Digital modes and weak signal setup: where decode thresholds, time sync and audio levels actually decide your results
- Replies 2
- Views 27
- Created
- Last Reply
Top Posters In This Topic
-
Mod_Mike 1 post
-
Mod_Chrissy 1 post
Most problems with FT8, FT4, JS8Call and the other digital modes come from the plumbing, not the mode: audio levels, clock accuracy, CAT/PTT behavior and filter settings. This thread is for comparing setups and sorting out which mode fits which job. Corrections are welcome.
Start with decode thresholds, because they are often misquoted. The WSJT-X documentation gives sensitivity as S/N in a 2500 Hz reference bandwidth, not the bandwidth the signal occupies. FT8 decodes down to roughly -21 dB on that scale, FT4 is a few dB less sensitive (around -17 dB), and WSPR goes down near -31 dB. A 2-minute WSPR cycle against a 15-second FT8 cycle shows the basic tradeoff: more sensitivity costs more time on air. Q65 is the flexible one. It has several T/R period lengths, from 15 seconds up to 5 minutes, and tone-spacing submodes tuned for ionospheric scatter, tropo and EME. MSK144 is built for meteor scatter, where the useful signal is a burst of a few hundred milliseconds. That is why it uses fast 15-second sequences and very short frames instead of long integration.
The mode families split roughly like this. FT8, FT4, WSPR, Q65 and MSK144 are structured modes with fixed message formats, forward error correction and strict timing. They are efficient, but they are not conversational. JS8Call keeps the FT8-style modulation and adds a keyboard-to-keyboard layer with selectable speeds, directed messages and relaying. It is slower than a keyboard mode like PSK31, but it can get a message through when PSK31 would not. PSK31 (31.25 baud, narrow) and RTTY (usually 45.45 baud, 170 Hz shift) are older, free-flowing modes. Olivia and other MFSK modes in fldigi trade speed for robustness, and Hellschreiber works almost like a fax, since you read the character shapes yourself. Winlink is a different category: it is message transport, not a QSO mode. Depending on the path, it runs over VARA, ARDOP, Pactor or packet, and VARA HF/FM are proprietary Windows-based modems.
Time sync deserves more attention than it gets. FT8 and FT4 assume your clock lines up with everyone else's to within about a second, and the closer to zero the better. If your decodes show consistently large DT values, or you decode people who never seem to answer you, check the clock before the antenna. A PC that only syncs occasionally can drift. Use a proper NTP client, or GPS-disciplined time if you work portable without internet, and watch the DT column as a free diagnostic. Lots of stations at DT +0.5 or higher, with yours different from the rest, points at your clock.
Audio levels are the other classic problem. Digital modes need a clean, linear transmit chain. Turn off the speech processor and EQ, set the rig to its data mode (USB-D, DATA-USB or whatever your radio calls it), and keep the Windows or Linux mixer from adding enhancements or system sounds. Drive the audio just far enough to reach the power you want, and keep the ALC reading at zero or barely moving. Running a 100 W rig at full power on FT8 is rarely necessary, and an overdriven signal splatters the neighboring audio slots and is just bad on-air behavior. On receive, set the input level so the WSJT-X meter sits around the middle of its range rather than pinned or near empty, which leaves headroom for strong signals and for noise to decode against.
CAT control adds its own choices. You can use CAT for frequency and mode and a separate method for PTT, or let CAT do both. Hamlib, the manufacturer's own protocol and a virtual COM port all work, but each adds a failure point. For split operation on FT8, WSJT-X offers 'Rig' and 'Fake It'. Both aim to keep the transmitted audio within the radio's cleanest passband, since audio harmonics get worse toward the edges. If your radio reads back frequency slowly or drops serial traffic during transmit, simplify it: lower the baud rate, check for USB RF pickup, and use ferrites before you suspect the software.
Grid squares matter more than they look. Maidenhead locators use field and square pairs (like FN31) and optionally subsquares (FN31pr). A 4-character grid is enough for most FT8 work and for award tracking, but 6 characters matter for VHF and microwave weak signal work, where distance and bearing precision count. Check that your grid is entered correctly in WSJT-X. A typo means wrong distance data on PSKReporter and wasted effort on tropo or meteor scatter schedules.
Questions for the thread: Which interface do you trust most, such as a built-in USB audio codec, a separate isolated sound card interface, or an external box? What do you use for time sync when you're away from home? And if you only had two modes for HF, which would you pick, and why?
Link to comment
https://www.hamradiobase.com/forums/topic/11293-digital-modes-and-weak-signal-setup-where-decode-thresholds-time-sync-and-audio-levels-actually-decide-your-results/Share on other sites