Deco XE75 v1 fw 1.4.3 — same disconnection issue as topic/738406, requesting beta firmware
Hello David-TP and TP-Link team,
I am experiencing the same issue described in topic/738406 (ThatGuy015) where you mentioned a beta firmware for Deco XE75 v1 that "optimized the 2.4 GHz connection by disabling the 2.4 GHz backhaul among Decos and allowing different Decos to use non-overlapping 2.4 GHz channels for mobile devices".
I would like to request access to that beta firmware. Below are my setup, evidence, and what I have already tried.
Setup
- 3x Deco XE75 v1 (one main, two satellites)
- Firmware 1.4.3 Build 20240926 Rel.38336 on all nodes
- Ethernet backhaul active on all 3 nodes (verified in Deco app)
- Region: EU (Italy)
- Operating mode: Access Point (Decos behind an upstream router on 192.168.1.1)
- Mixed clients: iPhones, Android phones, tablets, IoT (smart plugs, bulbs, robot vacuum, smart TVs), Linux laptops, Windows laptops
Symptoms
1. Frequent micro-drops on multiple devices during real-time activities (WhatsApp voice/video, video calls). Same as ThatGuy015: some devices drop dozens of times per hour, signal strength is excellent.
2. Continuous band-bouncing on stationary devices. One iPhone (XX:XX:XX:c1:f1:4f) switches between 5 GHz (wl02) and 2.4 GHz (wl22) dozens of times in 2 hours while sitting still in the same room. Example from main Deco log:
10:35:19 associate wl22
10:35:29 disassociate wl22
11:57:22 associate wl22
11:57:53 disassociate wl22
11:59:53 associate wl02
12:05:56 associate wl22 / disassociate wl02
...
3. Spontaneous roaming events on stationary clients overnight. From main Deco log between 02:35 and 08:15 (clients on charger, not moving):
stadbUpdateAssoc: Station XX:XX:XX:XX:DD:23 roams itself
stadbEntry_updateARBtmThreshold: update threshold ... reason 2
ar_config_set_thres: Null thres!
Multiple clients show associate / disassociate cycles every 30 to 60 seconds, with Null thres! warnings each time. Example sequence at 08:06-08:09:
08:06:32 XX:XX:XX:XX:DD:23 associate wl22
08:07:19 disassociate (Station roams itself, Null thres!)
08:07:49 associate again
08:08:45 disassociate (roams itself, Null thres!)
08:09:22 final disconnect
4. Recurring 802.11k / band-steering errors in the log:
estimatorDot11kIterateCB: Timeout waiting for 802.11k response
steeralgFindBestAPCallback: targetBand(X) != measuredBss->band(Y)
ar_pat_calc_cand_datarate: Cannot find <BSSID> in apinfo list
5. Deco app "Network Optimization" keeps changing the 2.4 GHz channel between 3, 4 and 5, never settling. It reports "channel 36 perfect" on 5 GHz, but an external scan shows all my Decos share channel 36 (see below).
External RF evidence (Linux laptop scan, fixed position)
2.4 GHz — 5 Deco BSSIDs all on channel 3:
XX:XX:XX:3b:3c:e6 ch 3 -55 dBm "MyWiFi"
XX:XX:XX:3b:3c:e7 ch 3 -55 dBm "MyWiFi 2.4"
XX:XX:XX:00:32:60 ch 3 -59 dBm (hidden)
XX:XX:XX:00:32:61 ch 3 -62 dBm "MyWiFi"
XX:XX:XX:00:32:62 ch 3 -60 dBm "MyWiFi 2.4"
5 GHz — all Deco BSSIDs on channel 36:
XX:XX:XX:3b:3c:e7 ch 36 -53 dBm (satellite A)
XX:XX:XX:00:32:6a ch 36 -55 dBm (satellite B)
This is co-channel interference between my own Deco nodes. The Network Optimization tool does not detect it because it does not see its own BSSIDs as external interference.
Independent monitoring
I run independent diagnostics outside the Deco system:
- SmokePing on a wired Proxmox container probing gateway, DNS, and WAN targets for 6+ hours: 0% packet loss, sub-millisecond jitter. The wired LAN and the ISP are completely clean.
- systemd Wi-Fi monitor on two stationary Linux laptops (Mint + Pop_OS), each connected to a different Deco satellite, logging RSSI, BSSID, bitrate and retries every 5 seconds. Both record band-bouncing and roaming events while the laptops never move.
The problem is isolated to the Deco Wi-Fi layer.
Cross-check with an alternative AP
To rule out client-side issues I temporarily activated a separate 802.11g AP (different hardware, channel 11, isolated SSID). My S25 Ultra phone made a WhatsApp voice call lasting more than 12 minutes with zero drop on that AP. The issue is not phone-specific — it is on the Deco side.
What I have already tried
- Network Optimization (multiple times)
- Disabled Fast Roaming
- Disabled Beamforming
- Power-cycled all Decos
- DHCP reservations / static IPs for problem devices
- Verified Ethernet backhaul on all 3 nodes
- Verified ISP and upstream router via independent SmokePing baseline
- Tested with an alternative AP (problem absent)
- Verified the issue is not phone-specific (occurs on iPhone, Android, Linux laptops, IoT devices)
Request
Could you please send me the beta firmware referenced in your reply on topic/738406:
"there is a beta firmware for Deco XE75_v1, which has optimized the 2.4 GHz connection by disabling the 2.4 GHz backhaul among Decos and allowing different Decos to use non-overlapping 2.4 GHz channels for mobile devices. (It could revert to the official 1.4.1 directly via web UI)"
I am happy to provide:
- Full system logs from all 3 Decos
- Deco app network map screenshots
- SmokePing 24-hour baseline graphs
- Linux Wi-Fi monitor event logs
- Before/after comparative tests once the beta is installed
Thank you very much for your support.
Best regards, Frappo
