Skip to content

20261002 - Tell the owner what to do when the radio stops sending data - #107

Merged
Purple10101 merged 1 commit into
mainfrom
20261002-no-signal-safe-settings
Oct 2, 2026
Merged

Purple10101 merged 1 commit into
mainfrom
20261002-no-signal-safe-settings

Conversation

@Purple10101

Copy link
Copy Markdown
Collaborator

What

When blah2 stops delivering samples, the Signal peak meter on the Configuration page turns red and reads "no signal", and until now that was all an owner saw. Once "no signal" has lasted a minute, a yellow box under the meter now says what to do:

  1. Check the USB cable between the SDR and the Pi is plugged in.
  2. Restart the radar on its safest settings: gain reduction 59 dB on both tuners and LNA state 9.
  3. Once the meter shows a signal again, run Quick Calibrate to find better settings.

If there is still no signal, contact support. A Use safe settings button sets the three fields and submits the form through the normal save and apply path. If the page already has other unsaved changes, it asks first, because they are submitted with it.

Why

  • "No signal" cannot say why the radio stopped. An unplugged cable, a wedged SDRplay service and a device that crashed on overload all look the same once blah2 delivers nothing, so the box does not call the receiver overloaded.
  • A plain restart is not offered. It comes back on the same config, so a tuning that overloads the radio at startup takes it down again. The watchdog also already restarts the stack every 5 minutes while blah2 crash-loops, so a restart has usually been tried by the time the box shows. Restarting on 59/59/9 removes overload as a cause too. The LNA state matters most: a node near a strong tower can overload at 59/59 with a sensitive LNA state.
  • 59/59/9 is almost deaf, so the next step is Quick Calibrate.
  • It waits 60 s and stays hidden while a calibration or an apply runs, because both restart the radar and show "no signal" for 30-60 s on their own. /calibrate/status and /config/apply/status are only fetched once the minute has passed.

This follows #104, which removed the confusing preflight note and made Auto-Calibrate save its own result.

Docs

docs/features/config-editor.md has a new No-signal help section under Signal peak meter, with the reasoning above. The page script points to it.

Verification

  • Full suite: 1,021 passed. ruff is clean, and the page scripts parse.
  • New tests check the box ships hidden, and that the button's values match calibrator.GAIN_REDUCTION_MAX and LNA_STATE_MAX and real form field names.
  • In a real browser (headless Chromium) on a test node with no SDR attached, where the meter reads "no signal" permanently. The node was restored afterwards.
    • The box was hidden at 30 s, while the meter already read "no signal", and appeared at 60 s.
    • Use safe settings saved and reloaded, and the normal apply ran. The box stayed hidden while the apply ran, and the apply finished cleanly.
    • After that, the fields read 59 / 59 / 9.
  • Not verified:
    • That the button brings a dead radio back on a node with an SDR. That case needs a real wedge.
    • That the box disappears when a signal returns. The test node can never produce a signal.

🤖 Generated with Claude Code

The Signal peak meter turns red and reads "no signal" when blah2 stops
delivering samples, and until now that was all an owner saw. Once "no
signal" has lasted a minute, the Capture section now shows what to do:
check the USB cable, restart the radar on its safest settings, then run
Quick Calibrate once a signal is back. A Use safe settings button sets both
gain reductions to 59 and the LNA state to 9 and submits the form through
the normal save and apply path, asking first if the page has other unsaved
changes.

The advice does not call the receiver overloaded, because "no signal"
cannot tell an unplugged cable from a wedged SDRplay service or a device
that crashed on overload. A plain restart is not offered either: it comes
back on the same config, so a tuning that overloads the radio at startup
takes it down again, and the watchdog already restarts the stack while
blah2 crash-loops. Restarting on 59/59/9 also removes overload as a cause.
The box waits a minute and stays hidden while a calibration or an apply
runs, since both show "no signal" for 30-60 s on their own.

Tests check the box ships hidden and that the button's values match the
calibrator's limits and real form fields. The reasoning is in
docs/features/config-editor.md under No-signal help.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
@Purple10101
Purple10101 merged commit e430134 into main Oct 2, 2026
3 checks passed
@Purple10101
Purple10101 deleted the 20261002-no-signal-safe-settings branch October 2, 2026 12:10
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant