Fix KiwiSDR CW BFO offset so a tuned carrier produces an audible tone (#4423)

[wa2n-code]

Summary

• Fixes CW tuned frequency offset with Flex antenna does not match KiwiSDR offset for audible tone aethersdr/AetherSDR#4423: on a KiwiSDR virtual antenna in CW mode, a carrier centered under the tuned-frequency pip produced no audible tone (or, after the first commit below, a DC-thump/heterodyne instead of a clean tone).
• KiwiSDR's SET ... freq= is the BFO/mixdown point, not a dial frequency — low_cut/high_cut are an audio passband applied relative to it. Sending freq=carrier puts the carrier at 0 Hz, which either falls outside the passband (no tone) or lands as an inaudible DC thump (tone, but not a tone).

Why two commits

This shipped in two steps because the first fix's assumption was disproved by a real session log attached to the issue, and I want that story visible rather than squashed away:

1. Offset KiwiSDR CW sound-stream BFO by the passband center — an automated triage comment on the issue had already root-caused the "BFO stays at the carrier" problem and proposed shifting freq by the CW passband's center (assuming Flex reports the passband already offset toward the pitch, e.g. 400..800). I implemented that. It compiled and looked reasonable, but testing on real hardware showed no change in behavior — still no tone, and audible "DC shifting" while tuning.

2. Use the radio's actual CW pitch for KiwiSDR BFO offset, not passband center — pulling the log files attached to the issue showed the actual commands AetherSDR was sending: SET mod=cw low_cut=-400 high_cut=400 freq=.... Flex's real CW passband is symmetric about the carrier, not offset — the sidetone pitch is a DSP shift Flex applies after that filter, not something baked into the reported passband. With a symmetric passband, (low+high)/2 is always 0, so commit 1's shift was a no-op — the carrier kept landing at 0 Hz, which is exactly the "DC thump" that was reported back.

   The fix: use the radio's actual configured CW pitch (TransmitModel::cwPitch(), the same value driving the Flex sidetone/monitor tone) instead of deriving anything from the passband. Move the Kiwi BFO down by the pitch (carrier demodulates to +pitch Hz) and slide low_cut/high_cut up by the same amount (so that +pitch Hz frequency is still inside the passband instead of getting filtered out). This reproduces, in software, the shift the Flex's own onboard DSP does natively — which is also why Flex antennas never needed this fix in the first place.

   Also plumbed cwPitch from MainWindow → KiwiSdrManager → KiwiSdrClient::setTrackedSlice(), and wired a live re-send when the operator changes CW pitch mid-session (MainWindow_Wiring.cpp), so an already-active KiwiSDR CW session doesn't go stale if the pitch is adjusted after switching to the Kiwi.

Test plan

• Added/updated formatSoundTuneCommand unit coverage in tests/kiwi_sdr_protocol_test.cpp (CW with real symmetric passband + pitch, CW with pitch disabled, USB/LSB unaffected).
• Verified against the real SET mod=cw low_cut=-400 high_cut=400 freq=... commands captured in the issue's attached session log.
• Manually verified on real hardware (Flex 6600 + KiwiSDR): tuned-frequency pip stays centered on the signal, CW carrier now produces an audible tone matching the configured CW Pitch.

🤖 Generated with Claude Code

[wa2n-code]

Closing — reopening against the upstream aethersdr/AetherSDR repo instead of my own fork.