Hi @menschel,
Sure, thanks. Added to the 4.6.0 issues list in the “Plane specific issues” section.
I’m seeing WDT resets on the bench. Indoors, no radio, pulled logs via mavproxy on linux, then fired up GCS to change a parameter and ran into 0x8000 which to my knowledge are watchdog resets.
14.04.2025 07:15:26 : PreArm: Waiting for RC
14.04.2025 07:15:26 : PreArm: Fence requires position
14.04.2025 07:15:26 : PreArm: waiting for terrain data
14.04.2025 07:15:26 : PreArm: Internal errors 0x8000 l:0 main_loop_stk
14.04.2025 07:15:26 : PreArm: Battery 1 low voltage failsafe
14.04.2025 07:15:26 : PreArm: GPS 1: Bad fix
14.04.2025 07:15:26 : Wait or rebo
14.04.2025 07:14:55 : PreArm: Waiting for RC
14.04.2025 07:14:55 : PreArm: Fence requires position
14.04.2025 07:14:55 : PreArm: waiting for terrain data
14.04.2025 07:14:55 : PreArm: Internal errors 0x8000 l:0 main_loop_stk
14.04.2025 07:14:55 : PreArm: Battery 1 low voltage failsafe
14.04.2025 07:14:55 : PreArm: GPS 1: Bad fix
14.04.2025 07:14:55 : Wait or rebo
14.04.2025 07:14:24 : PreArm: Waiting for RC
14.04.2025 07:14:24 : PreArm: Fence requires position
14.04.2025 07:14:24 : PreArm: waiting for terrain data
14.04.2025 07:14:24 : PreArm: Internal errors 0x8000 l:0 main_loop_stk
14.04.2025 07:14:24 : PreArm: Battery 1 low voltage failsafe
14.04.2025 07:14:24 : PreArm: GPS 1: Bad fix
14.04.2025 07:14:24 : Wait or rebo
14.04.2025 07:14:15 : IMU0: normal sampling 1.0kHz/1.0kHz
14.04.2025 07:14:15 : RCOut: DS600:1-2 PWM:3-10 NeoP:11-12
14.04.2025 07:14:15 : SpeedyBeeF405WING 0040001A 32335107 33343833
14.04.2025 07:14:15 : ChibiOS: 88b84600
14.04.2025 07:14:15 : ArduPlane V4.6.0-beta5 (b136aa2b)
14.04.2025 07:14:15 : IMU0: normal sampling 1.0kHz/1.0kHz
14.04.2025 07:14:15 : RCOut: DS600:1-2 PWM:3-10 NeoP:11-12
14.04.2025 07:14:15 : SpeedyBeeF405WING 0040001A 32335107 33343833
14.04.2025 07:14:15 : ChibiOS: 88b84600
14.04.2025 07:14:15 : ArduPlane V4.6.0-beta5 (b136aa2b)
14.04.2025 07:14:15 : IMU0: normal sampling 1.0kHz/1.0kHz
14.04.2025 07:14:15 : RCOut: DS600:1-2 PWM:3-10 NeoP:11-12
14.04.2025 07:14:15 : SpeedyBeeF405WING 0040001A 32335107 33343833
14.04.2025 07:14:15 : ChibiOS: 88b84600
14.04.2025 07:14:15 : ArduPlane V4.6.0-beta5 (b136aa2b)
thanks for the report! It isn’t a watchdog reset, it is “main loop stuck”, which is related but not quite the same thing. A main loop stuck means that the main loop didn’t run for 500ms. If the main loop doesn’t run for 2000ms it does a watchdog reset, but in your case it recovered.
Can you see if you can find the full tlog for this issue? I’d also like to know which parameter you changed (that should be in the tlog).
If you were running with LOG_DISARMED then the bin log would be good too.
Is the issue reproducible on your setup?
No use, system was in that state before changing the parameter which was just PITCH_TRIM_DEG.
Is it possible that due to ESC being unpowered some bdshot related action slows down the main loop. Other possibility is that ELRS receiver spams the system while already switched to WIFI.
I’m seeing occasionally “Link Lost, Link Down, Link Up” in mavproxy.
Another flight and tried to pull the logs
Requesting log list
Log 1 numLogs 500 lastLog 500 size 0
Log 2 numLogs 500 lastLog 500 size 0
...
Log 499 numLogs 500 lastLog 500 size 0
Log 500 numLogs 500 lastLog 500 size 0
AP: PreArm: AHRS: EKF3 Yaw inconsistent 162 deg. Wait or reboot
AP: PreArm: GPS 1: Bad fix
AP: PreArm: Battery 1 low voltage failsafe
AP: PreArm: waiting for terrain data
AP: PreArm: Fence requires position
AP: PreArm: Waiting for RC
AP: PreArm: AHRS: EKF3 Yaw inconsistent 162 deg. Wait or reboot
AP: PreArm: GPS 1: Bad fix
AP: PreArm: Battery 1 low voltage failsafe
AP: PreArm: waiting for terrain data
AP: PreArm: Fence requires position
AP: PreArm: Waiting for RC
Flight battery 100 percent
...
Then tried to look into it with MavFTP, found out that LOGS contains unspeakable sub folders, also could not be deleted via MavFTP, then operated out the SD Card from the highly integrated setup and plugged it into an SD Card Reader.
contents of APM/LOGS/
drwxr-xr-x 2 ubuntu ubuntu 16384 Okt 27 10:28 .
drwxr-xr-x 5 ubuntu ubuntu 16384 Jan 1 1980 ..
-r--r--r-- 1 ubuntu ubuntu 4294967295 Jan 1 1980 ■ç
-r--r--r-- 1 ubuntu ubuntu 4294967295 Jan 1 1980 ■ç
-rw-r--r-- 1 ubuntu ubuntu 4294967083 Jan 1 1980 ■.Ç
-rw-r--r-- 1 ubuntu ubuntu 4294967083 Jan 1 1980 ■.Ç
-rw-r--r-- 1 ubuntu ubuntu 4294967083 Jan 1 1980 ■.Ç
-rw-r--r-- 1 ubuntu ubuntu 721485824 Aug 1 2107 ''$'\001''+ . ■'
-rw-r--r-- 1 ubuntu ubuntu 721485824 Aug 1 2107 ''$'\001''+ . ■'
-rw-r--r-- 1 ubuntu ubuntu 4281008384 Jul 31 2044 + . ■Ç
-rw-r--r-- 1 ubuntu ubuntu 4281008384 Jul 31 2044 + . ■Ç
-rw-r--r-- 1 ubuntu ubuntu 4294912769 Apr 2 1980 .■Ç
-rw-r--r-- 1 ubuntu ubuntu 4294912769 Apr 2 1980 .■Ç
-rw-r--r-- 1 ubuntu ubuntu 4294912769 Apr 2 1980 .■Ç
fat32 filesystem was corrupted and read only. Formatted the whole SD Card afterwards.
No clue how this happened.
I see what you mean, @menschel . Thanks for the report!
I’ll try to reproduce this in SITL, and make a regression test.
Then fixing it should be easy enough.
Looks good to me but maybe this is the chance to align the speed reference value used for the rotate speed feature, it should be groundspeed or airspeed, not both. IMO groundspeed is correct but there have been other opinions expressed. Airspeed can be used if there is a sensor available.
I’ve been misusing that feature for hand launch and it works great BTW.
Ground speed makes no sense for an aircraft equipped with an airspeed sensor. You can have quite a bit of airspeed in headwind.
Yes, if there is an airspeed sensor.
Airspeed estimation without an airspeed sensor is erratic.
From what I learned in the field, unless you are free of obstructions in every direction, wind effect becomes less and less relevant close to the ground.
I’ve been flying/landing in close proximity to trees, bushes etc. and strong wind conditions usually dictate to approach with the correct ground speed as the plane otherwise stalls in flare. It is reversed for takeoff IMO.
When I throw the plane, I induce 10km/h of false wind speed and consequently during the climb it sees 10km/h too low airspeed until it turns and corrects the wind speed estimation.
The same thing happens in the other direction, wind speed and the consequent error on airspeed estimation does not bleed / align with groundspeed during flare.
So I see the plane lying in the grass showing 10km/h of wind and airspeed which is a 30-50% error in my case.