Typhoon H 480 PX4 v1.10 (Stability issues ;-)
- Thread starter Pöllö
- Start date
-
- Tags
- programming px4
Thank you for the notice. I will revert to the previous version for my next flight.
n8it
n8it
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
I am very sorry to hear about your crash, Helmut.
It seems that the firmware allows disarming in-flight if the left hand stick is held in lower left corner for a while.
The previous firmware is likely to have this issue too, it seems to be PX4's "intended" behaviour. Between the firmware versions, there are no changes regarding arm/disarm (and that is "stock" PX4 regarding it). I'll take a close look into it how we could go around this.
Last edited:
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
It seems, that it is "intended" (but nasty) behaviour of the PX4, to allow in-flight disarm if disarming stick combination is given (Lower stick to lower left corner) and held there for few seconds. As this is not specific to the Thunderbird, the previous version WILL disarm in-flight, too if it is requested to do so, even accidentally. Maybe we should switch to using a toggle switch instead of the stick combination. However, I don't quite like that option either.
It seems that we have few options here, which I have to consider:
- Avoid the full left rudder - throttle fully down -combination in-flight and keep the PX4's arming logic stock (I know that you probably won't like this approach...).
- Use the procedure described by H-Elsner: dial the speed down a bit to prevent the full input from the sticks, as the arm/disarm requires the full range of sticks
- Use Arm/Disarm switch: Clear, but a way more dangerous, read below. Immediate disarm, even in-flight when the switch is toggled. Knowing the quality of the switch and wiring on the ST16 and the possibility for accidental flipping of a switch... maybe not. (Edit: I am not going to configure this to the parameter set supplied with the FW. You can configure it by yourself if you want, but I feel this being even more dangerous than accidental stick combination disarm in-flight.)
- Modify PX4's code regarding the Arm/disarm logic. This is something that I would avoid doing, as I would like to keep the code base as stock as possible. And the amount of testing that is required with the modifications regarding this very critical functionality will be very high. Knowing the complex nature of the firmware and the high level of automation, this approach is very prone for very fatal bugs.
Arm/Disarm Switch
The arm/disarm switch is a direct replacement for the default stick-based arming/disarming mechanism (and serves the same purpose: making sure there is an intentional step involved before the motors start/stop). It might be used in preference to the default mechanism because:
It seems that we have few options here, which I have to consider:
- Avoid the full left rudder - throttle fully down -combination in-flight and keep the PX4's arming logic stock (I know that you probably won't like this approach...).
- Use the procedure described by H-Elsner: dial the speed down a bit to prevent the full input from the sticks, as the arm/disarm requires the full range of sticks
- Use Arm/Disarm switch: Clear, but a way more dangerous, read below. Immediate disarm, even in-flight when the switch is toggled. Knowing the quality of the switch and wiring on the ST16 and the possibility for accidental flipping of a switch... maybe not. (Edit: I am not going to configure this to the parameter set supplied with the FW. You can configure it by yourself if you want, but I feel this being even more dangerous than accidental stick combination disarm in-flight.)
- Modify PX4's code regarding the Arm/disarm logic. This is something that I would avoid doing, as I would like to keep the code base as stock as possible. And the amount of testing that is required with the modifications regarding this very critical functionality will be very high. Knowing the complex nature of the firmware and the high level of automation, this approach is very prone for very fatal bugs.
Arm/Disarm Switch
The arm/disarm switch is a direct replacement for the default stick-based arming/disarming mechanism (and serves the same purpose: making sure there is an intentional step involved before the motors start/stop). It might be used in preference to the default mechanism because:
- Of a preference of a switch over a stick motion.
- It avoids accidentally triggering arming/disarming in-air with a certain stick motion.
- There is no delay (it reacts immediately).
- Manual mode
- Acro mode
- Stabilized
- Rattitude
Last edited:
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
Whoa. This really got me! Luckily H-elsner found out this disarming behaviour when running the drone on hand without propellers, so no damage done.
But a very good find still, thank you very much for this. I will verify this behaviour in the PX4's Hardware In The Loop simulator in few days.
I would consider flying with both of the firmwares still safe regarding this. Just avoid the full left, full down combination of the throttle/rudder stick held for few seconds in flight.
But a very good find still, thank you very much for this.
I will verify this behaviour in the PX4's Hardware In The Loop simulator in few days.I would consider flying with both of the firmwares still safe regarding this. Just avoid the full left, full down combination of the throttle/rudder stick held for few seconds in flight.
Last edited:
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
I will disarm in flight check against the flight modes. Is it possible to increase the time for the disarm command down/left CSC command? It is much faster than down/right, regarding feeling immediately but maybe 1sec or lower to hold down.
I wonder how a PX4 acro flyer will handle that feature.
If it is PX4 specific, then let it as it is. I like to go down in spirals but I should do that any more.
Ahh, and I will check this CSC command if some pitch is added. Let's see if this helps.
I will report.
br HE
I wonder how a PX4 acro flyer will handle that feature.
If it is PX4 specific, then let it as it is. I like to go down in spirals but I should do that any more.
Ahh, and I will check this CSC command if some pitch is added. Let's see if this helps.
I will report.
br HE
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
To be honest, I have not been able to disarm it mid-flight. I do also those spirals, and I have tested that stick combination too. But my drone might have been saved by a bad potentiometer in the left stick...
I'll recheck this. Please avoid those full left rudder, full down spirals just to be sure.
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
In this case no drone was harmed
However, my drones have a hard life:
br HE
However, my drones have a hard life:
br HE
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
Toni is right. Switching off in the air is a safety feature of PX4. This is only possible in the manual flight modes. I tested it again. For Position hold or Altitude mode emergency shutdown mid-air is not possible. Even if the left stick is not pulled fully, i.e. the servo range was limited by the slider rabbit / turtle, it is also not possible to stop motors.
We have to be able to live with it. Seems PX4 racer drones can handle it too. Pitch / roll sticks don't matter. So spiraling down (with full servo range -100%/-100%) to the left could end up in a desaster.
The emergency shutdown comes after 1s hold this CSC command (left stick left-down).
Older guys like me have no problem because of the tremor. ?
br HE
We have to be able to live with it. Seems PX4 racer drones can handle it too. Pitch / roll sticks don't matter. So spiraling down (with full servo range -100%/-100%) to the left could end up in a desaster.
The emergency shutdown comes after 1s hold this CSC command (left stick left-down).
Older guys like me have no problem because of the tremor. ?
br HE
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
Thank you very much h-elsner for your testing! I tried it too, but I was not able to reproduce it indoors, at least with Stabilize mode. First I replaced my left gimbal to a new one (now on my third one..), calibrated it and verified that I can arm and disarm the drone when it is sitting on the floor. Then I armed it, throttled up, and swung and shook it around to simulate the flight. I was not able to disarm it as long I kept it in motion. When I stopped swinging it, it disarmed after few seconds, when it thought it was landed. The full left rudder and full throttle down combination was on all the time.
I tried it too, but I was not able to reproduce it indoors, at least with Stabilize mode. First I replaced my left gimbal to a new one (now on my third one..), calibrated it and verified that I can arm and disarm the drone when it is sitting on the floor. Then I armed it, throttled up, and swung and shook it around to simulate the flight. I was not able to disarm it as long I kept it in motion. When I stopped swinging it, it disarmed after few seconds, when it thought it was landed. The full left rudder and full throttle down combination was on all the time.
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
Wow, this is really cool looking custom frame! ;-)
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
Yes, and this was one of the last flights of it. But the evolution of the Blade 350QX gone ahead. The whole story is here:
Unfortunately only in German but the pictures are international.
The latest frame here in action:
Now you know why I love my Thunderbird...
br HE
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
Well.. It kinda is already. ? There is ACRO mode available to be mapped to one of ST16's switches with the QGroundcontrol, if you really want to push the TH to the limits...
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
Logs from yesterday test flight:
Flight Review
Description: Thunderbird test
Feedback:
Vehicle type: Hexarotor
Airframe: Generic Hexarotor x geometry, Hexarotor x
Hardware: YUNEEC_TYPHOON_H
Vehicle UUID: 000100000000353534303335510f00290034
Software git hash: ef114348bdb8f9c064012f2db862e904934960c8 (branch: RGB_LED)
Upload file name: log_18_2020-5-17-16-30-18.ulg
All went well with this flight above.
At the end of a second flight I have had a fast (but good) landing again. The battery is really bad and I have to drow it away. On the other hand - a good test what may happen when voltage drops.
Questions:
- What shall happen when BAT_LOW_THR (set to 20%) was reached? Can it be the status LED goes to red? I saw it flickering purple red sometimes.
- If BAT_CRIT_THR (=10%) was reached, RTH? Never happens.
- BAT_EMERGEN_THR (=7%) landing immediately? This is what I have had. But when it happen over water or wood or some other place where I don't want to go down? I can interupt the action but in danger of hard lading.
Another question that comes in my mind: In case of immediately landing, will it go up tho RTH height first? I think it did. But this makes not sense and is also not good for a poor battery.
But it could also be that RTH and land immediately was done at nealy the same time, I'm not sure.
br HE
Flight Review
Description: Thunderbird test
Feedback:
Vehicle type: Hexarotor
Airframe: Generic Hexarotor x geometry, Hexarotor x
Hardware: YUNEEC_TYPHOON_H
Vehicle UUID: 000100000000353534303335510f00290034
Software git hash: ef114348bdb8f9c064012f2db862e904934960c8 (branch: RGB_LED)
Upload file name: log_18_2020-5-17-16-30-18.ulg
All went well with this flight above.
At the end of a second flight I have had a fast (but good) landing again. The battery is really bad and I have to drow it away. On the other hand - a good test what may happen when voltage drops.
Questions:
- What shall happen when BAT_LOW_THR (set to 20%) was reached? Can it be the status LED goes to red? I saw it flickering purple red sometimes.
- If BAT_CRIT_THR (=10%) was reached, RTH? Never happens.
- BAT_EMERGEN_THR (=7%) landing immediately? This is what I have had. But when it happen over water or wood or some other place where I don't want to go down? I can interupt the action but in danger of hard lading.
Another question that comes in my mind: In case of immediately landing, will it go up tho RTH height first? I think it did. But this makes not sense and is also not good for a poor battery.
But it could also be that RTH and land immediately was done at nealy the same time, I'm not sure.
br HE
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
Q500log2kml updated for Thunderbird:
github.com
Flight modes and color coding are currently as follows:
Maybe there will be changes in future.
br HE
GitHub - h-elsner/Q500log2kml: Display and analyze FlightLog data from Yuneec drones
Display and analyze FlightLog data from Yuneec drones - h-elsner/Q500log2kml
github.com
Flight modes and color coding are currently as follows:
Maybe there will be changes in future.
br HE
WOW! I am completely impressed by what you guys are doing. To make a long story short, a few years ago at work I had been doing some stuff with the Intel Aero drone board, more tech documentation, not so much development other than software and getting Intel RealSense working with the Aero board.
For fun, I purchased a used DIY drone based on the Tarot 680 Pro. It was already built with DJI parts. I crashed and fixed the frame. But have never been able to get the radio to reconnect with the flight controller. Been sitting on the shelf for a few years.
Many times I have wanted to replace all the hardware with PX4 ecosystem. But, I thought to myself... Why? My drone does not have a camera. It's just a drone. The guy did sell me a little camera gumball, but no camera and I don't really want a gopro. I want a nice 4K camera.
I've heard that Yuneec (at least some models) are based/use PX4 technology and wondered if hacking these drones were possible.
So, lately I've been toying with the idea of purchasing a Typhoon H3. However, because of the hardships the company seems to be experiencing here in the US, I'm a bit fearful. However seeing that there are people who know how to hack/fix these even if Yuneec as a company leaves the US (not saying it WILL happen) is inspiring.
So this leads me to some questions...
For fun, I purchased a used DIY drone based on the Tarot 680 Pro. It was already built with DJI parts. I crashed and fixed the frame. But have never been able to get the radio to reconnect with the flight controller. Been sitting on the shelf for a few years.
Many times I have wanted to replace all the hardware with PX4 ecosystem. But, I thought to myself... Why? My drone does not have a camera. It's just a drone. The guy did sell me a little camera gumball, but no camera and I don't really want a gopro. I want a nice 4K camera.
I've heard that Yuneec (at least some models) are based/use PX4 technology and wondered if hacking these drones were possible.
So, lately I've been toying with the idea of purchasing a Typhoon H3. However, because of the hardships the company seems to be experiencing here in the US, I'm a bit fearful. However seeing that there are people who know how to hack/fix these even if Yuneec as a company leaves the US (not saying it WILL happen) is inspiring.
So this leads me to some questions...
- Outside of areal acrobatics, what are the benefits of this?
- Can the Typhoon H3 be hacked?
- If you flash the firmware, there is no going back... Right?
- What do you loose that a stock Typhoon has?
- What do you gain?
- Do you still have full camera control?
h-elsner
Premium Pilot
- Joined
- Mar 23, 2016
- Messages
- 1,947
- Reaction score
- 1,641
- Location
- Bavaria / Germany
- Website
- h-elsner.mooo.com
Let me try to answer as good as I can. @Pöllö as the expert and developer can say more.
1. Mission mode. This may help to replace expensive H520 for some tasks like fawn rescue by drone with infrared camera.
2. Maybe, but for no reason. H plus and H3 are already on PX4 autopilot.
3. Right.
4. Currently the serial connection between camera and drone is not yet working with the firmware. You cannot turn around the cam and do tasks like CCC, Orbit and so on.
5. The freedom to do anything, open source SW and the acrobatics which is my favourite at the moment.
6. Yes, except panning. Camrea is controlled by CGI commands via 5GHz WiFi. This is independent from drone firmware (valid for H480, not for H520 or H3!).
You are invited to take part at the project: tonirosendahl/Thunderbird
br HE
1. Mission mode. This may help to replace expensive H520 for some tasks like fawn rescue by drone with infrared camera.
2. Maybe, but for no reason. H plus and H3 are already on PX4 autopilot.
3. Right.
4. Currently the serial connection between camera and drone is not yet working with the firmware. You cannot turn around the cam and do tasks like CCC, Orbit and so on.
5. The freedom to do anything, open source SW and the acrobatics which is my favourite at the moment.
6. Yes, except panning. Camrea is controlled by CGI commands via 5GHz WiFi. This is independent from drone firmware (valid for H480, not for H520 or H3!).
You are invited to take part at the project: tonirosendahl/Thunderbird
br HE
Thanks h-elsner... I appreciate the feedback. I think the ONLY thing keeping me from buying Yuneec is the issues with supply here in the US, but that's a completely different topic.
- Joined
- Jan 20, 2019
- Messages
- 183
- Reaction score
- 142
- Age
- 40
Hi all,
I've encountered one bizarre bug today with the latest "untested" firmware with the CD, or "Crazy Destroyer", in its name. When I started the drone, and was going to insert the USB while powered on for uploading the mission, the drone armed itself, disarmed, armed and cycled its landing gear erratically, and went also to the radio failsafe mode for few moments, dropping the radio connection. It did not spool up the motors from the idle or pose any serious hazard, but this is something that I still take very seriously. The PX4 logs show something that could be weird RC behavior, and the logs of ST16 seem to be missing for that moment.
I restarted the drone by removing and reinserting the battery three times, repeating the very same behavior - the drone armed as soon as the GPS fix was got, but stayed in the idle. No mission was uploaded at this point. The ST16 was powered on, not resetted between these three attempts. Then, after touching the ST16, double-clicking the battery icon for channels check, all these issues went away and I was not able to trigger this kind of behaviour again. I'll do more testing this weekend. After the last incident, when the drone was still armed, but this time with no propellers and I was able to handle it more freely, it reacted perfectly to the commands and tilting it. Something triggered the arming, the drone was not messed up, and I do not yet know why this happened.
However, in the /src/lib/st24.cpp, it can be seen that there is CRC checksum checking in place for the serial data received from the SR24 receiver, so the channel data is very likely not to be corrupted between the ST16's internal radio module and the Typhoon's MCU. But, what data is fed to this module by the ST16 is another question... Also, as this kind of behavior is not seen previously with the PX4, it is pretty safe to assume, that the PX4 is also OK. Also, if there is something seriously wrong with the Thunderbird fork, this kind of behavior would have been appeared previously during all the testing, as I was able to repeat this three consecutive times before touching the ST16. This is going to be a hard one for me to track down as there is a possibility that this is something erratic caused by the ST16, and some erratic ST16 behavior such as channels getting messed up is not uncommon even in this forum.
At this point I am assuming that the ST16 was doing something stupid, especially as ~six flights ago I replaced my left gimbal (throttle and rudder), and this is why I am going to keep the firmware image available still.
I'll get back to this and your other questions as soon as I have resolved this one or tested it enough to be confident enough that the issue was a misbehaving ST16...
Be careful.
I've encountered one bizarre bug today with the latest "untested" firmware with the CD, or "Crazy Destroyer", in its name. When I started the drone, and was going to insert the USB while powered on for uploading the mission, the drone armed itself, disarmed, armed and cycled its landing gear erratically, and went also to the radio failsafe mode for few moments, dropping the radio connection. It did not spool up the motors from the idle or pose any serious hazard, but this is something that I still take very seriously. The PX4 logs show something that could be weird RC behavior, and the logs of ST16 seem to be missing for that moment.
I restarted the drone by removing and reinserting the battery three times, repeating the very same behavior - the drone armed as soon as the GPS fix was got, but stayed in the idle. No mission was uploaded at this point. The ST16 was powered on, not resetted between these three attempts. Then, after touching the ST16, double-clicking the battery icon for channels check, all these issues went away and I was not able to trigger this kind of behaviour again. I'll do more testing this weekend. After the last incident, when the drone was still armed, but this time with no propellers and I was able to handle it more freely, it reacted perfectly to the commands and tilting it. Something triggered the arming, the drone was not messed up, and I do not yet know why this happened.
However, in the /src/lib/st24.cpp, it can be seen that there is CRC checksum checking in place for the serial data received from the SR24 receiver, so the channel data is very likely not to be corrupted between the ST16's internal radio module and the Typhoon's MCU. But, what data is fed to this module by the ST16 is another question... Also, as this kind of behavior is not seen previously with the PX4, it is pretty safe to assume, that the PX4 is also OK. Also, if there is something seriously wrong with the Thunderbird fork, this kind of behavior would have been appeared previously during all the testing, as I was able to repeat this three consecutive times before touching the ST16. This is going to be a hard one for me to track down as there is a possibility that this is something erratic caused by the ST16, and some erratic ST16 behavior such as channels getting messed up is not uncommon even in this forum.
At this point I am assuming that the ST16 was doing something stupid, especially as ~six flights ago I replaced my left gimbal (throttle and rudder), and this is why I am going to keep the firmware image available still.
I'll get back to this and your other questions as soon as I have resolved this one or tested it enough to be confident enough that the issue was a misbehaving ST16...
Be careful.
Last edited:
Similar threads
