Below the Noise Floor — Episode 6: "Digital Modes on the Waterfall: FT8 and Friends"

FT8 has transformed HF amateur radio over the last decade - it is now the dominant mode on most bands. This episode explains what FT8 actually is, how it works, what it looks like on the waterfall, and how to set up WSJT-X alongside AetherSDR for decode-only reception. Plus WSPR, PSKReporter, and how to use digital mode decodes as a real-time propagation map.

Transcript

SPEAKER_00 0:00

Something changed on the HF bands around 2017 and the change was permanent. A new digital mode appeared called FT8, and within a few years, it had become the dominant form of activity on most amateur HF frequencies. If you tune across 20 meters on a typical afternoon and look at the waterfall, the majority of what you see is FT8. On some bands, during some hours, it accounts for more than half of all amateur radio contacts made worldwide on that day. That is a remarkable statement. And it deserves a real explanation. Because FT-8 is genuinely different from everything that came before it. Not just a faster or more efficient version of older modes, but a fundamentally different approach to what a radio contact is. This episode covers FT-8, WSPR, and how to set up decode-only reception using WSJTX alongside Ether SDR. No transmitting required. By the end of this episode, you will be able to watch the digital mode traffic on any band in real time and see contacts being made between stations on opposite sides of the planet. Let me start with the basics of what FT-8 actually is. FT-8 stands for Frankie Taylor Design, 8FSK modulation. It was developed by Joe Taylor K1JT, who is a Nobel Prize-winning physicist, and Steve Franca. Joe Taylor had previously developed a series of weak signal modes, JT65, JT9, WSPR, that used sophisticated forward error correction and narrow bandwidth to make contacts possible at signal levels far below what voice or traditional digital modes require. FT-8 extended that approach with a faster exchange cycle designed to make high-volume contact logging practical. Here is how FT-8 works in practice. Every FT-8 transmission is exactly 12.6 seconds long. Transmissions are synchronized to a strict 15-second cycle using GPS or Internet time. At the top of the minute, some stations transmit while others listen. At 15 seconds, the rolls reverse. At 30 seconds, they switch again. At 45 seconds, again. This rigid timing means that every station in the world running FT-8 is transmitting and receiving in synchronization. When you decode FT8 traffic, you are not just hearing one conversation, you are seeing an entire coordinated network of simultaneous exchanges. Each FT8 transmission encodes a small amount of information, typically a call sign, a grid square locator, and a signal report. The exchange to complete a contact requires a minimum of four transmissions and takes about 90 seconds from first call to logged contact. The encoding uses forward error correction, aggressive enough to decode signals that are 20 or 25 decibels below the noise floor, meaning signals you absolutely cannot hear by ear, signals that look like nothing on the waterfall, can be decoded by the software reliably. That last point is what changed everything. FT-8 can make contacts with signals so weak that no human operator could detect them by listening. A 5-watt transmitter with a modest antenna can work stations on the other side of the world under conditions where a 100-watt SSB station would hear nothing. The mode lowered the barrier to HFDX, long-distance contact, by an enormous amount. The trade-off is that FT-8 is not a conversation. The exchange is fixed. You cannot say anything that is not in the protocol. There is no rag chewing, no technical discussion, no net check-in. It is purely a contact log entry. I heard you, you heard me, here is the signal report. Contact confirmed. Some operators love this efficiency. Others find it hollow compared to a voice contact where you actually talk to another person. Both views are reasonable, and most active operators end up using both, depending on what they want, from a given session. Now let me describe what FT-8 looks like on the waterfall, because once you recognize the signature, you will never mistake it for anything else. FT-8 signals occupy approximately 50 Hz of bandwidth each. They appear on the waterfall as very thin, precisely defined vertical lines that are slightly taller than they are wide. But what makes FT-8 visually distinctive is the timing. Because every transmission is exactly 12.6 seconds long and synchronized to the same 15-second cycle, every FT-8 signal on the band starts and stops at exactly the same moment. On the waterfall, this looks like a comb, a row of thin vertical dashes, all the same height, all appearing simultaneously and all disappearing simultaneously, then a brief gap, then another row. The gap between transmissions is when the other half of the stations are transmitting and the first half are receiving. On a busy band like 20 meters during good propagation, the FT-8 segment around 14.074 MHz can contain dozens or hundreds of simultaneous signals, all blinking in unison. At normal waterfall zoom, it looks like a dense forest of identical vertical marks. Each one is a separate station somewhere in the world, and each one carries a call sign and grid square that decoding software can extract. The FT8 segments on each band are informal conventions, not regulations, but they are followed consistently enough that you can rely on them. On 20 meters, the primary FT8 frequency is 14.074 MHz. On 40 meters, it is 7.0074 MHz. On 80 meters, 3.573 MHz. On 15 meters, 21.074 MHz. On 10 meters, 28.074 MHz. There is a DX-only segment on 20 meters at 14.090 MHz. The pattern is consistent. On most bands, the FT8 frequency ends in 0.074 MHz above the band base. Now let me talk about WSPR, which is a different animal with a different purpose. WSPR, weak signal propagation reporter, pronounced Whisper, is a beacon mode. Unlike FT8, which is designed for making contacts, WSPR is designed purely for propagation research. WSPR stations transmit a low-power beacon on a schedule, typically for two minutes out of every 10, encoding call sign, grid square, and power level. Other stations around the world decode those beacons and report them to the WSPRNet database, along with the signal strength and time. The result is a real-time global map of what propagation paths exist at any given moment. On the waterfall, WSPR looks similar to FT-8, but on a slower cycle. WSPR transmissions are two minutes long and the tones are narrower. The total signal bandwidth is about 6 Hz per tone compared to FT8's roughly 6.25 Hz per tone, but the WSPR signal structure is different enough to be recognizable once you know what you are looking for. WSPR activity is concentrated on specific frequencies, 14.0956 MHz on 20 m, 7.0386 MHz on 40 m, 10.1387 MHz on 30 M. And the signals are often very faint, requiring the software decoder to pull them out of what looks like noise. The reason to care about WSPR, even as a new operator, is WSPRNet and its companion site, peacecreporter.info. These sites aggregate decoded signals from receivers around the world and display them on a map. You can look up your own call sign on PSKR Reporter and see which stations have heard you, or in receive-only mode, you can see which signals you are decoding and where they are coming from. It turns your station into a node in a global propagation sensing network, which brings me to the practical part of this episode. Setting up WSJTX alongside EtherSDR for decode-only reception. WSJTX is the free software from Joe Taylor's team that decodes FT8, WSPR, and a range of other digital modes. It is available for Linux, Mac OS, and Windows. The basic concept of the integration is this. EtherSDR receives the signal and produces audio. WSJTX takes that audio and decodes it. The two programs communicate through an audio pipe, a virtual audio cable on Windows and Mac OS, or a loopback device on Linux. On Linux, the audio routing uses a loopback interface. The most common approach is to use pulse audio or pipewire to create a loopback source, set EtherSDR to output to that loopback, and set WSJTX to take its input from the same loopback. There are specific instructions for this in the WSJTX documentation and in the Ether SDR GitHub discussions. The setup has been worked out by the community and is well documented. On Windows, the standard approach is a virtual audio cable application. VB Audio Virtual Cable is the most commonly used free option. You install it, which creates a virtual audio device. You set Ether SDR to output to that virtual device and WSJTX to receive from it. No physical cable required. On Mac OS, the equivalent is Black Hole, which is a free virtual audio driver. Same principle. Install it, route EtherSDR output to the black hole device. Set WSJTX input to black hole. Once the audio routing is in place, the WSJTX setup for receive-only operation is straightforward. Open WSJTX, go to File Then Settings, select the Audio tab, and set the input device to your virtual audio cable or loopback source. On the radio tab, you can optionally connect WSJTX to either SDR via cat control for automatic frequency tracking. But for decode-only listening, this is not required. You can simply tune Ether SDR to the FT8 frequency of your chosen band, 14 DAO 074 for 20 meters, and WSJTX will start decoding automatically at the top of the next 15-second cycle. When decodes start appearing in the WSJTX window, you will see rows of call signs, grid squares, and signal reports. Each row is one decoded station. The signal reports are in decibels relative to noise. A report of minus 10 decibels means the signal is 10 decibels below the noise floor and completely inaudible to human hearing. A report of plus 5 means it is a moderately strong signal. Reports of minus 20 or below are impressively weak signals being pulled out of the noise by the algorithm. If you enable reporting to PSK Reporter in WSJTX settings, which requires only entering your call sign and enabling the option, your decoded signals will be uploaded automatically. Within minutes of starting to decode, you can open PSKReporter.info. Enter your call sign and see a map showing every station you have heard and every station that has heard you if you are transmitting. For receive-only operation, it shows you your receive coverage and a good antenna on 20 meters during good propagation can produce a map that covers most of the hemisphere within a single session. This is one of the most satisfying early experiences in HF radio, and it requires no license to transmit, no microphone, no transmit setup of any kind. Just a radio, an antenna, ether SDR, WSJTX, and an internet connection to report to PSK Reporter. One note on time synchronization. FT8 is unforgiving about timing. If your computer clock is off by more than a second or two, D codes will fail. On Linux, the NTP service handles this automatically in most configurations. On Windows, make sure the system clock is synchronized. The built-in time sync is usually sufficient, but if decodes are failing unexpectedly, checking clock accuracy is the first thing to investigate. There are free tools to verify your clock offset against GPS time if you want to be thorough. Next episode, we start phase two of the series, transmitting. Before you key up for the first time, there are a few things to check and configure. Episode 7 is the pre transmit checklist microphone setup, audio levels, PTT options, and power settings. This is below the noise floor.