FT8 Can Hear Them — But Can You Work Them on SSB or CW?

You decode a distant station at −15 dB on FT8. The temptation is immediate: spin the dial, call on CW or SSB and expect the same path to work. That is useful optimism, but poor link engineering. The FT8 number has a precise convention, and changing mode changes far more than bandwidth.

What the WSJT-X Number Actually Means

The current WSJT-X User Guide defines displayed signal reports as signal-to-noise ratio in decibels using a standard reference noise bandwidth of 2,500 Hz. A report of −15 dB therefore means that the estimated signal power is 15 dB below the estimated noise power in that 2,500 Hz reference bandwidth.

That convention allows reports to be compared without pretending that the receiver’s audio filter is literally 2,500 Hz wide. It also means the displayed number must not be described as “the SNR inside FT8’s 50 Hz occupied bandwidth.” Those are different quantities.

The same official guide specifies FT8 as 8-GFSK at 6.25 baud, with 50 Hz occupied bandwidth and a 12.6-second transmission inside a 15-second transmit/receive sequence. Its protocol table gives an approximately −21 dB S/N threshold for at least 50% decode probability under the stated test convention. That is a decoder-performance reference, not a boundary between audible and inaudible speech or Morse.

Bandwidth Arithmetic Is Not Mode Equivalence

For white noise and otherwise identical conditions, changing effective noise bandwidth from B1 to B2 changes noise power by:

That 17 dB is a bandwidth normalization, not “the FT8 advantage over CW” and not an SSB prediction. FT8 does more than occupy 50 Hz: it synchronizes a structured waveform, integrates information over 12.6 seconds and applies forward-error-control coding with a purpose-built decoder. Conversely, occupied bandwidth is not necessarily the receiver’s equivalent noise bandwidth, and neither number alone describes how information is recovered.

CW copy depends on keying speed and shape, the receiver filter and ringing, tone pitch, QSB, interference, operator training and how much repetition is acceptable. SSB intelligibility depends on speech spectrum, articulation, processing, peak and average power, receiver bandwidth, AGC behaviour, noise type and the content that must be understood. No single offset from an FT8 report can absorb all of those variables.

A Decoder and an Operator Perform Different Tasks

FT8 also exchanges a constrained set of callsigns, locators, reports and acknowledgements. A decoder can exploit that structure. Human CW and SSB may carry less predictable content, and “I recognized the callsign” is a different criterion from “we completed the exchange without guessing.” Define the required information before declaring one mode equivalent to another.

Hearing Them Is Only Half the Contact

Your received FT8 report describes one direction of one link during one short interval. It does not tell you the other station’s noise floor, receiver bandwidth, antenna pattern, local interference, power, operating preference or willingness to change mode.

Propagation and passive antennas may be reciprocal under the same linear conditions, but a two-way station comparison is not automatically symmetrical. Transmit powers, feeder loss, polarization, local noise and receiver performance can differ at the two ends. FT8’s constant-envelope transmission and high duty factor also must not be compared with CW key-down power or SSB peak-envelope power merely because the front-panel wattage is the same.

So the practical statement remains sharp: decoding the DX is proof that an FT8 path existed toward your receiver. It is not proof that your CW or SSB signal will be copied in the reverse direction.

Why a Fixed dB Chart Fails in the Field

The displayed report is an estimate, not a laboratory power standard. Official WSJT-X performance notes explain that the estimate averages signal and baseline-noise power over the transmission, scales the result to 2,500 Hz and can carry an error of several decibels near the weakest decodable levels. A single −18 dB and the next −14 dB may reflect estimation scatter, QSB or interference rather than a clean 4 dB path improvement.

Decode success also changes with fading through the 12.6-second waveform, time and frequency error, overlapping signals, impulsive noise and decoder settings. AGC pumping or receiver overload from another signal can alter the result without changing the wanted path. CW and SSB then sample that changing channel with different time scales, bandwidths and decision processes.

This is why the useful question is not “What FT8 number guarantees SSB?” It is “How stable is the link margin, and what happens when I measure the intended mode under controlled settings?”

A Better On-Air Test

1. Record a run, not one decode. Log several minutes of reports, decode failures, QSB, interference and time-of-day conditions.
2. Freeze the receiver state. Record preamp, attenuation, RF gain, AGC, filter bandwidth, noise reduction and audio processing. Do not compare modes while those controls move unnoticed.
3. Measure the intended bandwidth. Use the receiver’s declared or measured equivalent noise bandwidth where available; do not substitute occupied bandwidth automatically.
4. Coordinate the far end. If possible, ask the same station to send FT8, steady carrier or keyed test material, CW and SSB close together in time. Record power at a stated plane and distinguish average from peak-envelope power.
5. Use an A/B/A sequence. Return to the first mode after the comparison. If the result does not return, propagation or receiver state changed during the test.
6. Define success. A callsign recognized once, solid conversational speech, contest exchange and error-free message are not the same requirement.

Without a coordinated station, use the FT8 report only as a scouting clue. A strong, stable series of reports usually offers more margin than a single marginal decode, but there is no honest universal number at which CW or SSB “starts to work.” Try the mode when the operator and band plan permit it; record what happened instead of treating a chart as propagation.

The Operator’s Bottom Line

A waterfall full of FT8 proves that the band supports FT8 traffic at that moment. It does not prove that the same stations are available, equipped or strong enough for your chosen CW or SSB exchange.

Read the report in its own units: S/N referenced to 2,500 Hz, derived from a 12.6-second FT8 transmission. Then account for receiver bandwidth, information content, integration time, fading, interference, AGC, operator skill and the reverse link. If you want to know whether you can work them on another mode, the best instrument remains a controlled attempt with both ends identified.

Primary and authoritative technical sources

• WSJT-X User Guide—current S/N reporting convention, FT8 timing, occupied bandwidth, protocol parameters and decoder behaviour.
• WSJT-X official downloads—current general-availability release and documentation entry point.
• Franke, Somerville and Taylor: “The FT4 and FT8 Communication Protocols”—message structure, coding, modulation, timing and channel-performance boundaries.
• WSJT-X: SuperFox and FT8 Weak-Signal Performance—FT8 S/N estimation method and uncertainty near the decode limit.
• ITU-R SM.328-12—current distinction between occupied, necessary and receiver bandwidth.

Questions, antenna-factor records or height trials to share? Contact RF.Guru.