ALC298 Samsung Galaxy Book3 Ultra (144d:c1cc): left speaker crackles with direct ALSA playback; Windows audio is clean

[pedrorach16]

Summary

The left internal speaker channel on a Samsung Galaxy Book3 Ultra consistently crackles under Linux. The right channel works correctly.

The problem reproduces with direct ALSA hardware playback while PipeWire and WirePlumber are completely stopped. Both speakers work perfectly under Windows on the same laptop.

Hardware

Component	Detail
Laptop	Samsung Galaxy Book3 Ultra NP960XFH-XA2BR
CPU	Intel Core i9-13900H
Audio controller	Intel Raptor Lake-P/U/H cAVS 8086:51ca
Codec	Realtek ALC298 10ec:0298
Codec/controller subsystem	Samsung 144d:c1cc
Kernel driver	sof-audio-pci-intel-tgl
Distribution	openSUSE Tumbleweed 20260724
Kernel	7.1.4-1-default
PipeWire	1.6.8
WirePlumber	0.5.15
ALSA	1.2.16.1
SOF firmware	2025.12.2

The subsystem ID is already matched by:

SND_PCI_QUIRK(0x144d, 0xc1cc,
              "Samsung Galaxy Book3 Ultra (NT960XFH)",
              ALC298_FIXUP_SAMSUNG_AMP_V2_4_AMPS)

Symptoms

• The left speaker channel crackles during the spoken channel test.
• On openSUSE Tumbleweed, crackling is audible on both words of "Front Left."
• On a Fedora KDE live USB, the same channel crackles on "Front" but not "Left."
• The right channel works normally.
• Both internal speaker channels work perfectly in Windows.
• The problem has existed across multiple Linux versions and configurations.
• Testing from a Fedora live USB rules out accumulated openSUSE configuration or user-session state.

PipeWire-independent reproduction

PipeWire, PipeWire Pulse and WirePlumber were stopped before testing:

systemctl --user stop \
    wireplumber.service \
    pipewire-pulse.service \
    pipewire.service \
    pipewire-pulse.socket \
    pipewire.socket

The standard ALSA front-left sample was then played directly through the normal hardware PCM:

speaker-test \
    -D plughw:0,0 \
    -c 2 \
    -r 48000 \
    -F S16_LE \
    -t wav \
    -w /usr/share/sounds/alsa/Front_Left.wav \
    -l 1 \
    -s 1

The command completed successfully and the left channel crackled. The hardware Speaker mixer channels were equal and unmuted. The test was repeated at an audible hardware level.

This rules out PipeWire, WirePlumber and the desktop channel-test implementation as the root cause.

Expected result

The left speaker should reproduce the sample cleanly, as it does under Windows and as the Linux right channel does.

Suspected area

The shared digital audio path appears functional because the right channel is clean. The problem seems specific to the Samsung/Realtek per-amplifier programming used by ALC298_FIXUP_SAMSUNG_AMP_V2_4_AMPS.

Possible areas to investigate include:

• initialization coefficients for left-side amplifier IDs;
• amplifier enable/disable ordering;
• missing stabilization or volume-ramp delay;
• differences between the simplified Linux sequence and the Windows vendor sequence;
• amplifier state after HDA codec power transitions.

I am not claiming a particular register-level fix; assistance capturing and comparing the relevant amplifier state would be appreciated.

Related upstream work

• Samsung ALC298 tracking issue:
  [BUG] Samsung Galaxy Book2 Pro 360 no sound through speaker #4055
• Previous report for this exact 144d:c1cc subsystem:
  ALC298 (Samsung Galaxy Book3 Ultra NP960XFH): speakers silent after s2idle resume — SOF DSP audio path broken #5730
• Samsung amplifier fixup refactor:
  torvalds@7e4d4b3

Please let me know which additional codec dumps, dynamic-debug logs or controlled playback tests would help isolate the incorrect left-amplifier state.