Weird loss of throttle control.

[jmoore3274]

I was out flying today testing camera placement to find a place which I like the best. While flying the quad acted like it lost connection to the TX but fail safe did not initiate. I still had control of roll, yaw, and pitch just no throttle control. I was only flying 600 feet 181 meters from where I took off with an altitude of no more than 40 feet. Once the throttle stopped responding it did climb to about 200 feet at some points. Not far at all and I had line of sight the whole time.

Once the problem started the quad was acting funny. I had to fight it to keep it from spinning and I used this spin to keep the quad over a field so it would have a place to come down. After trying everything I could short of cycling the TX I switched GPS hold on which also turns on Baro mode then turned it off which returned some throttle control to me but it was still not responding correctly and it came down pretty fast. I checked everything and there are no loose connections so this has to be a TX or FC hardware or software issue.

Has anyone else run into something like this? Thankfully the flight was recorded and I need someone with more expertise with multiwii to scrutinize the video. I do apologize for the shaky video. The gopro was temporarily mounted till I found an angle I like. The quad was not damaged badly at all. Dented up a battery and slightly cracked an arm no big deal its a quick fix. I just want to get this issue figured out so it doesn't happen again.

Here is the video:



Quad Specs:

450 frame
4x FPV model MC2212 1000KV
10" props
Plush ESC
Multiwii Pro FC
Tactic TTX650 TX binded to Tactic RX
Fatshark attitude 5.8ghz
Zippy 3s 3000mah 30C battery

Low vibration in the frame and PID's are tuned.

Weather was 33 deg F with 8-12 mph winds. Looks like wind was stronger at higher altitude.
Quad was given about 20 min acclimation time for the colder weather.

[Noctaro]

Pew pew, what a struggle.
Your transmitter got lots of mixer functions. Is it possible that anyhow you activated one of these during flight?
Greetz

[vpb]

I also had this similar issue just 2 days ago. It happened to my tri with r-1739 version, fortunately I was hovering, tuning the copter at that time. I couldn't control only the throttle, pitch & roll & yaw still ok. The copter wanted to jump high and I just did everything to make it land/crash. It's never happened like that before.

Feel like I had baro activated but couldn't turn it off, because when baro is activated, throttle control has no effect. Or maybe something went wrong, I can't reproduce that bug.

[EOSBandi]

What was the state of the checkboxes : "Ignore throttle during NAV and RTH" and the "Takeover BARO mode" ?

[vpb]

Hi,
I use MultiWiiConf on Mac, so dont see those checkboxes, but in config file I get

//Navigation engine will takeover BARO mode control
#define NAV_TAKEOVER_BARO 1 //(**)

//Throttle stick input will be ignored (only in BARO)
#define IGNORE_THROTTLE 1 //(**)

[EOSBandi]

I went through the code a couple times and I cannot see any case when the IGNORE_THROTTLE can be activated and stuck without user intent. However I could be wrong... please set IGNORE_THROTTLE to 0. Let's see that is it happens again or not

[jmoore3274]

I know this is a few months late and I noticed I never updated what my findings were and how I fixed it. After racking my brains out over this crash I eventually figured out it was the fence distance that caused the BARO to hold control for some odd reason. So I fixed up the Quad, cleared the EEPROM and disabled the fence distance and I did not have this issue again.

Just recently I started playing with the fence distance. Once again I had a similar problem and noticed that the auto pilot took control. Once the quad started coming back to me I changed the AP modes then turned off all AP and it relinquished control of the craft to me.

Even in the video you can hear the throttle cutting in and out while I was switching modes but the BARO was still taking control. At the end it finally gave me back throttle but I don't think the battery had enough juice left in it to keep it in the air.

Seems the EEPROM clear worked for me. I have not had the issue of a stuck BARO in 2.4 at all thankfully.

[Ciki]

Hi. I had the same problem you showed in the video. I was flying well when i lost throttle control and the copter started to take altitude. Pitch, roll and yaw worked but i couldn't do anything to land the quad because even if the gas was 0 the copter didn't descend.Motors started to spin on and off (i can't explain better but the sound is the same when you have high pids values that causes wobble and flips).I tried to stop the motors with a switch that i have on the transmitter (motor signals in this case go to minthrottle), turn off the radio but nothing happened.It seems to be a random problem and it came out when i installed the crius gps v3.1 with i2c gps nav board. Before that no issue, nothing. I don't know what is it but i think that it's a problem of lack of memory on Multiwii SE v2.5 (my flight controller) when you add gps line code (arduino says that can be instability problems due to small quantity of memory available). PID are good, ESC propellers and motors balanced, always fly in horizon mode, no baro or compass activated (default PIDs for them and never used). Next days i'll try to clear EEPROM, reload Multiwii 2.4 without gps lines code and fly. Any ideas to try to delete this problem?

This is my configuration:
Wooden Homemade Quad X (550 with 9x4.7 propellers)
QM2812 Motors 980kV
30A Simonk ESCs with BEC (one BEC powers all the component, one powers LEDs and only signals connected to the board)
Multiwii 2.5 SE with Multiwii 2.4
HC-05 Bluetooth Module
Crius GPS v3.1 with i2c GPS NAV BOARD connected to flight controller
3s 4000Mah turnigy nanotech 25c-50c battery
Mobius v3 camera (no connection with battery)

[edsimmons3]

Hi Ciki,

This sounds like the same issue I was experiencing - the remaining ram is very precious and you need to pay close attention to not running out!

The copter takes on a life of it's own if the ram runs out... I have had RTH activate randomly, not due to the fence distance but due to corruption of the ram.

Do you have VBAT or POWERMETER enabled? If so, try the settings in config.h that are shown on this thread: viewtopic.php?f=16&t=6712 (make the changes to PSENSOR_SMOOTH etc to reduce the amount of ram that is used).

Hope that helps...
Ed

[Ciki]

Hi and thanks for the reply. I have no VBAT and POWERMETER enabled because i haven't FPV goggles (i think to buy them in december), so i don't need these informations (i have a little lipo tester on my quad that control the battery voltage and rings when it goes down 3.5 v). I thought that BARO took control but it's strange that it activated without my intent and without throttle control. Unfortunately i don't have telemetry (it is coming from china) because it would have been interesting getting an eye on fc data. I don't know enough about arduino code but maybe someone can explain if there is the possibility to switch the gps code on i2c gps nav in order to free flight controller ram. It's only an idea .

[edsimmons3]

I haven't had the time to properly read the code that deals with the altitude hold and throttle stick.... I would like a proper explanation of what the throttle stick does in the baro mode as the wiki seems really sparse on details in this regard.

All I've seen that I can remember clearly is an instruction to 'not touch the throttle any more' while the baro mode is enable. Surely this should be elaborated on in the docs and the actual behaviour described?!

I will update the info myself once I have a wiki account... *nudges Alex*

[edsimmons3]

Try the config file here (http://pastebin.com/D6XUR5xw) and ground test your quad with the props off or just powered by USB.

Arm it using the TX once it has a GPS lock with more than 6 satellites, simulate flying it, try all the different modes for a while...

Check the behaviour of the GUI is not erratic, watch for resets of the GUI. Repeat all this testing with the battery attached and props off, arm and give some throttle to see how everything works.

Do you have bluetooth for your FC? EZgui on android is excellent for setting parameters and all the pre-flight checks.

[Ciki]

Yes. I've HC-05 Bluetooth module installed on the quad and EZ-GUI on my phone.Thanks for the code, I'll try when i'll be at home.I'll post the test results as soon as possible. I red the code quickly and i found quite the same uncommented lines (quadx,HK_Multiwii_v2 as board). I saw that MINThrottle uncommented is 1250 (i use 1150) and i have to change the mag declination and the gps baud (it's 57600 on the code but mine works at 115200). I also have to compare them for other changes that i don't remember. Which are other changes?

[edsimmons3]

First, don't worry at all about GPS_BAUD 57600, this isn't used if you have the I2C GPS board. If your GPS is working with the I2C board, you don't need to do anything to that part.

The main settings are:
I2C GPS (only allows RTH and poshold, no waypoints, see below)
USE_MSP_WP *disabled* due to memory contstraints
Failsafe enabled
Headfree enabled
Normal RC setup
Aux pin 2 enabled on pin 8

Feel free to change the minthrottle etc but the modes that are enabled (eg headfree and the I2C gps) use up as much ram as I'm comfortable to risk using!

I have this config file in a copy of the multiwii code that I've made a few changes to (including adding in EOSBandi's code for checking ram usage) and I've tested this to make sure it doesn't run out of ram.

You might want to choose different settings for these to suit your preference...

Code: Select all

// copter faces toward the navigation point, maghold must be enabled for it
#define NAV_CONTROLS_HEADING       0    //(**)
// true - copter comes in with tail first
#define NAV_TAIL_FIRST             1    //(**)

[Ciki]

This is the video of the crash. After few seconds motors started to make a strange noise and i lost throttle. I tried to land the copter which wanted to take altitude even if the gas was 0.
https://youtu.be/43gDpp50PWM

[edsimmons3]

You need to make the video public for me to see it

Thanks!

[Ciki]

Done!! Sorry!! ☺☺☺

[edsimmons3]

Looks and sounds pretty much like my incident... although my quad looked as if it was going to fly at me, it yawed to face me and I crashed it for my/its own safety...

One further suggestion... please add a good quality capacitor (eg 220uF or 330uF 16v electrolytic with low ESR) to the main 5v of your flight controller.

There are settings in the 328p hardware that control things like brownout detection, this should reset the IC if the voltage drops to a level that could cause corrupt ram. I think some flight controllers are lacking in power supply capacitors and the ram can be corrupted by a power glitch of really short duration. Adding this capacitor will help prevent these brownout conditions that *might be* occurring.

Since it's pretty much impossible to hook up an oscilloscope to a flying multirotor, we have to fix/eliminate possible errors until it doesn't occur any more...

Because GPS hold mode has the option to take over alt control (enabled by default) if your controller somehow flips into GPS hold/home modes by accident the altitude control will be handed to the flight controller....

[FightingFlight]

I'm just curious, when you run avr-size on the multiwii elf file how much bytes of data and bss do you have?
I know it's not the real way to tell but just something that will give us a ball park figure.

I'm also going to be adding i2c gps to my multiwii with the baro and mag, it's 80 data and 1669 bss so getting to close to the limit

[edsimmons3]

With my now stable config I get 82 data and 1738 bss.

With my previously dodgy config (which I compiled just now to see....) I get 82 data and 1808 bss. This dodgy config spuriously "does stuff"... it is not to be trusted!

The test I carried out on these two configs revealed that *no* ram was free after a while on the dodgy one, whereas the now good config gives no less than 30 bytes untouched ram.

[Ciki]

Hi ed,
I've just finished to try your config.h file in the conditions you suggested. Nothing!!! I tried again my configuration and nothing happened. Quad on gui seems to reflect each stick movements perfectly. I don't know how happened in the field but tomorrow and next days i'm going to fly without gps code in order to see if i'll have problems. I also thought that a magnetic interference reset fc but it happened only after i installed gps. I haven't added a capacitor on 5v line yet as you suggested but i'll do it as soon as possible.