 I want to do it for a really long time, which is to build radio transceivers for Amateur Radio. So the generally Amateur Radio is done with commercially built equipment, which is great, but that's sort of a bit divorced from the hardware itself. So that's an off-the-shelf radio. It does some absurd range of frequencies and modes, not a lot of power, but extremely flexible device, but it's also not very practical to muck around inside it. So I've wanted for a very, very long time to build my own radios from scratch, and so I've cheated a little bit. I've started with a kit, but this is the first one I built, which is this guy here. It's a single-band receiver, and it's only the front half of a receiver, so it backs on to a sound card, and then you do most of the work in software and the computer. So this is no longer a particularly popular designer, it was popular about a decade ago. Jeremiah Swong bought a bunch of these things and never had time to build them, so he gifted them to me on the condition that I do actually eventually build them. It's taken me two or three years to get to that point, but I've finally started. So the context was the reason to make a fair or make an extra vaganza, as we now call it. Because it was held at the Science Centre, I felt the need to build another ridiculous antenna. So it did, although the speaker is in the way, the top of the antenna is just barely in frame, but that's only because we didn't raise it to its full height. This is only 20 of the 26 metres, so the deal is that you lay out 20 metres that are on the ground, like this, you're supposed to get a fishing pole, you stand it up, and then you push the remaining 6 metres up. I think I have a shot with someone on the ladder, which will make clear how that works. Well, it gives you that idea. So, yes, okay, not so obvious. The segments, oh for God's sake, stop that. Why is it? No, I think I've pressed F5, which is interpreted as slideshare rather than full screen, so let's not press F5. We'll solve that problem. So it packs, the whole thing packs into a box about this long, somewhat impractical, it means you need to bread a van to move it around, which is the reason for the ladder. So the segments are here, and so in this configuration, the bottom about five segments are all collapsed. So we've pulled out the top 20 metres, and there are basically hose clamps with a bit of lining rubber, and then covered in heat shrink. When you have one piece going into another, like this, you put a hose clamp around the top and tighten it to stop it from falling, because the usual way of doing it fails is to fall into itself. So the top 11 were tightened, the bottom four are all loose about here somewhere, and the pieces are all collapsed. And it was just too hot and too unpleasant. Not much is obvious in the photos, but we were drenched. There was not even the slightest breeze, which is great for bringing up the mast, but very, very inconvenient for the people who were putting it up. You were with us on the previous Saturday or the... I was before the event itself. Yeah, so that's this day. So you are somewhere in these photos. Sorry, the photos are really out of order. I'll show you what I've got in some sort of rational order. So the way the receiver works is a little bit unusual. The way most contemporary receivers work is what is called a super heterodyne. So you've got a front section, which is the antenna and ground, maybe a preamp, I think called a frequency mixer with a variable oscillator. And so the tuning of the radio is achieved by changing the frequency at that point. And the mixer creates some indifference. So you end up typically, you have a... If you've got a FM radio receiver, you're tuning your 98 MHz. Typical what's going on is that the VFO is at 87 MHz. And so the mixer, what then goes out into the intermediate frequency stage is 11 MHz. At about a center of 11 MHz. And so what you do is you have a fixed frequency intermediate section after the first mixer. So you're varying the VFO when you're turning the tuning dial, but the variable frequency oscillator's output is 11 MHz below or above the frequency that appears on the dial, meaning that the thing that is in the intermediate section is an 11 MHz signal. And then you do all of your filtering and demodulating and other processing on a signal that's always at the same frequency. You then have another mixer and an 11 MHz oscillator to bring it down to baseband to get the audio out. So this is a typical superheterodyne arrangement for a radio. Superheterodyne meaning multiple energies or frequencies. This one works a bit differently. So most importantly, it uses a fixed frequency oscillator. It's a crystal. And there's no bending, the crystal is simply running at its resonant frequency, which is approximately four times the frequency of interest. So for the one that I've built, I'm interested in approximately 7 MHz or what amateurs call 40 MHz, which is because the wavelength is about 40 MHz long. Therefore the crystal is at about 28 MHz. So you bring in from your antenna and counterpoint system or your antenna system, there's a low pass filter and then a simple power splitter, which is just a hand wound inductor. It's not exactly the one I photographed, but it's a similar idea that in this radio there are two hand wound inductors that are something like that. It's a little yellow torus that is perhaps a centimeter in diameter and you just sort of winding wire around it by hand. This is part of the low pass filter. So that's this one here, which has a single winding. There's another that has three windings. And all this one, this splits the power in half. And the reason for that is that this is a front end to a digital signal processing system where it's in general DSP is done with in-phase and quadrature measurements. So instead of just digitizing the way a sound card digitizes for sound or MIDI setup does to put out a stream of samples to a speaker where if you're at 44.1 GHz then it's every 44,100th of a second you're taking a sample. In DSP you take one at that rate and then one offset by a quarter of a cycle. That's why the crystal has to be at four times. So in essence you're taking two 7 MHz samples, one offset by a quarter cycle from the other. And then that passes through an analog switch and into the left and right audio of a high-definition sound card. And so then the radio itself is a piece of software. So it's really a very simple bit of electronics. And also later the dividers are just a pair of flip flops to get a two-bit number that goes into an analog switch. As described you've got nothing more than a low-pass filter, a coil to split it into two loading resistors and then some biasing for the two op amps that feed the sound card. There's literally nothing else on the board. So it's sort of a while to work it out and various things like learning how to solder surface mount. That was exciting. These are fiddly things and I've started working on the transceiver which is a bit more complicated and has even more fiddly chips on it, including chips that are smaller. So there are three and five-link chips that are smaller than this and others which have links that are embedded inside the case and all sorts of complicated stuff. But that was cool. It all worked. No components damaged, not one joint failed, but it took me a number of nights to get it together. The winding of the coils was I think the most interesting piece but actually it's pretty simple. It really is just winding wire and inductors and at this level it turns out not to be very precise. You can't have a partial winding. If you're putting 11 windings, I think the left one's 11 windings, on a coil you can't have 11.1 windings or 10.9 windings. It's exactly 11 windings. So whatever your circuit needs the inductance to be, there's got to be somewhere else in the circuit to deal with the error. And so there's two things going on. One is we don't care all that much. So the fact that this thing is plus or minus 10%, simply because you can only have an integer number of turns, the capacitor is chosen to cope with that. In fact I measured it with an LCR bridge. It's almost 15% higher than what the design of the circuit calls for. I don't know why, but I'm reasonably convinced that the meter is correct. So yeah, this is an interesting piece that's avoided because there's no meaningful counterpart in solid state. In solid state you can have resistors, you can have diodes, you can have transistors, you can even have small capacitors. Granted you often need additional capacitors to compensate, but there's just no good way to do an inductor. And so in general you sort of, where are possible, say well if I need a series inductor I'll just make a bigger parallel capacitor. The reason that doesn't work very well for the front-ender of a radio is that these two form what is called a tank circuit. And so there's actually energy being transferred back and forth between the capacitor and the inductor. And so if you, which is part of the circuit running at resonance, if you replace the inductor with just a bigger capacitor that doesn't occur. And so you end up transferring less of the energy that's coming in from the antenna. So there are reasons for the use of inductors at the sort of very low power parts or low signal parts of a circuit. In addition to the use of power supplies for smoothing noise, you'll also find them right at the front of RF systems because you can't do the same things with capacitors. But yeah, once you get past this very front-ender of the radio there are no more inductors that they're fiddly to work with. I just wanted to show you. So this was, so the program that I use is a thing called Qisk. The one that's normally used for this is I think called Rocky, which is a Windows app. But as I don't use Windows, that's not very practical. So I'm using Qisk instead. This is a conventional way for amateurs to look at, not just amateurs, to look at signals in a digital domain. The axes are frequency this way, power this way, and then negative decibels relative to whatever full power would be. If you've got a 24-bit ADC, which this does, then 2 to 24 minus 1 would be the top. At minus 100, say minus 90 dB, you're at about a billionth of that. Which is, I should be able to lift my head, but I can't. I think you're down to about two bits. No, I can't be right. A bit more. But you get the idea that with a 24-bit ADC, your dynamic range is substantial so long as the device can really do it. Allegedly, this, which is an ASUS zone art, can. The other thing you need, and this is because of the strange design of the, or the unusual design of the radio, that because words conventional ideas have a variable frequency oscillator up front to bring the input signal into a fixed frequency IF, and that then mentioned all your filtering and demodulation at a fixed frequency. In this case, the oscillator, it's not quite a mixer, but it behaves a bit like one. The oscillator is fixed frequency. If you wanted to tune, what you actually have to do is operations in software. And I haven't got it working, but the, you can tell at the time, I'm probably a bit other. But if you, in order to receive a signal, which is more or less what's happening here without much success, is you tune the radio program to say, operate at 10 kHz above carrier or above center. And you also specify a width, how much spectrum to convert. The thing below is a time series. So in this case, the right now is the top, this is 10, 20, 30, 40, 50, 60, perhaps it's, it might be 45 seconds, top to bottom. And then color is used to indicate intensity. So what you can see is off the edges here, there's basically nothing. There's something here, even though you can't, and this is again a normal part of a spectrum display, and it's why am I just use waterfalls, because the waterfall will allow you to see a signal that you can't see just by looking at the instantaneous intensity graph. But this yellow here tells you there's something happening around here. And so if you sort of tune your radio to the middle and set the width to about that width, you'll get, you'll demodulate whatever it is. It doesn't look like voice, and it doesn't look like a clean digital signal. I would hazard a guess. It's a piece of industrial machinery making noise. But that's the sort of thing that allows you to do. The other big thing that this allows you to determine is how wide the response of the sound card is. The fact that the thing is rolling off quite sharply, it's not quite vertical, but it's certainly almost 45 degrees, means that there's something going on the sound card whereby at approximately 24 kHz either side of the centre, there's nothing, meaning that the passband is about 48 kHz wide. It is supposed to be 192 kHz. In this configuration, I don't know if you can see it. Nope, doesn't show it. In theory, it's supposed to be something at 192 kHz and it's supposed to have sufficient passband in the analog front end to do that, and the radio certainly has it. This is interesting because, with one exception, all of the amateur bands in HF fit in 200 kHz. So if you've got a 200 kHz wide ADC, then you need one crystal per amateur band of which they're about height. So you sort of build, as this is designed, actually build a separate radio per band. So that's the guts of it. This is various construction things. Ah, this is the power splitter. There's a problem with loading, which is why they're unbalanced. I haven't yet got a very good explanation of what happened there. But this is measured, basically these two red dots. So I put in a carrier, just had my radio transmitting with no modulation, and a line of attenuators, so it's not to damage anything. And then measured here and here, and what you're seeing is what you expect to see, which is two signals that are sine waves, give or take the quantization noise in the instrument, and that are exactly out of phase with each other, which is important for the design of the rest of the circuit. The reason why the amplitudes are not the same, I am not sure. But I swapped them over, and the curves stayed the same shape. If there was a problem in the device, the yellow curve would be small, the blue curve would be bigger, but that didn't happen when I swapped them, they stayed exactly in this relationship. So there's some sort of instrumentation problem, I don't know what. And the other, yeah, if you look at the centers, I was transmitting about 7 MHz. I haven't got the screenshot, but it comes out correctly. So very pleasing to have, and the oscilloscope showed very simply that the thing was doing what it was supposed to do. The other thing was the... This is a bit less clear, but I wanted to get a sense of what the filtering at the front end was doing. And it actually has a bit of a band pass rather than purely a low pass character. But I realized the other reason that the precision of the inductors doesn't matter is that all the filter has to do is prevent a leasing at half or double. And so it doesn't have to be within 10%, it only has to be within 100%. So it can be very, very rough. And so although I don't trust the calibration of the spectrum analyzer, what I did here was put a... I have a wideband noise source, cheap things, I suppose I could diode with some amplifiers in front of it, connected it to the input of the radio front end, and then once again tapped... I think that I tapped here or here, but we were looking at the frequency response across the band, and it's doing more or less what you would expect, which is to cut off... That's 4 MHz of spectrum and the center... I can't read the center in this photo. Is that any clearer? The center is 7.1, OK. So this is more or less the plan. The crystal is about 28 MHz. The middle diameter band is 7.1 MHz, which is about there. The whole thing is about 4 MHz wide, and as you can see, there's quite a clean roll-off, or quite a clear roll-off, to call it 2.5-3 MHz either side of the carrier, meaning that ALSs don't get in. What this does not solve is if you have a nearby powerful transmitter, so up until fairly recently, BBC ran its shortwave transmissions from here in Singapore, and so if I was listening to an ammeter, it would pop across the world, and at a frequency very close to the BBC's transmissions, then there would be harmonics and modulation distortion and other spores that risk swamping the signal and after, despite that they're actually separate, because this filter is very broad. But enough for the purposes of running the radio. That's... Ah, yes. So the antenna analyzer is a tool for analyzing antennas. This is what the antenna looked like up on top of the mast. So it was just wire, this yellow wire you can see, cut... No, it wasn't, it was two other guys who were there with me, one from the Science Centre and the son of one of my friends. I said, yeah, cut up these 10 metres long and two bit 20 metres long, and you've got the 10 metre vertical and then the two 20 metre bits coming from the middle, and that's the two halves of the antenna, and there's nothing else. There are no loading coils, there's no magic, there's just a bit of coax, and you can see the centre of its resonance is at 7.11 MHz. The amateur band centre is at 7.1. This is within a quarter of a percent of the, or eighth of a percent of the middle of the band, which was pretty awesome. At its standing wave ratio was 1.13, which is getting pretty close to practical optimum. A theoretically perfect antenna is a 1.0 standing wave ratio, meaning all energy that goes out goes out and nothing comes back. In this case there's an amount of reflection, but ceramic is anything below 1.3 is generally considered near enough to perfect, and with help most radars can work with anything all the way down to about 3. So this was, but this is two and a half, so the amount of band is 100 kilohertz is here, so it's below 2 across the entire band, beautifully clean antenna. Part of that is it was far above the ground, so often you make an antenna by having a vertical wire and then you just lay radars on the ground, which they're then coupled to the ground and messy. In this case they were 10 meters above the ground and therefore behave the way the mathematics say they behave. Worked fairly well to receive with the commercial receiver, transmitting mixed. Unfortunately I had software problems on the day and so I was unable to use the receiver, so a lot of mucking about. This is how we got Wi-Fi out of the field. Basically I put a 100 meters of wire from our booth, the Hackerspace booth, in the make of extravaganza out to the loading bay and then that's, the coax goes to the booth and then that's just a Wi-Fi access point with a 10-fold gain antenna. It's basically pointing 100 meters out to the field where there's like a Raspberry Pi and the sound card a bit. So yeah, all of that worked. The antenna worked, the mast worked. There's no software running on the laptop. Such is life. I think that was all. Yeah. Oh yes, and because it was a science tender of course it was appropriately signposted. We put a big coordinate around it. Because it was unattended. I didn't want people wandering in there and complaining about that they'd tripped on the rope and had the thing fall on their head. But also it's Singapore, right? You get thunderstorms. If we'd got this up to 26 meters it would have been more than anything in the vicinity including lightning conductors. So it's basically a freestanding lightning conductor. As a result, this is the other reason for using Wi-Fi rather than running Cat5 all the way out to the antenna. Because if it's struck by lightning like this then okay, you vaporize the mast and equipment and I'd be very sad. If it's struck by lightning and there's a Cat5 cable going back in to make a fair, then we have a problem. Fires, deaths, all that sort of thing. So yeah, this comes up and the Science Center is now getting a bit more thorough about risk management. This year they actually required risk assessments like formal documented risk assessments which they've never done before. But yeah, I was concerned enough that I did something similar last year. This is the somewhat underpopulated booth and this was our testing access point inside the fair. There's like dozens of Wi-Fi networks being right next to the access point meant that it worked. I think that was all. Questions, comments? Didn't you have a whole adventure with the ADC side of things on the... Well, I haven't resolved it yet. Sorry, that was the point of the... The Facebook post about that. Right, so the difficulty here is that quite clearly this runs from about minus 23-ish kilohertz to about plus 23. So that's 46... 48 kilohertz wide. So this is a... It's called an audiophile gamer's sound card. I'm like, yes, okay, fine. Mostly it's about output. There's a super high-frequency, super high-resolution output and 7.1 so you can do all the effects and surround sound. So it claims 192 kilohertz on the input, but this is probably fake. In fact, Adnan spent some time looking at chipsets and it's not just at the card level, it's always down to the chips that it's likely to be untrue. And so I had configured it to 192 kilohertz and yet you can see from this that it's very clearly only processing a 48 kilohertz wide slice of spectrum if it was feeding... So in other words, although the software is handing over a hundred and two thousand seconds, that samples per second what's really going on is it's 48,000 samples, each of which is being handed over four times. Just the same sample again and again and again. And so that then means that there is no component of the signal that has a more than 24 kilohertz deviation from the center. And so you therefore, we're only decoding 48 kilohertz around the middle of the band. And the point I haven't got out and sort of bought this thing was that it should be capable of processing 192 kilohertz for the entire 7 to 7.2 megahertz that Amit has used. Give or take where the crystal is actually operating. So yeah, I just think that really sucks. So I posted this on Facebook and quite a few people hopped on and commented and went looking around and it turns out that this thing really can do what it's supposed to do. There's a data acquisition community who uses basically sound cards for lab use and therefore they care a lot about the behavior of these devices which is the best and revised and someone found a review of six or seven such devices. This one topped the review. So it's credible. The software community, which is the community people who make these particular category of radio, several people claim that yes, they are able to get the full 200 kilohertz or 192 kilohertz width, which means this graph would be four times as wide. But yeah, I haven't run that one down yet. There's a bunch of things. There are multiple inputs. There's questions about how to configure it. And there's certainly the possibility that the performance that the software guys are seeing is only possible with the ASUS drivers on Windows. So I don't know. I just sort of fired it up. I said yes, do 100 kilohertz because obviously if I tell you that, you will do that, won't you? Pulse audio is definitely supposed to be capable of it although it behaved the same way but from looking at it, even though it's claiming 192 and providing 192, it's very clearly only providing 40 kilohertz width of samples, which means basically repeated samples. So yeah, I haven't put any time to it since Maker Faire. I hold all the gear back, put it all away, went to bed, then got sick. So it was unwell for most of the following week. Thankfully that didn't occur until the day after Maker Faire. That would have been appalling. So no, tonight's the first time I've played with it since. Yeah, no progress yet. But given, yes, that this was a post to Facebook and a whole level jumped up and found useful stuff, so I'm fairly optimistic that it will be possible to do the full 200 kilohertz wide slice and therefore to, using this super simple design, listen to the full 40 meter amount of demand. Don't know whether I'll build the rest. I've got enough of the kits to build about five so to do different bands. But what I really want to do is transmit. That's kind of the point of the exercise. And so I've started work on the transceiver. It was not ready for Maker Faire and still isn't, but I'm working towards. Which then points to the next part of this puzzle, which is regulatory. And IMGA either didn't understand what I said or didn't care. It's not yet entirely clear to me, but as it stands, they have been advised in writing of my intention to build an operator set of soft rock kits. No one's done it. I've seen people who have built their own radios before. In fact, Jeremiah's would have been the first had he done it. So there are two ways you can deal with this. One is not bothered to mention it to IMGA. Actually, so it's not quite true. You've built your own radius. Started to. So almost no one, but certainly no one had any idea how to get through IMGA's hoops because we are still settled with an obligation to have every radio that's part of our license or part of our station listed on our license right down to serial numbers. I believe Matt has spoken to IMGA and they say that as long as you can come up with test reports showing the performance of the radio, you can get it approved. But I guess for most people that may not be a fairly drill pass especially don't want to test. So yeah, there's a couple of things there. I had a careful look at the IMGA spec and there's not much. There's only four things you've got to verify that the frequencies are correct and that's now easy to do with a GPS discipline oscillator. It used to be the case you had to have really sort of expensive standards and you had to have a traceable path. Now you just get an oscillator. It can be rubidium, it can be quartz and a GPS receiver and a feedback loop to discipline and you will get a frequency standard correct to around 10 to minus 11 or 10 parts per billion and this is like a hundred dollar item, maybe ten dollar item. So it's very easy to do frequency. So let's really just sit down and verify that the radio doesn't sort of only set to frequencies that the spec sheet says it sets to. Power levels similarly are not that challenging. You need a power meter, you need a dummy load. Strictly speaking I've got a dummy load. But they're easy enough to get and certainly I've got most of the, everything here except the power meter actually. I can prove by construction that I can't, I don't exceed the legal limit but of course they want to test it against the spec sheet. Emission modes, I've just made a written submission to IMD age to remove that, that doesn't make any sense at all. It made sense in the 1960s. The standards that are part of the regulation in the handbook appear to date from pre-independence and like I even found, coincidentally while trying to decode one of the phrases which says something very specific on the footnotes I'm like I don't understand what those words actually mean. I mean individually yes, but I don't know what point is being made. So I punched the phrase into Google and what I found was a 1965 Amital Radio Handbook from the UK. Like, and like even the weird things like the footnotes have both letters and numbers for no really apparent reason. Like seven footnotes with numbers and then forward letters like why? I repeat it. From, yeah, from this 50 year old document. So, yeah, at the moment I'm, it's just, you need to basically say yes I tested it and it can only use modes that are listed in the Amida grants within Singapore, but hopefully they'll just go away. The only difficult one is spurious emissions. It's not even clear from the way the standard is written whether you're obliged to have testing done for spurs. It's a very strangely worded sentence, but the best assumption is that you are obliged but the standard for HF gear is not difficult. It's minus 40 dBc. So all that means is if you've got a spectrum analyzer that's not completely broken what you are looking for is this is a broadband noise source that's close to flat but if you're testing, so you've got your radio plugged into a dummy load and you're tapping or attenuated and you're tapping to input to the spectrum analyzer you'll key down for the transmitter and so your carrier appears for the sake of argument here. So you've turned the power level down and maybe you've put some attenuators in line to prevent damaging the instrument. What the standard is is that the spurs must be 40 dB below the carrier. So what you then do is count down one, two, three, four and as long as any spurs, usually they start at double at the first harmonic, are below that line, then you've passed. And so that's all. And so, yeah, what I said to IMDA was yes, I will... I'll find it actually, shouldn't be too hard to find. These attenuators you talked about come in 10 dB steps. Various. Usually 3 dB and 10 dB steps, yeah. So you just string them up for 20-30 dB? Yep. I didn't think it was photographic but I had them set up for exactly that purpose when I did the power splitter test because I didn't want to damage the instrument. Now that's the old one. Right, so that's my equipment register that I have provided to IMDA and they said that's awesome, thank you. This is basically proposed to build because I haven't built them yet and all I did was add it a condition, no use for it until the IMDA technical standard amateur radio compliance with technical requirements is met and that test records be retained for inspection. So I didn't even enter into the argument about who tests. Historically IMDA wanted radios sort of taking away to a lab to be tested like because of the three tests I've just described all you've got to do, it seems ridiculous to involve a testing lab. So I've just said I'm going to, before I put the where, perform the tests and keep records. We've got to keep records anyway. We've got to keep logbooks and we've got to keep a copy of the license. So this is not particularly onerous. So that's how I tackled it. And so I also provided spec sheets. Like okay, you want a spec sheet? All right, I'll give you a spec sheet. No problem. Hooray! Voila! Right, so these are the modes and these are the piece that I had suggested that should just disappear. What would this stand for? Okay, so A1A, okay, basically a code that stands for for what kind of modulation it is. So A1A will represent amplitude modulation of what kind. Okay, so there's a... There's a... Built-in... There's a key somewhere that explains... So all of this stuff harks back to the ITU radio regulations which are updated after each World Radio Conference. There's one happening now, or last week. And so they'll be... So the last conference prior to what's just happened was in 2015 and therefore the last version of the rigs came out in 2016. So in essence all the actions that were taken during the conference get sort of editorially assembled over the following six months. So I'd expect that by about sort of April we'll have a new version of the rigs. And so frequency allocations, use limitations, and in some cases emission modes. But I went through... It's a monster. It's a 200,000 page document. But there are no constraints on amateur emission modes anywhere in the world. There are all sorts of constraints on power in particular cases and a bunch of things. But there isn't a single case where there's an amateur service limitation on emission modes anywhere. So there's a combination of validation schemes and... So it's three things. The first letter is the modulation. And so A is double-sided band, or AM essentially. The second is how many modulated signals there are and also what they are. So if you're doing continuous wave or Morse code, one way to do it is literally continuous wave where you just key your transmitter as a sine wave at a particular frequency. Another fairly popular way to do it is using a subcarrier. And this is perhaps more relevant for... It's a bit meaningless if you're doing a single-side band or... Oh, sorry, for AM, yes. So if you have an AM transmitter, then you put, say, a three kilohertz tone on top. So you then got your carrier plus sidebands three up and down that are one quarter power of the carrier. It's kind of a wasteful way of doing it. You'd rather have all the power in the carrier and then reconstructing the receiver. So that's what that one is. One channel analog is things like single-sideband and various digital multiples, et cetera. And then the third character is a letter describing what it is that's being transmitted, whether it's beacon... Application layer, kind of. Yeah, more or less. You're transmitting alien data. Yeah, hard to say. I can't think of... There might be some examples. More detail, so images can be grayscale, color, et cetera. When this stuff was adopted, there was no question of having 200 different frame formats for images. So, right, but not relevant. Multiplexing, so bear in mind these codes are used to describe everything, in particular to describe mobile phone systems where this matters. How multiple parties sharing a piece of frequency cooperate is the... And then finally, although not listed in the Wikipedia page, you often get a number which is the bandwidth in kilohertz of the signal. So broadcasters usually use A3E. So, AM, single channel analog, voice phone, by the way. F80 is the same thing, but it's FM instead of AM. There's TV things. Sorry, so these are the bandwidth that comes before rather than after. So 11k2 is 11.2 kilohertz. FM3E, I forget what that is. NON is basically beacons. So, nothing, no modulation, no modulation form and no information content. It's a beacon for direction finding. How do you think planes work out where they are? We depend upon the set nav, in case the satellites work. So, in Singapore... We have transmitters for these. You bet. We have a number of them. Three or four? What we call VORs and DBs. I'm sure King Wing can tell you all about them. With relation to ships. Aircraft. So that is a ring of antennas. This is bizarre. This is... Using AM and FM on the same thing at the same time, so that the receiver knows... For the plane to work out the bearing to a beacon, it doesn't have to depend on its own instrumentation. Instead, it measures the phase angle in the two parts. No. Yeah. So, with just one beacon, you can immediately put yourself on a line. You know, just by measuring the phase angle between the two parts of the signal, that you are 222 degrees, maybe it's 5 degrees, 235 degrees, from a particular beacon which is identified by a call sign being played occasionally. And so there's... They're all similar sizes. No, the Alaskan one is harp. That's a different thing. But there are about four of these in Singapore, and they're all set up more or less this way. In probably, the trees have just disappeared for one little bit, and you've got this circular antenna and next to it, basically, a trans-modestation. What are they called? Directional beacons, but they're also... We need a non-directional beacon or VHM one-way directional range of VOA. But these are likely NDVs. No, no, no. This is what I'm saying. They have a two-part modulation, meaning that merely by being able to hear it, you can work out what direction you are from it by doing a phase angle comparison between the two parts of the signal. This is a directional beacon, which the VHF one is typically out. I guess you still need altitude, right? So you need to... Yeah, but you can get that with a barometer. Or enough. But yeah, so that's... I understood why it makes a lot of sense, but yeah, I'd never heard of it before. So that's... Can we go on and treat that? I don't know what the deal is. I sort of assume it's got one of those signs with like the stencil like this. And you know, please... Don't step on the grass. But I just... I was curious that we're looking, and all four of them are... Because of course their locations are public information. The pilots have to know exactly where a beacon is located, so it's not hard to find them. And that all four of them are just like this. They're a circular antenna in a clearing surrounded by trees. That's all right, maybe we made the point. Right. Probably enough rambling. Anything else at this point? No? I think that's all right. Let's go. Is it safe?