RZ616 Wi-Fi module: one year of driver-initiated link drops (Framework Laptop 13, AMD Ryzen 7040)
Prepared 2026-09-04 from the machine's own records: Windows event logs, setupapi.dev.log (nine retained files, 2025-08-29 → 2026-09-04), the driver store, and the Windows Update history. Times are local (CEST). Every number below is reproducible from the logs listed at the end.
Summary
- The Wi-Fi card (AMD RZ616 = MediaTek MT7922) drops its link by itself. Every disconnect the WLAN log holds — 27 since 2026-08-29 — carries the same reason: "The network is disconnected by the driver." It happens on three different networks (home fibre-box, a public hotspot, the university's
uzh-5GHz). - Windows has had to reset the card 82 times between 2025-08-29 and 2026-09-04 (
setupapi"Restart Device" records forPCI\\VEN_14C3&DEV_0616). The rate is similar under two successive Framework bundle drivers — 3.4.0.1268 (bundle v2.05: 8.4 resets/month) and 3.4.0.1335 (bundles v2.07/v2.08: 5.8 resets/month). It is not one bad driver version. - On 2026-09-04 the card dropped 13 times and refused to associate 25 times in a single morning. Four manual adapter restarts did not help. A rollback is not an option: the only older package in the store is 3.4.0.1268, which had the worse rate.
- Downstream: the IKEv2 VPN tunnel dies about 40 s after each link stall (RAS termination codes 828/829, 17 times since 2026-07-08), and the card's Bluetooth half logs "cannot store link keys" (BTHUSB event 18) 48 times since July.
- Request: replacement of the Wi-Fi module.
Device
| Field | Value |
|---|---|
| Laptop | Framework Laptop 13 (AMD Ryzen 7040 Series), SKU FRANDGCP07, mainboard FRANMDCP07 rev A7 |
| CPU | AMD Ryzen 7 7840U w/ Radeon 780M |
| BIOS | 03.18 (released 2026-01-08, installed 2026-01-28) |
| OS | Windows 11 Pro 25H2, build 26200; installed 2024-12-18 |
| Wi-Fi module | RZ616 Wi-Fi 6E 160MHz (MediaTek MT7922) — PCI\\VEN_14C3&DEV_0616&SUBSYS_E61614C3&REV_00; first installed 2024-12-18 |
| Bluetooth half | USB\\VID_0E8D&PID_E616; driver 1.1043.0.555 (2025-08-13) |
| Current Wi-Fi driver | MediaTek 3.4.0.1335, INF date 2025-08-26 (mtkwl6ex.inf → oem162.inf), installed 2026-01-11 14:25 from Framework driver bundle v2.07 |
| Firmware in package | WIFI_RAM_CODE_MT7922_1.bin 752,052 bytes (3.4.0.1335) vs 751,860 bytes (3.4.0.1268) |
| Adapter settings at time of report | Roaming Aggressiveness: Medium-High · Preferred Band: 2.4 GHz · Power Saving: Auto · Transmit Power: Highest (Windows defaults) |
Driver history
| Date | Event | Result |
|---|---|---|
| 2024-12-18 | Device first installed with driver 3.3.0.908 (INF 2023-11-27), from the Framework bundle of that era (2024_10_02) | active until 2025-08-25; before the retained logs |
| 2025-08-25 | Framework bundle v2.05 (2025_07_18) and BIOS 3.16: driver 3.4.0.1268 installed (package staged 11:48 that day; shown as Installed in the 2025-11-17 record) | active until 2026-01-11 |
| 2025-11-17 12:31 | Bundle re-run: Wi-Fi driver 3.4.0.1268 | "No better matching drivers found — no devices were updated" (already installed) |
| 2026-01-11 14:25 | Framework bundle v2.07 (2025_11_20): Wi-Fi driver 3.4.0.1335 installed | SUCCESS; 908 removed by disk cleanup on 2026-01-12 |
| 2026-01-28 | BIOS 3.18 installed | — |
| 2026-06-23 09:19 | Framework bundle v2.08 (2026_06_01): Wi-Fi driver 3.4.0.1335 | "No better matching drivers found — no devices were updated" (v2.08 ships the same driver; the AMD chipset parts installed normally) |
| 2026-09-04 11:44 | Device Manager → Roll Back Driver dialog opened | cancelled (exit 0x4C7); no driver change. The only older package available is 3.4.0.1268 |
Windows Update has never delivered a MediaTek driver to this machine (no MediaTek entry in the update history, and a policy on the machine blocks driver search on Windows Update); every Wi-Fi driver came from a Framework bundle.
Failure history over one year — card resets recorded by Windows
A "reset" is a setupapi Restart Device record for the card. It is written both when Windows' WLAN AutoConfig service recovers the adapter on its own (System event 4003, "limited connectivity, attempting automatic recovery") and when the user disables/enables it by hand. Both mean the card stopped working.
| Month | Resets | Driver era |
|---|---|---|
| 2025-08 (from 08-29) | 4 | A · 3.4.0.1268 (bundle v2.05) |
| 2025-09 | 4 | A |
| 2025-10 | 3 | A |
| 2025-11 | 13 | A |
| 2025-12 | 13 | A |
| 2026-01 | 0 | changeover 01-11 |
| 2026-02 | 2 | B · 3.4.0.1335 (bundle v2.07 = v2.08) |
| 2026-03 | 0 | B |
| 2026-04 | 6 | B |
| 2026-05 | 5 | B |
| 2026-06 | 3 | B |
| 2026-07 | 17 | B |
| 2026-08 | 9 | B |
| 2026-09 (4 days) | 3 | B |
- Era A (3.4.0.1268, 2025-08-29 → 2026-01-11): 37 resets in 134 days = 8.4 per month. (The original 3.3.0.908 ran 2024-12-18 → 2025-08-25, before the retained logs.)
- Era B (3.4.0.1335, 2026-01-11 → 2026-09-04): 45 resets in 235 days = 5.8 per month.
The last five weeks in detail (event logs)
System log from 2026-07-29; WLAN AutoConfig operational log from 2026-08-29 (older WLAN entries have rotated out). Only days with adapter events are listed.
| Date | Link drops by driver | Failed connects | Auto-recoveries | Driver-module restarts | VPN drops | Resumes from standby |
|---|---|---|---|---|---|---|
| 2026-07-31 | · | · | 1 | 1 | 1 | 4 |
| 2026-08-01 | · | · | 1 | · | 1 | 5 |
| 2026-08-15 | · | · | 1 | · | 1 | 3 |
| 2026-08-17 | · | · | · | 2 | · | 11 |
| 2026-08-19 | · | · | · | 2 | 2 | 4 |
| 2026-08-20 | · | · | 1 | 1 | 1 | 7 |
| 2026-08-26 | · | · | 1 | 1 | 2 | 16 |
| 2026-08-27 | · | · | 2 | 1 | 3 | 11 |
| 2026-08-28 | · | · | 1 | 1 | · | 13 |
| 2026-08-29 | 3 | · | 1 | 1 | · | 8 |
| 2026-08-30 | 3 | 1 | · | 2 | · | 4 |
| 2026-08-31 | 1 | · | 1 | · | 1 | 9 |
| 2026-09-01 | 4 | 2 | 1 | 1 | 3 | 5 |
| 2026-09-02 | 2 | 3 | 1 | 1 | 2 | 13 |
| 2026-09-03 | 1 | 3 | · | · | · | 8 |
| 2026-09-04 | 13 | 25 | 1 | 4 | 2 | 2 |
- Disconnects by network (whole WLAN log):
FibreBox_X6-277357-Guest24 (home),publicVB2,uzh-5GHz1. - 15 of the 27 disconnects occurred within 15 minutes of a resume from Modern Standby. A hint, not a proof: the 2026-09-04 storm happened with the laptop awake.
What one failure looks like (2026-09-04, local time)
11:25:32 WLAN 11004 Wireless security stopped (on a link that was up)
11:25:39 WLAN 11010/11005 security started, succeeded
11:25:45 WLAN 11004 security stopped <- re-keys every 6-8 s from here
11:26:02 WLAN 8003 DISCONNECTED -- "The network is disconnected by the driver."
11:26:06 WLAN 8000 reconnect started
11:26:07 WLAN 8001 connected
11:26:14 ... 11:26:31 security stopped / started, three more cycles
11:26:45 RAS 20226 VPN "INI" terminated, reason code 828
11:31:19 RAS 20226 VPN terminated again after the same pattern (8003 at 11:30:51)
11:34:04 WLAN 8003 disconnected by the driver
11:34:08 ... 11:36:33 WLAN 8002 six consecutive association failures
Earlier that morning: Windows' own adapter recovery at 11:04:40 (event 4003, recovery type 4, trigger reason 5), driver-module restarts at 10:57, 11:02, 11:04 and 11:05 (manual disable/enable), a system reboot at 10:59.
What has been tried
- Manual adapter disable/enable: four times on 2026-09-04 alone; 82 resets over the year including Windows' own.
- Framework driver bundles v2.05 (→ 3.4.0.1268), v2.07 (→ 3.4.0.1335) and v2.08 (same driver, nothing to update).
- Driver rollback considered and not done: the only older package is 3.4.0.1268, whose era was worse.
- BIOS kept current (3.07 → 3.16 → 3.18).
- Working over phone tethering when the card is down.
Ruled out
- The access point. Three unrelated networks show the same driver-initiated drop.
- The VPN. The RAS terminations (codes 828 idle-timeout and 829 link-failure) follow the link stall by \~40 s and the profile auto-redials; the other codes in the log (631, 830, 832) are user disconnects, logoffs and hibernation.
- A single driver version. Both bundle drivers fail at comparable rates; the latest bundle (v2.08) ships the same driver; the older one is not an improvement.
- The OS build. 25H2 build 26200, kept current.
In parallel, the Bluetooth half of the same module logs "cannot store link keys" (BTHUSB event 18): 3 in July, 35 in August, 10 in the first four days of September.
Request
Replacement of the RZ616 Wi-Fi module under warranty, or the Intel AX210-based module Framework supplies for the AMD Ryzen 7040 mainboard. The fault follows the card across driver versions, BIOS releases and networks, and the Bluetooth half of the same module misbehaves in parallel.
Evidence sources (all on the machine, available on request)
- Windows event logs:
System(from 2026-07-29),Application/RasClient (from 2026-07-08),Microsoft-Windows-WLAN-AutoConfig/Operational(from 2026-08-29),Microsoft-Windows-NetworkProfile/Operational(from 2026-07-30). C:\\Windows\\INF\\setupapi.dev.logand the eight rotatedsetupapi.dev.<date>.logfiles (2025-08-29 → 2026-09-04).- Driver store packages
mtkwl6ex.inf_amd64_c79f23b85b5ce577(3.4.0.1268) andmtkwl6ex.inf_amd64_bfac85d912f87f94(3.4.0.1335). - Windows Update history (
IUpdateSearcher.QueryHistory). netsh wlan show wlanreportoutput (wlan-report-latest.html, generated 2026-09-04).
All of the above is packaged and available on request: raw .evtx exports of the four event logs (plus CSV), the nine setupapi.dev logs with a card-only extract, both MediaTek packages' INF files and file listings, the adapter's properties and advanced settings, every PnP property of the Wi-Fi and Bluetooth halves, systeminfo/BIOS/hotfixes/Windows Update history, the list of Framework bundles installed with dates, and the wlanreport HTML.