endotherm
 First Officer
Flight distance : 503241 ft
Australia
Offline
|
ghettoracer Posted at 2017-7-10 11:58
I guess my question is with good solid 12+ GPS lock at the beginning of the flight why did it go into ATTI mode just seconds into the flight? Is it because of the compass error? Also what are the causes of yaw error and what exactly does that mean? If you look at the playback the first ~22 seconds I was heading towards South East on the lake and I'm pretty sure I was (trying to) flight straight ahead but according to the log I was flying sideways at 45 degree angle looking right. I wasn't doing that I was trying to replicate the path and orientation of the previous flight pointing and looking straight ahead South East on the lake...
[view_image]
There are lots of error messages contained in the software. It is hard to know why lots of them appear from time to time. As best I can tell most of these errors appear when the system can't make sense of the data it is receiving. There are many systems that need to report data and interpret it, then send it to a central brain and have that unit make sense out of it. Errors can and will occur many times per second, and need to be taken care of.
Consider the GPS sub-system. It could be reading a dozen or more satellites at any one time. They all aren't going to agree with each other. Sometimes a satellite has to be ignored for a short time or altogether if it is unreliable. The data from each is going to be different and doesn't exactly line up each time. There could be signal reflections or obstructions, weather and rain for the signal to penetrate, which affects the timing of the signal and causes error. The GPS system has to apply error correcting algorithms to work out what to keep, what to throw away, decide whether to take an average or just rely on the majority concensus of good satellites etc. There is lots of comparing, calculating, decisions on what to keep etc. In the end, a single lat/lon coordinate is the result. Many errors are discarded and ignored. That's just ONE sub-system. This data has to agree with data coming from all the other systems -- compass, accelerometer, barometer... Later aircraft like the P4 have additional redundant systems in order to reduce errors, and have better confidence in the data etc, and hopefully successfully eliminate these errors.
At times, the systems can't decide which data is correct because it is in conflict with other data which appears to be correct and is corroborated by other data. That is when it will throw up an error message. Often the numbers you are looking at in the log is correct and sufficiently accurate to perhaps locate a crashed aircraft, to within 20 feet or so. But that level of accuracy is insufficient to hold the aircraft steady if you are flying under a bridge or next to a steel tower etc. I think people panic far too much when they see these error messages, especially if it is flown in the open away from buildings/trees and other obstacles. Whenever I see errors, I know that the data currently isn't accurate and I keep an eye on it until conditions change, but I know it isn't just going to burst into flames and crash or fly away for no reason. If I am close to an obstacle, I know there is a danger of a collision because the coordinates aren't very accurate.
What are the error messages? A compass error means the compass data doesn't agree with the GPS data and the accelerometer and barometric data. Don't rely on the compass heading to be accurate to 0.1° like it might normally be. It's approximate direction will be good enough though -- if it was facing north before, you can be pretty sure it's still facing north, not spinning wildly like you are flying through the Bermuda Triangle. A yaw error is similar, it can't calculate the direction it is facing from the small x/y/z turns the aircraft is making every few milliseconds and reported by the accelerometer, that doesn't agree with the data coming from the GPS and compass sub-systems. The other systems might be reporting accurate data, but it is being degraded by bad data from one source. We can assume the resultant data isn't highly accurate and dependable, but it is still in the ball park.
These errors come and go all the time, and it is up to the values set in the firmware to determine whether you are alerted to them or not. I think they have become more frequent in recent releases which suggests the tolerance for error has been reduced. Higher accuracy is usually a good thing, but it seems to lead to more error messages being produced (perhaps unnecessarily) and a lot of pilots getting nervous. The same error was always there, it is just that you are being alerted to them more frequently.
Why do they occur at all? If you do something silly like take off from the roof of your car, or from concrete that has steel reinforcement in it, or you decided to do a compass calibration/IMU calibration whilst wearing your watch or with a pocket full of coins or while handling your phone, it can corrupt the settings. It's a bit like using a digital kitchen scale -- if you zero or reset the weight and you forget to put the bowl on it first, the readings will be all out of whack. They should all still be relative though and perhaps still slightly useful, but they are inaccurate. If you "zero" your aircraft in a bad location with magnetic properties like steel being present, your aircraft will happily report everything is fine in that environment, but when you leave the area and go somewhere there isn't interference, errors will occur and be reported. Sort of the reverse of what you expected--you haven't flown into a magnetic interference environment, but you have left one and told the aircraft that that was "normal".
What happened in your case? I'd suspect for whatever reason you had a bad calibration or your aircraft was affected by parking it on a magnetic field or whatever. You initially took off and everything reported fine. However when you flew away from that environment everything started reporting bad data and the sensors couldn't agree on who was right. We can see on the flight log everything was still functioning and recording stick movements, headings, altitudes etc. The data may have been a few feet out or a few degrees off true, but it was relatively accurate.
When your aircraft decided that GPS was not to be trusted, it changed mode to ATTI and notified you to take manual control and fly it manually, no longer relying on the GPS to hold position. When it does this, you are at the mercy of external influences like the wind. It can take you away from your position very fast making it look like a fly-away at high speed, having a mind of its own etc. It's not, it's just being blown away. This can be very confusing especially if you have not practiced flying ATTI before. It just goes to show you what a sophisticated job the GPS system is doing to hold it still and steady when you are flying normally. You miss it when it is gone.
Once again, I looked at your logs and it reported accurate records of what you were inputting and how it was reacting. It also recorded the direction you were facing. The only thing it doesn't show is the wind speed and direction, although sometimes that can be derived with some clever math based on the speed/direction of the aircraft and its tilt.
In your case, if we look at 13 seconds into the flight, we see compass and yaw errors being reported.The aircraft switches to ATTI. You were facing 146° before and after the errors occur, which is perfectly correct. There was mainly forward propulsion from you recorded, no sideways movements of the sticks apart from small short bursts and small rotations. The compass heading showed the compass orientation changed due to the (left stick) rotations, and the aircraft faced that new direction. However you still had in your mind that pushing right stick forward would produce forward movement on a heading of 146°. Instead what you saw was the aircraft moving in a different direction of about 60° and not facing that new direction. You didn't fly it appropriately as you should in ATTI, countering the unwanted direction, to make it go in the direction you wanted. You should have had a LOT more manual right stick right input. If you had been in GPS mode, that right stick right input would have been fed to the motors invisibly without your intervention and it would have remained on course. Looking at the pitch and roll of the airframe, we can see it tilt when it changed to ATTI mode, as well as when you tilted the body by moving sideways or back or forward. We conclude that the aircraft was responding correctly to your stick inputs and it did have some course influence (just not as much as you expected). The big conclusion is that there was a strong wind at altitude that came from the SW, pushing your aircraft SE. The wind might not have been apparent--sometimes the air on the ground can be quite still and it can be quite turbulent a few feet above. As it is invisible, you often don't know. You also wouldn't have known because the GPS normally counters any wind and takes the effect away. However in this instance, ATTI mode made it obvious. The fact that you were reporting 38mph shows that you had wind assistance at the time, as it is faster than the aircraft can produce by itself.
Later in the flight you decided to RTH. For whatever reason the flight systems behaved. The aircraft responded to the RTH command. RTH relies on having a reliable GPS signal to navigate back home. You had this and it came back to you under control as you expected. This would tend to indicate the problem wasn't permanent, and with no further errors reported on the leg home, it shows the systems were capable of operating accurately once reliable sensor data returned.
I bet if you take the aircraft somewhere else and do a proper recalibration following all precautions and requirements, you will find the aircraft behaves perfectly and reliably. It is frustrating when things like this happen and it is hard to identify why things went wrong. I recall examining a flight from the forum where a perfectly calibrated and functioning aircraft flew close into a harbour where there were a number of large boats docked and concluded the maritime radar from the vessels upset the air navigation systems. This is a possibility in your flight, but unlikely.
Looking at the data we can see (green arrows) that your left stick side to side movements were rotating the aircraft (yaw) as they should. We can also see (red arrows) full right stick forward resulted in a large nose-down pitch, as expected. The flight log shows the correct data for all of your movements, except where you haven't allowed for the wind and flown differently for ATTI mode flight. You should have pretty much been leaning full right stick right for the whole SE leg of the flight.
|
|
lvl.2
Captain
Core User of DJI
