Digital or A/D converter, and to transmit, the reverse uses a Digital to Analogue or D/A converter to make an electrical signal appear on your antenna.

The specific A/D or D/A converter determines what you can do. The sampling rate of such a converter determines what frequencies it can handle, the sample size determines the range of signals it can handle. You can compare it with a video screen. The sample rate determines how many pixels on the screen, the sample size determines how many colours in each pixel.

The sample rate of an A/D converter is measured in samples per second. If the device only has one channel, you could think of this as Hertz, but if there are multiple channels, like say a sound-card, the sample rate is likely equally divided across each channel.

You might have a sound card capable of 384 thousand samples per second, or kilo-samples, but if it supports simultaneous stereo audio input and output, only 96 of those 384 kilo-samples will be allocated to each channel and only half of those will actually help reconstruct the audio signal, leaving you with 48 kHz audio. In other words, the advertised frequency response might not have a direct and obvious relationship with the sample rate.

At the moment I have access to a few different A/D and D/A converters. The simplest one, a USB audio sound card, appears to do up to 192 kilo-samples at 16 bits. The next one, an RTL-SDR tops out at a theoretical rate of 3.2 million or mega-samples at 8 bits. The Analog Devices ADALM-PLUTO, or PlutoSDR handles 61.44 mega-samples at 12 bits.

Now, to be clear, there are other limitations and considerations which I'm skipping over. Consider for example the speed at which each of these devices can talk to a computer, in this case over USB. I'm also going to ignore things like mixers, allowing devices like the RTL-SDR and PlutoSDR to tune across frequency ranges that go beyond their sample rate.

Each of these three devices can convert an analogue antenna signal into bits that can be processed by GNU Radio. All of them can also be used to do the opposite and transmit. Yes, you heard me, several amateurs figured out that an RTL-SDR can actually transmit. Credit to Ismo OH2FTG, Tatu OH2EAT, and Oscar IK1XPV.

The point being that whatever Bald Yak looks like, it will need to handle a range of A/D and D/A converters. As I've said previously, I'm aiming for this to work incrementally for everyone.

This means that if you have a sound card in your computer or an $8 USB one, this should work and if you have an $33,000 NI Ettus USRP X410 lying around, this too should work.

Also, if you have an X410 lying around not doing anything, I'd be happy to put it to use, you know, for testing.

So, kidding aside, what about the rest of the Bald Yak experience?

GNU Radio works with things called blocks. Essentially little programs that take data, do something to it, then output it in some way. It follows the Unix philosophy, make each program do one thing well, expect the output of every program to become the input to another, design and build software to be tried early and use tools rather than unskilled labour.

Amateur radio transceivers traditionally use electronics blocks, but if we move to software, we can update and expand our capabilities as the computer we're using gets faster and the GNU Radio blocks evolve, and because it's all digital the computer doesn't actually have to be in the same box, let alone the same room, it could be in multiple boxes scattered around the Internet.

So, the idea of Bald Yak is a collection of blocks that allow you to do radio things. You might have a separate box for each amateur radio mode, AM, FM, SSB, RTTY, CW, WSPR, FT8, FT4, Q65, but also modes like Olivia, FreeDV, SSTV, Packet, PSK31 or Thor. Instead of having to figure out how to wire these modes into your radio and your computer, the infrastructure is already there and you just download another block for a mode you want to play with.

We'll need to deal with variables like which A/D and D/A converter is being used and what their limitations are. We'll also need to build a command and control layer and probably a few other things.

I'm considering a few other aspects. For example, GNU Radio is mostly run with text files. We might distribute those using something like a web store. GNU Radio is proving hard to install, perhaps a LiveCD is the way to go. We'll need to come up with a base level of functionality and the documentation to go with it. I'm still contemplating how to best licence this all, specifically to stop it from being exploited. Feel free to get in touch if you have ideas.

I'm Onno VK6FLAB

Have you ever come across a solution to a problem that you sort of knew you had, but didn't really appreciate until that moment? I had one of those recently. To set the scene, fair warning, we're not going to solve this today, we're still very much shaving yaks, but there's plenty to take away.

So, the scene.

I'm hosting my weekly net. It's going well. All the internet links are up and running again, thanks to the hard work behind the scenes of several unsung heroes, I can name a few, Bob VK6ZGN, John VK6RX and Rob VK6LD, but there are plenty of others whom I don't know and who have yet to stick up their hand to say, I was there. Regardless, thank you.

Anyway, I'm hosting my weekly net, F-troop. A curious thing is occurring. Two of the stations are emitting a tone during their transmission. I'm pretty hot on how things sound, so I ask. We talk about it for a bit when Allen VK6XL comes in and tells us that according to his spectrum analyser it's a 1 kHz tone with harmonics and it's on all transmissions, just audible on two.

This starts a conversation about spectrum analysers when Allen mentions that he's using an audio spectrum analyser, a piece of software running on his computer. The software has a copyright from 1999 and based on the documentation I saw, has lots of excellent functionality. I might even be able to run it on a Linux machine using WINE, but that's an adventure for another day.

Randall VK6WR points out that I could use the spectrum display on Audacity. This is a much more current piece of software, but it's not intended for real-time use, it's what I use to edit the audio after recording my podcast. Not even sure if the spectrum display can show during recording, I've never tried.

In the past I've used SoX, the Swiss Army knife of sound processing to create sonograms, but that too isn't real-time.

Then it hits me. I have a real-time tool. I've been playing with it for weeks. GNU Radio. Surely it has a spectrum display, and indeed it does, several.

So, I already have a tool, purpose built for processing signals, that can do all the things I'm looking for and some I've not yet imagined.

Before I proceed, I'll remind you that we're in the middle of the Bald Yak project, so named because by the time we're done there won't be much hair left, if any. In case you're unfamiliar, the Bald Yak project aims to create a modular, bidirectional and distributed signal processing and control system that leverages GNU Radio.

So, boldly clicking about, I set on the notion of making a block called "fosphor" work. Depending on which description you use, it's an Open Source, GPU-accelerated FFT and Waterfall display tool. What that means is that it uses a graphics processor to do the heavy lifting and has the ability to show signal levels across frequencies and on a waterfall display. Apparently it's a block for RTSA-like spectrum visualisation. I'm fairly sure that doesn't mean Railway Technical Society of Australasia or has any relationship with Reverse Total Shoulder Arthroplasty or the Road Transport and Safety Agency of Zambia.

I'll admit that I didn't see the GPU part of that description until several days later. Had I seen it at the time, I would likely have carefully backed away and shelved the idea, but that's all water under the bridge.

To cut to the chase, I have yet to make this show a single pixel. I smelled trouble for the first time when I discovered a post asking if anyone had gotten this to work on a current release of Debian.

I came across a lovely post by what appears to be the author helping some hapless user, and I'll confess that's the camp I'm currently in, to make it work. I have no doubt that I can make it work, but that's going to take some effort.

Now, at this point you might ask me why I wasted your time with this tale of woe?

Well, the answer is simple. This is what "Yak Shaving" looks like. You solve a thousand little problems, one at a time, and if you manage to keep track of what you're doing and why, you can get stuff done.

This applies here, but it also applies in your life, in radio, in antenna building, in making a contact, in participating in a contest, in activating a park.

Each activity reveals myriad issues that you'll each need to resolve. The more practice you have at this, the better you'll get. I will point out that for me it's not without stress. When I go though intractable problems I'm often as grumpy as a bear with a sore tooth whilst my brain is running like a hamster in a wheel generating kilowatts of power.

This too shall pass.

Oh, because I know it's bothering you. RTSA, Real Time Spectrum Analyser, obvious, right?

I'm Onno VK6FLAB

Recently I built a first attempt at a noise cancelling circuit, on my couch, in GNU Radio, without holding a soldering iron and running the risk of the room smelling like burnt chicken, because if you believe the Internet, sometimes holding a hot piece of metal by the hot end is not the best way.

The idea behind the circuit, or more accurately, flowgraph, is that you take a signal from two sources, invert one, combine them, and they cancel each other out. If the signal with the noise only contains noise, then you can, at least theoretically, remove the noise from the actual signal.

Before you think that I'm inventing something new, I'm not. I'm merely attempting to recreate the same notion I came across decades ago, where you combine the signals from two microphones, preferably identical, reversing the wiring in one and talking into a microphone whilst holding the other one away from your mouth.

I did essentially the same thing using RF signals from two RTL-SDR dongles.

Randall VK6WR pointed out that, aside from misusing the word "mix", which in electronics really means multiply, but in audio means combine, Randall suggested I use "add" and "subtract". I'm still working out how best to name things, because we're talking about audio and RF, sometimes at the same time. Perhaps that's where I went wrong. I'm currently using "combine" as my technology neutral word, but I'm happy to take suggestions.

All that was the side show, because as Randall points out, doing this in RF is much harder than in audio. This is already something I knew. At the time I didn't really know how to get two different but the same sources of audio to experiment with, so I started in the deep end at the RTL-SDR dongle side.

Now, armed with the encouragement from Randall I built a horrible thing, which is easy when you just drag and drop blocks on a screen. I built two independent FM decoders that use the exact same parameters, so they're tuned to the same frequency, they're amplified and tweaked identically. The only difference is that they each decode a different dongle. I then piped each of those into my magic noise cancelling circuit and tried again.

Aside from dealing with hardware restrictions, causing things like buffer under-, and over-run, that's when the computer isn't processing all your samples, or is getting ahead of itself and is running out of samples, I can make audio come out of the speaker in my computer. I can prove that there are two signals, by setting the amplification of either to zero, and still get sound from the other source, however, noise cancelling, no matter what I tried, didn't work.

Then I decided to simplify, rather than trying to cancel out "the Heat is on", word of honour, I'm not making that up, that's the song that was playing, I went back to basics starting with a tone. I fed the same tone into the noise cancelling block twice, once as signal, once as noise. Magic, the cancelling works.

I also learned that changing the frequency of the noise and changing it back gets you into all kinds of problems and even if you send the same tone, one shifted in phase by a known amount, getting the two to cancel each other out is non-trivial.

You might think that this was all a complete waste of time and if you're just driving past it looks like a swollen electrolytic capacitor about to burst your bubble, but it's not that bad.

Here's what I learned from this little adventure.

I can make hierarchical blocks out of flowgraphs. This is important because at some point all the functionality associated with Bald Yak will likely end up being implemented like this. I also learned that such a block can contain user interface elements, which means that we can build blocks that know how to do stuff and tweak how they operate without having to build a user interface every time we use such a block.

I learned that we can implement an idea that would be hard using physical components and test it really quickly, in this case my available time was the limiting factor, not the testing. If I'd done this with components I'd still be trying to figure out where to get them from, let alone turn up the heat. Another bonus is that I didn't spend a single dime and I can dispose of it with the click of a button, rather than trying to figure out how to recycle components and circuit boards.

I also learned that the idea as I built it doesn't work quite as I expected and that things that I didn't anticipate, like changing the frequency, buffer under-, and over-runs, impacted my efforts in unexpected ways. There's a delay between making a change on the user interface and the effect becoming audible, and I learned I can make a dongle work on my computer and that installing GNU Radio is a challenge at the best of times.

In other words, even though I'm unlikely to use the noise cancelling efforts in their current form, there was plenty I learned from the experience.

From my perspective, this was a success. What have you experimented with and learned?

On a completely unrelated matter, long overdue, and music to the ears of some, can you spell SKCC, I've finally put all the Morse Code versions of my podcast on a thumb drive and plugged it into my car. During the week I've managed to listen to about two hours of Morse. While I don't know most of the letters of the alphabet, I can still detect letter and word boundaries.

I'm Onno VK6FLAB

Scientists have long struggled to accurately predict solar flares, but a recent breakthrough using NASA’s Solar Dynamics Observatory has uncovered a promising new sign: flickering coronal loops.

I use gedit for most of my text editing, but markdown support is very limited.

Things I've tried:

• vscode, too heavy and intrusive
• Google docs, only renders, doesn't show the plain text, need to manually export to see markdown
• Eclipse, haven't actually tried markdown, but I have no doubt that it's supported, but heavier than anything else
• atom, no longer developed last time I checked
• online editor, don't want to share my text and functionality is poor
• type markdown, save it and render with pandoc, lots of effort, but the results are good

Over to you.

Edit: Had some issues with my Lemmy client, moved to Voyager and hopefully I can fix things.

I was asked what functionality I require, which to be fair, I hadn't considered because I use my editor for pretty much everything.

Ideally I'd be able to use it to either see the raw markdown or the rendered version of whatever I'm writing, code in a dozen languages, articles, websites, legal documents, books, all of which I do pretty regularly.

The side-by-side view doesn't do it for me, I'd more likely than not have multiple windows open with different documents instead.

It should do autocomplete, syntax highlighting, bracket closing, live spell checking in a variety of languages, launch quickly, be rock solid when faced with a massive log file and allow me to add menu-items to run bash scripts that do things like calculate the time it would take me to read out the text at my normal podcast reading voice or covert weird characters into html-entities.

There have been many wonderful suggestions, most of them do the preview side-by-side which pretty much eliminates them as a candidate.

There are many suggestions to use a vscode floss version, but the biggest issue with vscode is its weight and I'm not sure if it changes by moving to the floss version. I note that my search for that tool brought me many AI features, which is why I did a hard pass and why I can't remember its name ATM. (Edit: Codium)

I've been using Debian since 1999 and still struggle with remembering the vi control codes, so emacs is unlikely to get in the door.

So, with that in mind, whadayagot?

As you might recall, recently I stumbled on an excellent list of 52 weekly challenges put together by Fabian, DJ5CW and friends. You can find it at hamchallenge.org. As I've previously mentioned, it contains activities right across the amateur spectrum, from designing a QSL card to making a contact on 80m or 160m, with everything in between. It's an excellent tool to set a weekly goal to achieve and I recommend that you have a go.

It's not the only interesting tool around.

Amal VU3FTH and Steph Piper, whom you might know as MakerQueen AU, have put together an "Amateur Radio Skill Tree". It's a collection of hexagonal tiles, each with a skill, displayed together in an attempt to track what you know and could know about amateur radio. The idea is that you print it out and colour in each tile as you complete it. You'll find things like "Explore D-Star", "Build a cw key", "Teach a friend about Amateur Radio", and plenty more. Can you get four activities in a row and which skills could do with more effort? There are 68 tiles ready for your colouring pencil .. Bingo!

The Radio Amateur Training Planning and Activities Committee, better known as the RATPAC, put together an "Amateur Radio Challenges Checklist for 2025" and published it on mastodon.radio. It's a list of bands, modes, activities, builds and clubs that help you track what you've been up to throughout the year. While we're here, I should mention that the RATPAC in their words "comprises Amateur Radio Operators of a wide variety of backgrounds and experiences", and hosts two weekly Zoom meetings so anyone can participate in the talks published on the ratpac.us website. If you're unable to attend, you'll find the presentation on YouTube.

Over the years I've been part of this amazing hobby, I've been telling anyone who will listen that there is plenty to do and see in our community, a thousand hobbies in one. The tools I've mentioned represent around one hundred and fifty activities and pursuits, but hidden behind each one is plenty to explore. For example, hamchallenge asks you to spend week 21 creating a GNU Radio flowgraph. That topic alone could fill a decade worth of exploration if you were so inclined. Similarly, one of the Skill Tree tiles is "Study RF propagation", something which you might realise is easier said than done. The RATPAC checklist has a tick box for "Work 100 countries", not something you're likely to achieve in an afternoon.

My point is, you can do as much or as little of this as you like, to what ever degree floats your boat. As you might know, I'm deep in the weeds with GNU Radio and I expect to stay there for plenty of time to come, but you are under no obligation to follow me down the rabbit hole. In other words, it interests me, but it might not do the same for you.

One final comment. None of these activities require you to upgrade your license, well, other than the Skill Tree tile "Upgrade your radio license". You can do most if not all of the activities I've shared with any amateur license and plenty of it can be done without a license at all.

So, what are you waiting for?

I'm Onno VK6FLAB

In case you're unsure why this has anything to do with Amateur Radio, the ionosphere is directly impacted by space weather, as is Earth's magnetic field.

The other day I was sitting on the couch lounging about when I came up with an idea and there and then I picked up a circuit board, soldered down a hundred or so components and built a noise cancelling gadget, all within about an hour .. right there, on my couch.

Yeah, that didn't happen, least of all, my soldering skills are not up to scratch, never mind the couch.

I did however build a noise cancelling circuit, on my couch, in about an hour, but no soldering iron was hurt and the room didn't smell of chicken afterwards. Instead I used GNU Radio to put together a series of blocks that allow me to apply noise cancelling to a signal.

How does this work?

Well, imagine a signal mixed together with the negative version of that signal, think of it as swapping positive and negative if you like. If you mix two identical signals together, one of them inverted, you end up with nothing, because they cancel each other out, where one is high, the other is opposite and low and vice versa. Mix a high signal with a low signal, you end up half-way. If they're the same but opposite, half-way is zero.

Hearing nothing is not helpful, but what if the two signals were slightly different, let's say, one is connected to a proper 10m antenna and the other is connected to an antenna that picks up local noise. We'll call the antenna on 10m the "signal" and the local noise antenna the "noise". If you mix these together, you end up with a signal and noise combined, but if you invert the noise signal, you can, at least theoretically, filter out the noise that's common to both antennas.

Now I did say "theoretically" and that's because while it sounds simple, it's far from it. Unless you have a special radio, the two signals are not coming from the same device and won't really be identical where it matters. For example, let's call it the volume or gain of one might be higher than the other. One might arrive slightly later than the other if the coax isn't the same length or the electronics in both radios are different. There are other things too, but let's just stay with this for a moment.

You could amplify or attenuate one or both signals to make them the same or similar levels. You could change the phase, think of it as the time when a signal starts, and synchronise the two signals in time manually.

Of course whilst you're doing this, inside GNU Radio, the computer is doing some serious math, which takes time to make these changes, which introduces further delays you'll also need to account for.

Building it was simple. Testing it much less so.

After coming to grips with the USB port on my computer, which for reasons best known to the manufacturer, cough, Apple, switches off ports that are in use, I managed to get two RTL-SDR dongles connected and working. This involved removing GNU Radio, which was installed using a tool called "homebrew", then installing it instead using "radioconda", twice, since the first time the installer failed with an error, actually three times, because the failed installation left all manner of rubbish behind, so that needed to be removed; then I had to disconnect my keyboard and track pad, because for reasons only known to the manufacturer, yeah, the same one, they won't play nice if there's an RTL-SDR dongle plugged into the USB-C hub, I finally got this running, which in turn involved figuring out what the GNU Radio "Device Arguments" look like for a locally connected dongle, in case you're wondering, it's "rtl=0" for the first one and "rtl=1" for the second.

Clearly this project is living up to its name, Bald Yak.

Now I can invert a signal, I can amplify and attenuate it, change the phase, shift the frequency, swap the I and Q, do a complex conjugate, and have a user interface that can change these settings as required.

I'm going to ignore the hour of my life I'm not getting back to understand how variables, parameters and user interface items hang together and how they interact. It's logical, but it takes a bit to wrap your head around it and I'm a software developer, so I don't envy you if you're not.

Anyway, I tuned to a local FM broadcast radio station and couldn't make any noise go away. I then discovered that my noise antenna was picking up the station just fine, so that didn't help. Then I swapped radios, actually, I just swapped the zero and the one in the "Device Argument" fields and tried again.

In the process I discovered how you can create a so-called "Hierarchical Block" in GNU Radio and to my delight also discovered that there is one that can have a user interface, so I can make a stand-alone block, that has a user interface, that I can use in another project, which is how I intend Bald Yak to function.

So, changing stations, I could finally hear noise, but still no reduction.

Then I realised that I was using FM, not single side band, so I started hunting for an SSB decoder but had to abandon ship to go and do life. Overnight I realised that if everything went to plan, the two FM signals should have cancelled each other out, making silence, which they didn't, despite my efforts.

So, at this point I have a thing that can alter noise and mix it with a signal for the purposes of noise reduction. That it doesn't work is potentially because I have not enough range in my adjustments, or too much range with not enough steps, just as likely it's because I'm missing some fundamental understanding somewhere. If you have any documentation I should read, please don't be shy.

As life adventures go, I'm hooked, but it's not without shaving Yaks, Bald or otherwise.

I'm Onno VK6FLAB

A coronal mass ejection is expected to arrive in the second half of UT day 24-Jan, possibly inducing G1 geomagnetic activity in the Australian region. A second coronal mass ejection is expected to arrive on 25-Jan, prolonging geomagnetic activity. Auroras may be visible in dark skies late on UT day 24-Jan in high latitude locations, and then there is an further chance for them on 25-Jan.

Aurora alerts will follow if significant geomagnetic activity actually occurs.