Skip to main content

FT8 vs JS8Call vs PSK31: How Weak-Signal HF Digital Modes Reach Long Distances

How narrow signals, structured decoding, and clean station setup extend practical HF reach

FT8 vs JS8Call vs PSK31: How Weak-Signal HF Digital Modes Reach Long Distances
Topic Technology
Updated
Author Daniel Odoh
Read Time 14 min

A modest HF station can complete contacts that sound implausible on voice, but the software is not manufacturing a path. It is making better use of the signal that reaches the receiver.

The useful question is therefore not which digital mode is “strongest.” It is what kind of exchange you need, how much timing discipline you can support, and what your radio can transmit cleanly. Those choices separate FT8, JS8Call, and PSK31 more clearly than a sensitivity number ever will.

Quick Take

  • Why they work: All three concentrate information into narrow signals. FT8 and JS8Call also use fixed timing and coding, while PSK31 gains efficiency from a very narrow live-typing signal. Raw power is only one part of the link.
  • FT8: 15-second structured exchanges built for grid and DX hunting. Start here if you want contacts fast with a modest antenna.
  • JS8Call: Keyboard messaging and store-and-forward relay on a related weak-signal transport. Choose it when you want to type real sentences or pass messages.
  • PSK31: Live keyboard-to-keyboard ragchewing in roughly 60 Hz. Choose it for conversation, not for automated weak-signal work.
  • Two setup failures beat any band opening: An overdriven transmit chain and a drifting computer clock prevent successful decodes no matter what the ionosphere is doing.
  • No guarantees: Low power and a good decoder do not promise a worldwide contact. Frequency, antenna, noise, solar conditions, and the other station still decide whether the path exists.

Why Weak Signals Can Still Decode

The shared mechanism is narrow bandwidth: less noise enters the receiver’s passband, so a modest signal is easier to separate from it. FT8 and JS8Call add fixed transmission windows, known symbol patterns, and coding that help software recover structured data. PSK31 takes a different route, using a continuous, very narrow signal for live typing.

ARRL’s explainer on weak-signal modes describes the underlying idea through coherent CW: stable, precisely timed transmitters let receivers apply extremely narrowband filtering that raises the signal-to-noise ratio to a usable point, even at very slow sending rates. The same principle helps explain why narrow digital signals can remain usable when voice cannot.

Keep every signal-to-noise figure tied to the bandwidth that defines it. A number quoted without its reference bandwidth cannot be compared to another mode’s number.

Documented bandwidth and sensitivity figures, each with its stated reference basis.
ModeOccupied bandwidthSensitivity figure as documentedReference basis stated?
PSK31Approximately 60 Hz at −26 dBNarrow bandwidth reduces susceptibility to noiseYes — but −26 dB is an occupied-bandwidth limit, not a receiver noise reference
JS8 40160 Hz−18 dBPartially — speed-table value, no receiver noise bandwidth stated
JS8 60250 Hz−16 dBPartially — speed-table value, no receiver noise bandwidth stated
WSPRTwo-minute sequencesDecodes to −31 dBYes — in a 2500 Hz bandwidth
Sources: ARRL PSK31 specification, JS8Call User Guide, and WSJT-X documentation.

Two different kinds of number appear in that table, and mixing them is an easy way to misjudge a mode. The PSK31 specification uses −26 dB to mark the edge of its occupied-bandwidth envelope; it is not a receiver noise baseline. The JS8Call speed table publishes tone sensitivity per speed without naming a receiver noise bandwidth. WSPR’s documentation states its −31 dB decoding figure in a 2500 Hz reference bandwidth. Those measurements describe different things, so they should not be ranked as though they came from one test.

Comparison Criteria That Actually Matter

Six characteristics separate these modes in ways that change how you operate. Judge each mode on all six before you commit your station time to it.

  • Transmission timing. Fixed-interval protocols require your clock and the other station’s clock to align. Free-running modes do not.
  • Occupied bandwidth. Narrower signals concentrate your power and admit less noise, but they also carry less information per second.
  • Message type. Structured exchanges (callsign, grid, report) versus free text versus live typing change what you can actually say.
  • Error correction. Coding recovers corrupted data, but it can also cost sensitivity when the coding scheme is inefficient for the path.
  • Clock tolerance. How far your computer’s clock can drift before decoding fails.
  • Typical use case. The operating goal the protocol was designed to serve.

The Three Modes Side by Side

FT8, JS8Call, and PSK31 on shared criteria.
CriterionFT8JS8CallPSK31
Timing structure15-second transmit/receive sequences12.6 s frames in a 15 s window at normal speed; faster and slower speeds availableContinuous; no fixed intervals
Message typeStructured exchange (callsign, grid, signal report)Free-text keyboard messaging plus directed messages and relayLive keyboard-to-keyboard conversation
Error handlingStructured coding with repeated framesLDPC coding with soft combining of repeat framesBPSK without error correction; QPSK with convolutional coding and Viterbi decoding
Clock toleranceSync to UTC within ±1 sWithin ±2 s of UTCNone required
Duty cycleShort structured transmissions84% at normal speed; 79% fast; 65% JS8 40; 75% JS8 6080%
Best fitFast DX and grid huntingTyped messages, relay, emcomm experimentationConversational ragchewing
Sources: WSJT-X documentation, JS8Call User Guide, and ARRL PSK31 specification.

FT8: Structured Speed

FT8 uses 15-second transmit/receive sequences and is roughly four times faster than the one-minute slow modes in the same software family. That speed costs a few dB of sensitivity compared with those slow modes, which is a deliberate trade. The same documentation says worldwide HF QSOs are possible with a few watts or even milliwatts and compromise antennas. Possible, under the right conditions, is the operative phrase.

FT8’s practical advantage is activity. ARRL describes FT8 as an extremely popular HF mode, so a new operator is more likely to find active stations than on a quiet niche mode. That makes it a sensible first choice for learning weak-signal operating with a modest antenna.

JS8Call: Weak-Signal Transport With Typing

JS8Call is a derivative of WSJT-X, restructured for message passing using a custom FSK modulation called JS8. The project states plainly that it is not supported or endorsed by the WSJT-X development group. What it adds is a directed calling protocol layered over the base transport, which supports free-form and directed messages.

The timing is the tell. According to the JS8Call User Guide, normal speed transmits for 12.6 seconds inside a 15-second window, an 84% duty cycle. Faster speeds compress that: JS8 40 runs for 3.95 seconds in a 6-second window at 160 Hz and 40 words per minute, while JS8 60 runs for 3 seconds in a 4-second frame at 250 Hz. The guide flags JS8 60 as experimental and more demanding of the computer. JS8 40 and JS8 60 cannot participate in Heartbeat networking, which is the relay system that makes JS8Call interesting for message passing.

PSK31: The Conversational Classic

PSK31 runs at 31.25 baud, about 50 words per minute, in roughly 60 Hz of bandwidth at −26 dB. ARRL’s specification describes it as intended for live keyboard-to-keyboard conversations, similar to radioteletype. There is no fixed timing, so your clock does not matter.

The error-correction story is more nuanced than it first appears. BPSK carries no error correction. QPSK adds convolutional encoding and Viterbi decoding, but the specification describes a 3 dB signal-to-noise penalty. Coding can help on fading paths with bursty errors, yet it can perform worse when the path is simply noisy. There is no universal winner; BPSK is the simpler starting point, and QPSK is a path-dependent alternative.

Where FT4 and WSPR Fit

FT4 and WSPR sit outside this comparison as supporting tools. FT4 uses 7.5-second sequences and is designed with contesting in mind. WSPR is a one-way propagation beacon that carries callsign, grid, and power in dBm, feeding the WSPRnet database. Neither is a conversational mode, and WSPR is not a QSO mode.

Trade-Offs: What You Give Up for What You Get

Every one of these modes buys weak-signal performance by surrendering something else. Understanding the exchange prevents disappointment.

  • FT8 gains speed and structured decoding but gives up a few dB of sensitivity versus slower modes and cannot carry a conversation outside its fixed exchange.
  • JS8Call gains free-text messaging but gives up speed and simplicity; its long transmit periods also demand attention to the rig’s thermal limits.
  • PSK31 gains live conversation but gives up automated structured exchanges, while BPSK carries no error correction.
  • QPSK adds error correction to PSK31 but accepts a 3 dB signal-to-noise penalty that may not help on a simply noisy path.

Long transmit periods deserve attention because they affect hardware. The JS8Call guide tells operators to follow the manufacturer’s power restrictions for sustained digital transmissions and recommends conservative output when the rig’s limits are unclear. ARRL’s PSK31 guidance likewise warns that the radio still has to dissipate heat. The safe setting is the one your transceiver can sustain cleanly, not the largest number on its display.

Which Mode Fits Which Operator

Match the mode to your goal and your constraints. These rules resolve most of the decision.

  • Newly licensed, wants contacts fast, limited antenna: FT8. The band is busy and the exchange is short.
  • Contester: FT4. The 7.5-second sequences exist for exactly this.
  • Wants to type real sentences over weak signals, or experiment with relay messaging: JS8Call. Heartbeat networking is the reason to choose it over FT8.
  • Wants live ragchewing with a classic feel: PSK31. No clock sync, no fixed windows, just typing.
  • Wants to map propagation without a QSO: WSPR. Run a beacon and read the reports.

What Still Decides Whether the Path Exists

Digital processing helps you decode a marginal signal. It does not create a path where none exists. Frequency choice, antenna efficiency, local noise, solar conditions, and the other station’s willingness to be there all gate the outcome.

ARRL’s weak-signal page frames this honestly: many hams operate QRP at 5 watts or less and still get very good signals when band conditions are right. The conditional clause carries the weight. A quiet location and efficient antenna can make modest power surprisingly effective, while local noise can erase that advantage.

If you are estimating what a given setup might achieve before you invest in hardware, a practical overview of click to visit is worth reading alongside your own band observations. Treat any range estimate as a starting point for your own testing.

Keep the claim in proportion: no combination of low power, narrow bandwidth, and good decoding guarantees a worldwide contact. WSJT-X documentation says worldwide QSOs are possible with a few watts; possible is not the same as reliable. Plan your expectations around the path, not the protocol.

Station Hardware and Audio Chain

Most digital-mode failures trace back to the audio chain, not the radio. A dedicated soundcard interface keeps transmit and receive audio isolated from your computer’s noisy ground. Digirig and SignaLink are common examples; both present the radio with clean audio and give the software a predictable device to select.

CAT control lets the software change frequency and mode without you touching the dial. WSJT-X includes Hamlib for rig control and supports nearly all modern amateur radios. If your radio is very new, you can update Hamlib from within the program.

RFI on USB is the other frequent culprit. Shielded cables, a common ground point, and ferrite chokes on the USB run solve most cases. If your waterfall shows broadband hash that rises when you transmit, you have a ground loop or a radiating cable, and no amount of software adjustment will fix it.

Waterfall fills with broadband noise when I transmit

This is RF feedback into the audio chain. Add ferrite chokes to the USB and audio cables, verify a single common ground point, and shorten cable runs where possible. Recheck the waterfall with the transmitter keyed into a dummy load to confirm the noise source.

Decodes are intermittent even with a strong signal

Check your computer clock offset first, then your input audio level. The JS8Call guide recommends setting input audio to the low-to-mid range on the application’s meter and turning AGC off or to fast. A clock outside the decoding window can produce this symptom.

Other stations report my signal is wide or splattering

Your transmit audio may be overdriving the rig. Reduce the audio output, disable speech processing or enhancement, and follow the transceiver’s digital-mode guidance for ALC. ARRL’s PSK31 tips target a clean intermodulation figure of −20 dB or better.

Set Up a Clean Transmit Chain

Do this once per mode and per band before you call anyone. A clean signal decodes when a dirty one does not, regardless of propagation.

  1. Set the audio output level low, then raise it slowly.Start with the software’s transmit audio near minimum and increase it while watching the radio’s ALC meter. Stop as soon as the meter begins to move.ALC behavior varies by transceiver, so use the manufacturer’s digital-mode instructions rather than treating one meter reading as universal. Raise the software level only until the rig produces stable, clean output without showing active compression.

    With the rig keyed into a dummy load, the power output should remain stable and the ALC indication should match the manufacturer’s digital-mode guidance.

  2. Verify your intermodulation distortion.Ask a nearby station for an RSQ report, or use a local receiver to inspect your signal’s cleanliness. Send a steady tone and have the other operator report your IMD and how your waterfall trace looks.ARRL’s PSK31 guidance targets an IMD of −20 dB or better. A splattering trace means you are still overdriving somewhere in the chain.

    Your trace should appear as a single clean line on the receiving waterfall with no sidebands.

  3. Set your receive audio and disable processing.Turn AGC off or to fast, disable any speech processor or audio enhancement, and set input audio so the application meter reads in the recommended range. The JS8Call guide suggests roughly 30–40 dB.ARRL‘s PSK31 tips add that a rig volume setting below 25% of maximum helps, and that a separate soundcard with a signal-to-noise ratio above 100 dBA outperforms most onboard audio.

    Your waterfall should show a defined noise floor with clear, separated signal traces.

  4. Check your rig’s duty-cycle rating before sustained operation.Confirm what your transceiver manufacturer recommends for sustained digital transmissions before using either mode for long sessions.Start conservatively, watch for thermal warnings or automatic power reduction, and lower output if the radio cannot hold a stable level. The JS8Call guide and ARRL’s PSK31 tips both emphasize conservative operation.

    The rig should sustain the transmission without a thermal warning, power derating, or a worsening signal report.

Time Synchronization

Fixed-interval protocols decode only when your clock aligns with the transmission window. WSJT-X requires synchronization to UTC within ±1 second and recommends a proper NTP client such as Meinberg NTP. Its documentation warns against tools that make occasional time steps because the clock should advance smoothly.

JS8Call allows about ±2 seconds, but the same principle applies. The guide recommends an internet, NTP, or GPS time source and includes a manual drift tool for portable operation away from connectivity.

Synchronize Your Clock for Fixed-Interval Modes

Do this before your first FT8 or JS8Call session, and again after any operating system update.

  1. Install a proper NTP client.Use a full NTP implementation rather than the operating system’s periodic time correction. WSJT-X documentation names Meinberg NTP as a recommended option and warns that SNTP tools applying periodic corrections can break decoding.The requirement is a clock that advances monotonically, never stepping backward mid-window.

    Your time service should report as active and continuously running.

  2. Verify your offset against UTC.Compare your system clock to a reliable UTC reference and confirm the offset is well inside the tolerance for your mode: a sub-second margin for WSJT-X modes, and a couple of seconds for JS8Call.Both applications display UTC directly, so you can watch the offset in real time.

    The displayed UTC time should track the reference without visible drift over several minutes.

  3. Use the manual drift tool only as a fallback.When you are portable and offline, JS8Call’s manual drift tool lets you align the internal clock to signals you can see or hear. You can also offset deliberately to match a station whose clock is wrong.This is a fail-safe, not a substitute for proper synchronization at a fixed station.

    Decodes should begin appearing consistently once the drift adjustment brings you inside the window.

The Bottom Line

Choose the mode for the message, synchronize the clock when the protocol requires it, and transmit a clean signal. Get those three right and the software can show you what the path supports; neglect them and even a capable protocol may produce silence.

Verify the result

  • Your system clock offset sits inside the tolerance for your chosen mode.
  • The ALC indication matches the manufacturer’s digital-mode guidance and does not show active compression.
  • Your receive audio meter reads in the range your software recommends.
  • A local station confirms your signal is clean with no splatter.
  • Your rig stays within its rated duty cycle during sustained transmissions.
Daniel Odoh

About the Author

Daniel Odoh

A technology writer and smartphone enthusiast with over 9 years of experience. With a deep understanding of the latest advancements in mobile technology, I deliver informative and engaging content on smartphone features, trends, and optimization. My expertise extends beyond smartphones to include software, hardware, and emerging technologies like AI and IoT, making me a versatile contributor to any tech-related publication.

View all posts by Daniel Odoh →
Comments

Be the First to Comment