Getting a clean digital-mode chain: sound card levels, CAT, time sync and what decode thresholds actually mean
- Replies 2
- Views 28
- Created
- Last Reply
Top Posters In This Topic
-
Mod_Chad 1 post
-
Mod_Chrissy 1 post
Starting a thread to collect the practical details that cause most digital-mode problems. The modes are different, but the failure points are mostly the same: audio levels, rig control, clock accuracy and a misunderstanding of what a published decode threshold promises.
The signal chain. For WSJT-X, FLDIGI and the Winlink clients, the path is radio, audio interface, sound card, software, then back out through the same interface to the transmitter. Most problems sit in one of three places. The rig is in the wrong mode (USB or a data mode, not FM or AM, with compression, EQ and noise reduction off). The operating system is processing the audio (disable enhancements, AGC and system sounds on the interface). Or the transmit drive is too high. For FT8, FT4 and other WSJT-X modes, the usual rule is to set the drive so the ALC barely moves, even if that means less than full power. Overdriven audio splatters across the band, which is much more noticeable on a crowded FT8 segment than it would be on a single PSK31 signal.
CAT versus VOX versus serial PTT. CAT control gives the software frequency readout and PTT through the radio's USB or serial port, and it enables WSJT-X's Split Operation options. If audio is clean but the radio sits at 1500 Hz offsets with harmonics showing up, try 'Fake It' split so the audio tones stay in a comfortable part of the passband. VOX is a last resort because it can clip the first symbols and trigger on stray audio. RTS/DTR PTT works but can cause a transmit glitch when the serial port opens, so check what the radio does when the program starts.
Time sync. FT8, FT4, WSPR, Q65 and MSK144 all depend on a shared clock. FT8 works in 15-second periods, FT4 in 7.5 seconds, and WSPR in 2-minute slots. The software can tolerate roughly a second of error in FT8, but the tolerance is narrower the shorter the period. Watch the DT column in the decodes. If most stations show a consistent offset, your computer clock is probably the one that is off. Use NTP or a GPS-based source and confirm the sync service is actually running, since a system that was right yesterday can drift after sleep or a lapsed sync service. JS8Call has extra timing slack because of how its submodes are laid out, but it is not immune.
Decode thresholds. The usual figures are quoted as signal-to-noise ratio in a 2500 Hz reference bandwidth, not in the bandwidth of the signal itself. Roughly, FT8 decodes near -20 dB, FT4 a few dB higher (around -17), and WSPR much lower, near -30 dB, because its transmissions are long and slow. Q65 is designed for weak-signal paths such as EME and scatter, and its performance depends on the chosen submode and period length. MSK144 is designed for meteor-scatter pings, so it trades long integration for speed. These numbers are sensitivity under ideal conditions with a good clock and a clean receiver. They do not account for QRM, local noise, fading or a poorly set AGC. A 'report' of -18 on FT8 is a measurement by the decoder, not a promise that the QSO will complete.
Comparing the older and keyboard modes. PSK31 is narrow, at about 31 Hz for BPSK31, and works well for casual keyboard QSOs, but it is not a weak-signal mode in the way FT8 is. RTTY (commonly 45.45 baud, 170 Hz shift) is the contest standard, and contest software leans heavily on it. Olivia and other MFSK-family modes (Olivia, MFSK16 and similar) are built to work through noise and fading with forward error correction, at the cost of slow typing speed. Hellschreiber displays the signal visually instead of decoding it, which makes it a different kind of experiment. JS8Call is derived from FT8 but is aimed at keyboard-style messaging, with selectable speeds, heartbeat and relay features.
Winlink and VARA. Winlink is an email-over-radio system. It can use HF or VHF/UHF links and several transport modes, including VARA, ARDOP, Pactor and packet. VARA is a proprietary modem with its own licence terms, so check the current options before you assume it is free to run at full speed. The practical issue is the same as everywhere else: the TNC audio levels and the PTT method. If a Winlink session connects but fails on larger transfers, look at audio drive and the compression the radio applies before blaming the gateway.
Grid squares. Maidenhead locators use letters and digits, such as FN31 (four characters) or FN31pr (six). WSJT-X exchanges four-character grids in most standard messages. A mistyped grid ruins distance and award tracking, so verify yours against a map or a locator calculator, and check the rig or logger is not using an old value.
Which of these do you find harder to pin down: audio levels, CAT setup or the clock? And does anyone have a checklist for confirming a chain is clean before calling CQ?
Link to comment
https://www.hamradiobase.com/forums/topic/11218-getting-a-clean-digital-mode-chain-sound-card-levels-cat-time-sync-and-what-decode-thresholds-actually-mean/Share on other sites