C6: SYN/arbitration errors over USB and Wi-Fi with flexoTHERM, VRC700 and VR900

[christianspuller]

Description

I am trying to integrate a Vaillant flexoTHERM exclusive with Home Assistant using an eBUS Adapter Shield C6 and ebusd 26.1. Active reads intermittently fail over Wi-Fi and repeatedly fail over direct USB with ERR: SYN received followed by arbitration won in invalid state ready.

I would appreciate guidance on a minimal supported configuration and the specific trace needed to distinguish a firmware/interface problem from bus arbitration or configuration issues.

Environment

• ebusd: 26.1.26.1, official Debian 13 amd64 package with MQTT support.
• Adapter: eBUS Adapter Shield C6, ESP32-C6; firmware build 20260917, reported as 1.1[6911].1[6911].
• Host: Proxmox VE 9.1.6; unprivileged Debian 13 LXC, ebusd running as root.
• Original transport: enhanced high-speed TCP over Wi-Fi.
• Comparison: enhanced USB serial, ens:/dev/ttyACM0, with the adapter switched to Enhanced protocol / USB serial. Proxmox device passthrough gives container root access to the character device (166:0).
• Heating system: Vaillant flexoTHERM exclusive, VRC700 and VR900. Exact VWF model has not yet been verified from the nameplate.
• Existing masters seen: 03, 10, 71, f1; ebusd uses its default master address 31.
• Identified slaves:
  • 08: HMU00, SW0303, HW0403
  • 15: 70000, SW0209, HW4103
  • 76: VWZ00, SW0303, HW0403
  • f6: NETX0, SW4039, HW3103

Configuration used for the USB comparison

The intended options retained from the Wi-Fi configuration, with MQTT credentials omitted:

--device=ens:/dev/ttyACM0
--scanconfig
--scanretries=1
--pollinterval=10
--latency=50
--accesslevel=service
--mqttjson
--mqttint=/etc/ebusd/mqtt-hassio-hmu.cfg
--log=all:notice

Broker connection/authentication options are also configured. The exact full running argument list has not been independently captured after the USB switch.

The custom MQTT integration file is a copy of the supplied mqtt-hassio.cfg with filter-seen=5 retained, the circuit filter restricted to hmu/700, and exact message-name filters for Status01, two LiveMonitor power values and selected VRC700 temperatures/modes. It also allows the service level. There are no intentional heating-control writes in these tests; entries such as received update-write ... QQ=10 originate from the existing controller.

Observations

Wi-Fi: Passive reception works and forced reads sometimes succeed. During a real heat-pump run, forced LiveMonitor reads returned 2.6 kW electrical consumption and 8.9 kW environmental yield, and these reached HA. Automatic reads often failed, leaving old zero values displayed for over ten minutes. SYN, arbitration-lost and timeout errors occurred. Reported symbol latency ranged from 6 ms up to approximately 72 ms, with zero reconnects.

The Vaillant app repeatedly stopped loading heating data during active ebusd operation and recovered after ebusd was stopped. This is a repeated user observation, not a proven root cause.

USB: The device opens correctly, firmware is detected, the bus signal is acquired, and symbol latency is consistently 4 ms. However, repeated SYN/arbitration errors remain. The Vaillant app continued working during the brief USB test, unlike the earlier Wi-Fi tests. No long-term USB test has been completed.

Latest USB info:

device: /dev/ttyACM0, serial high speed, enhanced, firmware 1.1[6911].1[6911]
access: service
signal: acquired
min symbol latency: 4
max symbol latency: 4
scan: running
reconnects: 0
masters: 5
messages: 226
poll: 2
update: 23

Only 08.hmu.csv was loaded at this point; 15 was identified but its configuration scan failed. Status01 continued arriving from QQ=10.

Representative USB log (UTC)

2026-10-21 15:06:31.295 [bus error] poll hmu LiveMonitorCurrentConsumedPower failed: ERR: SYN received
2026-10-21 15:06:31.298 [bus notice] arbitration won in invalid state ready
2026-10-21 15:06:37.056 [main error] scan config 15: ERR: SYN received
2026-10-21 15:06:39.298 [main error] scan config 76: ERR: SYN received
2026-10-21 15:06:39.302 [bus notice] arbitration won in invalid state ready
2026-10-21 15:06:42.301 [bus error] poll hmu LiveMonitorCurrentYieldPower failed: ERR: SYN received
2026-10-21 15:06:42.306 [bus notice] arbitration won in invalid state ready
2026-10-21 15:06:53.708 [main error] scan config 15: ERR: SYN received
2026-10-21 15:06:53.755 [bus error] poll hmu LiveMonitorCurrentConsumedPower failed: ERR: SYN received
2026-10-21 15:06:53.759 [bus notice] arbitration won in invalid state ready
2026-10-21 15:07:00.417 [update notice] received read hmu Status01 QQ=10: 24.0;23.5;16.750;-;50.5;off
2026-10-21 15:07:00.948 [main notice] scan completed 1 time(s), check again
2026-10-21 15:07:04.259 [bus error] poll hmu LiveMonitorCurrentYieldPower failed: ERR: SYN received
2026-10-21 15:07:04.263 [bus notice] arbitration won in invalid state ready
2026-10-21 15:07:07.480 [main error] scan config 15: ERR: SYN received

Questions

1. Is this pattern known with C6 firmware 20260917 and ebusd 26.1, including direct USB?
2. What minimal settings and targeted debug/raw trace would you recommend? In particular, should the retained --latency=50 be removed for this USB comparison?
3. Is there a known interaction with a parallel VR900/VRC700, or anything specific to check before trying another master address? No address conflict has been established.
4. How should the failing 76/f6 configuration scans be handled while preserving automatic configuration for 08/15?

I found older reports such as #990, but am not assuming the older v5 Wi-Fi issue is the same as this C6 USB issue.