Audio Expansion Card disconnects and won't recover on Framework Desktop

[PatrikScully]

Device Information

System Model or SKU

• Framework Desktop (AMD Ryzen AI 300 PRO Series)

BIOS VERSION

03.04

DIY Edition information

Memory: Not applicable (integrated)
Storage: WD_BLACK SN850X 1000GB (Firmware 620361WD)

Port/Peripheral information

• Peripheral: Framework Audio Expansion Card (USB ID 32ac:0010, bcdDevice 0.02)
• Port: Expansion slot (USB path: Bus 3, Port 2.2, behind Genesys Logic hub 05e3:0610 on xhci_hcd at 0000:c4:00.0)
• Card was present at boot (not hot-plugged)

Adjacent port devices:

For Desktop only

1. Audio Expansion Card (the affected device)
2. Empty / not applicable

Standalone Operation (Laptop Only)

Not applicable (Framework Desktop)

Describe the bug

The Framework Audio Expansion Card disconnects during active audio use and enters an unrecoverable state. The card goes through a disconnect/reconnect loop with UAC2 clock negotiation failures before disappearing from USB entirely. Only a full reboot restores it. Software USB hub resets, replugging headphones, and modprobe cycling have no effect.

This is the same class of issue as #161 (Laptop 16), but occurring on the Framework Desktop, confirming it is not laptop-specific.

Also potentially related to #158 (USB-C port issues) given the PD controller read failures observed.

Steps To Reproduce

1. Boot with Audio Expansion Card seated in expansion slot
2. Card enumerates correctly, audio works via headphones
3. Use audio for several minutes (e.g., Slack meeting with headphones connected to the expansion card)
4. Audio cuts out - card disconnects and enters a reconnect loop
5. Card eventually vanishes from USB entirely (lsusb no longer shows it)
6. Only a full reboot recovers the card

Expected behavior

The Audio Expansion Card should maintain a stable connection during audio playback and not disconnect or require a reboot to recover.

Screenshots

Not applicable. See kernel logs below.

Kernel Log Timeline

08:16:55  Card connected successfully (device #4, 32ac:0010)
08:24:37  usb_set_interface failed (-110)                    <-- ETIMEDOUT, initial trigger
08:24:42  USB disconnect (device #4)
08:24:42  usb_set_interface failed (-19)                     <-- ENODEV (device already gone)
08:24:44  Card auto-reconnected (device #5)
08:24:48  USB disconnect (device #5) - immediate clock failure
08:24:51  Card auto-reconnected (device #6)
08:26:00  Final USB disconnect (device #6) - after ~1.5 min of repeated failures

Error Pattern (233 error lines total)

The repeating error loop between each reconnect attempt:

usb 3-2.2: uac_clock_source_is_valid(): cannot get clock validity for id 9
usb 3-2.2: clock source 9 is not valid, cannot use
usb 3-2.2: 1:1: cannot get freq (v2/v3): err -71
usb 3-2.2: 1:1: cannot set freq 48000 (v2/v3): err -71

• Error -110 = ETIMEDOUT (USB interface set timed out)
• Error -71 = EPROTO (UAC2 clock source negotiation protocol failure)
• Error -19 = ENODEV (device gone during cleanup)

PipeWire / WirePlumber Errors

pipewire: (alsa_output.usb-Framework_Audio_Expansion_Card-00.analog-stereo-114) suspended -> error (Start error: No such device)
wireplumber: spa.alsa: Error opening low-level control device 'hw:3': No such file or directory
wireplumber: spa.alsa: can't open control for card hw:3: No such file or directory

Recovery Attempts (all failed)

• Replugging headphones: No effect (card itself is gone from USB)
• Software USB hub reset (echo 0/1 > /sys/bus/usb/devices/3-2/authorized): Hub re-authorized but card did not re-enumerate
• Card is not visible in lsusb after failure - completely absent from USB bus

Operating System (please complete the following information):

• OS/Distribution: Ubuntu 25.10 (Questing Quokka)
• Version: 25.10
• Linux Kernel Version: Linux SPATRIK-PC 6.17.0-20-generic #20-Ubuntu SMP PREEMPT_DYNAMIC Fri Mar 13 20:07:29 UTC 2026 x86_64 GNU/Linux

Additional context

Firmware Details (framework_tool 0.6.1)

Component	Value
Board	FRANMFCP06, Revision A6
EC Firmware	dogwood-3.0.4-a4ef660 (2025-11-18) - running from RO image
PD FW	0.0.0E (Back only - Right/Left fail to read Silicon ID)
Audio stack	PipeWire 1.4.7 / WirePlumber 1.4.7
USB Hub	Genesys Logic GL3590 (FW 10.0.1) via 05e3:0610/0625

PD Controller Status (framework_tool --pd-info)

Right / Ports 01:  Failed to read Silicon ID/Family
Left / Ports 23:   Failed to read Silicon ID/Family
Back:              Silicon ID 0x30A0, FW 0.0.0E (MainFw)

All 4 USB-C ports return [ERROR] EC Response Code: InvalidCommand for port status queries.

Note: EC is running from RO (Read-Only) image, not the main RW image.

Additional Observations

• The card worked for ~8 minutes before the first failure - consistent with reports in Audio Expansion Card breaking at least on a weekly basis for over a year now. #161
• The failure pattern (ETIMEDOUT -> disconnect -> reconnect -> EPROTO clock loop -> disconnect) matches exactly what Laptop 16 users report
• The Genesys Logic USB hub remains functional after the card dies - other devices on the same bus are unaffected

References

• Audio Expansion Card breaking at least on a weekly basis for over a year now. #161 - Audio Expansion Card breaking at least on a weekly basis (Framework Laptop 16)
• Periodic loss of USB ports - Desktop #158 - USB-C PD port failures / Lost rear USB-C ports
• Community tracking thread - 120+ posts, 18 months, no fix
• Debian Bug #1100105 - same kernel errors reported