Beta Testers requested to fix Compass Issue

[d4ddyo]

Do we still have a forward tilt/airspeed limitation?

 

[damoncooper]

Airspeed in GPS mode at least yes, not sure about tilt

 

[ianwood]

Altitude change is really only bad when you drive it hard and then let go. If you're doing lighter operations, it's not at all a big deal. At least for me. I've shot a bunch of sites with the new gain settings and it hasn't been an issue.

Airspeed is limited only indirectly by the tilt limitations.

 

[droneranger]

Altitude change is really only bad when you drive it hard and then let go. If you're doing lighter operations, it's not at all a big deal. At least for me. I've shot a bunch of sites with the new gain settings and it hasn't been an issue.

Airspeed is limited only indirectly by the tilt limitations.

Still wondering if an IMU cal is recommended before continued beta testing and what gains are best to use. Any info would be appreciated.

 

[nzvideoguys]

The altitude drop started for me with beta 1 after I hit the default button on the gains page. Beta 1 with the default gains from 1.08 didn't do that from what I can tell. The only other firmware I had was 3.02 and I didn't hit the default button then.

Can someone post the 1.08 gains please?
Cheers

 

[BigBadFun]

Still wondering if an IMU cal is recommended before continued beta testing and what gains are best to use. Any info would be appreciated.

ANY change to Phantom whether it be firmware, new gimbal or other additions, or even after a heavy landing - always do an advanced calibration.

 

[John Shaw]

Has anyone noticed any decrease of the initial hook on subsequent flights/batteries? Or does it appear the same each time?

Hello Ian et. al.
After reading this note and spending a couple of hours kicking myself for not thinking about this when I was testing yesterday, I decided to go take a look. My able assistant is "on the road" and the temps are around 87degF. I know in LA you consider this a cold snap, but here in the Pacific Northwest it's considered HOT"

I am a bit confused but I'll give you my raw impression...

Yesterdays testing went through the normal learning sequence and I did fly down 2 batteries. So on my first run today I watched to see if it remembered anything from yesterday. My first reaction was no - There appeared to be left drift but there was no JHook. I did a few more runs and it seemed to go generally through the usual cycle but more slowly. I tried CL and noticed some minor drift. I landed, shutdown, pulled the battery, and waited a few minutes.

I put the battery back in and powered up. My first reaction was it was over compensating on drift and went slightly right but again no JHook. The next few runs were pretty well behaved - no drift - no JHook. Again I shut down, removed the battery, and waited.

I did this cycke a few more times and it did look like it was learning. Later runs went straight right from the first run (no drift) and had no JHook. and when I tried course lock with the takeoff lock still in effect it went straight down the line.

So I am hopeful that it is learning and the first few odd behaviors were part of the learning process. I don't know how long I should have kept the battery out as the board undoubtedly has some capacitance. I would say I was shut down for a least 5 minutes and the battery was out for 3-4 minutes.

On all runs TBE was minimal, hardly detectable, wanting to stand in some shade I often took off and landed in very confined space and there was no stress. I would then fly out over the soccer field and conduct the test.

So don't view this as absolute, but it is my impression so far. I will now watch my videos and see what else I can glean. I'll be watching to see if others see the same thing - and if I see the same thing on my next tests.

Any word from DJI on what we should be expecting?

I have quoted my test comments from yesterday for easy reference.
I have added the video of the 4th run, note...
Takeoff from a confined space with no TBE. I am sitting in the shade under some trees but with no TBE it's stressless.
First runs done soon after takeoff have no drift or JHook.
I am cautious on the returns because I am sitting under the trees and don't want to come zipping back and damage the vegetation.
There are also runs in CL with no drift, even with takeoff locked course. I should not have rotated so as to keep the line in sight - lessons learned. ( I did it to distinguish it as a CL run).

I find this very encouraging. Even if it isn't perfect, if I wasn't testing I probably wouldn't notice any problems.

https://www.youtube.com/watch?v=BeS0ZTb ... ObwC0HUbmg

 

[robby666111]

Testing this morning flew straight on first flight (16hrs since last flight). So it looks to me that the learning is remembered.
Also tried GS, and yay, the video is recording more or less center. I knew it was right as soon as she took off. It actually flew straight following the line of the course! This is a big win for me with 305.
Did quite a bit of observation wrt the ramping. I was able to pilot from an elevation and observe the Phantom flying on the same plane as me. What it looks like is when sticks are released the Phatom drifts up initially, then ramps down. I presume that the ramp down is to compensate for the drift up. But it ramps down too far.
The drift up is purely a conservation of momentum equation due to the requested stop and tilt of the Phantom during forward motion. But the ramp down is a engineered compensation. The overall change in height (drift up - ramp down) I would say is in the region of -1m (overall decrease in altitude of 1m). Not a big issue for me. And no one should be flying with 1-2m of the ground/water anyway. That is asking for trouble is so many "other" ways!

Cheers,
Rob.

 

[sar104]

Testing this morning flew straight on first flight (16hrs since last flight). So it looks to me that the learning is remembered.
Also tried GS, and yay, the video is recording more or less center. I knew it was right as soon as she took off. It actually flew straight following the line of the course! This is a big win for me with 305.
Did quite a bit of observation wrt the ramping. I was able to pilot from an elevation and observe the Phantom flying on the same plane as me. What it looks like is when sticks are released the Phatom drifts up initially, then ramps down. I presume that the ramp down is to compensate for the drift up. But it ramps down too far.
The drift up is purely a conservation of momentum equation due to the requested stop and tilt of the Phantom during forward motion. But the ramp down is a engineered compensation. The overall change in height (drift up - ramp down) I would say is in the region of -1m (overall decrease in altitude of 1m). Not a big issue for me. And no one should be flying with 1-2m of the ground/water anyway. That is asking for trouble is so many "other" ways!

Cheers,
Rob.

I'm not sure about your momentum argument. If the flight path (and thus the linear momentum vector) is horizontal then how does subsequent vertical momentum result from momentum conservation?

 

[Andy T]

Testing this morning flew straight on first flight (16hrs since last flight). So it looks to me that the learning is remembered.
Also tried GS, and yay, the video is recording more or less center. I knew it was right as soon as she took off. It actually flew straight following the line of the course! This is a big win for me with 305.
Did quite a bit of observation wrt the ramping. I was able to pilot from an elevation and observe the Phantom flying on the same plane as me. What it looks like is when sticks are released the Phatom drifts up initially, then ramps down. I presume that the ramp down is to compensate for the drift up. But it ramps down too far.
The drift up is purely a conservation of momentum equation due to the requested stop and tilt of the Phantom during forward motion. But the ramp down is a engineered compensation. The overall change in height (drift up - ramp down) I would say is in the region of -1m (overall decrease in altitude of 1m). Not a big issue for me. And no one should be flying with 1-2m of the ground/water anyway. That is asking for trouble is so many "other" ways!

Cheers,
Rob.

The ramp thing can just be a flying characteristic of the machine... if it is a consistent characteristic, then it is no different than descending quickly and the machine descending another 2 meters after the sticks are released. Nobody seems to complain about that. Lets concentrate on getting this thing released.

 

[robby666111]

I'm not sure about your momentum argument. If the flight path (and thus the linear momentum vector) is horizontal then how does subsequent vertical momentum result from momentum conservation?

Yeah your right, but I think it contributes.
My further thought process around this is (stand to be corrected).
The copter is flying forward thus the rear rotors are spinning faster than the front rotors causing a tilt (pitch). All of a sudden the copter is "requested to stop".
Now momentum will cause the copter to continue forwards, to counter act this the front rotors are sped up to match the speed of the rear rotors, thus casing the copter to come to a stop quicker than if the rear rotors simply slowed down to match the speed of the front rotors. This overall increased speed will cause the copter to begin to rise once the momentum is eliminated.
So my original statement that the rise is due to momentum is not entirely accurate. The rise is due the the problem momentum creates! IMHO
Cheers
Rob.

 

[robby666111]

The ramp thing can just be a flying characteristic of the machine... if it is a consistent characteristic, then it is no different than descending quickly and the machine descending another 2 meters after the sticks are released. Nobody seems to complain about that. Lets concentrate on getting this thing released.

Agreed, sorry for the diversion!

Cheers
Rob.

 

[Navman]

Lets concentrate on getting this thing released.

Multi-rotor physics aside, YES, Release the Frickin' thing DJI! :mrgreen:

 

[robby666111]

In short, let's get this released.
In previous posts some (including myself) have been a little critical towards those that have the beta and are flying straight, thus are not in such urgency to get this released into the public.
As a recent addition to the beta test group, I am now in the fortunate position of having a Phantom that flies straight, and it's great!
So I think we should really now apply pressure on to DJI to release this. So far I have not heard of any instances of catastrophic behaviour with 305 (crashes, flyaways, lost control etc) so it seem to me the firmware is pretty stable.
I'm not sure how to exert pressure on DJI to release, but that could be a focus of Drunken Sparrow in the coming days. I realise that I am privileged to have been added to the beta test group, and I am not going to abuse this privilege so any action should be group related and probably headed by ianwood.
Cheers
Rob

 

[sar104]

The ramp thing can just be a flying characteristic of the machine... if it is a consistent characteristic, then it is no different than descending quickly and the machine descending another 2 meters after the sticks are released. Nobody seems to complain about that. Lets concentrate on getting this thing released.

Agreed, sorry for the diversion!

Cheers
Rob.

Agreed - it was not my intention to divert but I think we have been tasked with assessing all performance aspects, so we should not necessarily ignore this. Personally I suspect that it is just a gains issue. Specifically, in order to shed horizontal speed, the quad applies reverse thrust by pitching backwards (relative to its direction of travel). As the pitch angle increases the thrust vector becomes more horizontal and less vertical, reducing the available thrust component to maintain altitude. Too much pitch and the quad no longer has sufficient lift, and descends. So, if the pitch gain is too high the quad tries to stop too quickly and sinks in the process.

I see this very clearly at 7500 ft where there is less available lift anyway. As a worst case, if I put it into ATTI mode (which has a higher maximum pitch and thus higher top speed), run it up to speed, release the right stick and then engage GPS mode, it will drop 5 m trying to stop. I have not yet experimented with the pitch gain to see if I can reduce the effect though, so just a hypothesis at this time.

 

[Andy T]

I'm glad the ramp issue came up and was discussed but I think we can probably all agree that it is a non issue? just have to decelerate a bit less aggressivly

 

[riblit]

I got a short flight in this morning in gusty conditions. I can report that the "J" hook has been to all intents eliminated although the Phantom was blown around for the first half minute or so after takeoff before settling into a hover. Takeoff in GPS mode with 8 Satellites.

I noticed a small increase in altitude then a decrease when I centred the pitch control after a forward run into the wind. I need a calm day to remove the wind variable to check into this. Apart from that, stability was good into, across and with the wind.

I flew GPS and course lock only as the wind effect on ATTI would have made any tests invalid.

Course lock is still off, even with resetting it in the air I couldn't get it to stay on top of my test path.

Phantom Vision with Taco Gimbal. Flight data recorded temperature of 54F

 

[landonkk]

In short, let's get this released.
In previous posts some (including myself) have been a little critical towards those that have the beta and are flying straight, thus are not in such urgency to get this released into the public.
As a recent addition to the beta test group, I am now in the fortunate position of having a Phantom that flies straight, and it's great!
So I think we should really now apply pressure on to DJI to release this. So far I have not heard of any instances of catastrophic behaviour with 305 (crashes, flyaways, lost control etc) so it seem to me the firmware is pretty stable.
I'm not sure how to exert pressure on DJI to release, but that could be a focus of Drunken Sparrow in the coming days. I realise that I am privileged to have been added to the beta test group, and I am not going to abuse this privilege so any action should be group related and probably headed by ianwood.
Cheers
Rob

The best way to exert pressure is to simply anonymously leak the beta. If DJI even notices (which I am almost positive they wouldn't since they clearly don't monitor these forums anymore) it will simply force them to release it publicly a little sooner. I know that this idea will draw the typical harsh and negative response from those 'in charge', but is it really such a big deal at this point? Everyone agrees that it is a huge improvement, has no catastrophic errors, and is more than likely the version they will release anyway. So somebody just anonymously post it somewhere already and let those of us who have been involved with this thing from the beginning, but weren't lucky enough to be included in the testing, have some relief finally!

 

[robby666111]

Leaking it is not the right approach in my opinion, and that was also my opinion prior to being added to the beta test group.

 

[Pazz]

Has anyone noticed any decrease of the initial hook on subsequent flights/batteries? Or does it appear the same each time?

Hello Ian et. al.
After reading this note and spending a couple of hours kicking myself for not thinking about this when I was testing yesterday, I decided to go take a look. My able assistant is "on the road" and the temps are around 87degF. I know in LA you consider this a cold snap, but here in the Pacific Northwest it's considered HOT"

I am a bit confused but I'll give you my raw impression...

Yesterdays testing went through the normal learning sequence and I did fly down 2 batteries. So on my first run today I watched to see if it remembered anything from yesterday. My first reaction was no - There appeared to be left drift but there was no JHook. I did a few more runs and it seemed to go generally through the usual cycle but more slowly. I tried CL and noticed some minor drift. I landed, shutdown, pulled the battery, and waited a few minutes.

I put the battery back in and powered up. My first reaction was it was over compensating on drift and went slightly right but again no JHook. The next few runs were pretty well behaved - no drift - no JHook. Again I shut down, removed the battery, and waited.

I did this cycke a few more times and it did look like it was learning. Later runs went straight right from the first run (no drift) and had no JHook. and when I tried course lock with the takeoff lock still in effect it went straight down the line.

So I am hopeful that it is learning and the first few odd behaviors were part of the learning process. I don't know how long I should have kept the battery out as the board undoubtedly has some capacitance. I would say I was shut down for a least 5 minutes and the battery was out for 3-4 minutes.

On all runs TBE was minimal, hardly detectable, wanting to stand in some shade I often took off and landed in very confined space and there was no stress. I would then fly out over the soccer field and conduct the test.

So don't view this as absolute, but it is my impression so far. I will now watch my videos and see what else I can glean. I'll be watching to see if others see the same thing - and if I see the same thing on my next tests.

Any word from DJI on what we should be expecting?

I have quoted my test comments from yesterday for easy reference.
I have added the video of the 4th run, note...
Takeoff from a confined space with no TBE. I am sitting in the shade under some trees but with no TBE it's stressless.
First runs done soon after takeoff have no drift or JHook.
I am cautious on the returns because I am sitting under the trees and don't want to come zipping back and damage the vegetation.
There are also runs in CL with no drift, even with takeoff locked course. I should not have rotated so as to keep the line in sight - lessons learned. ( I did it to distinguish it as a CL run).

I find this very encouraging. Even if it isn't perfect, if I wasn't testing I probably wouldn't notice any problems.

https://www.youtube.com/watch?v=BeS0ZTb ... ObwC0HUbmg

I think your last sentence says it all. Reviewing all your videos, its not the machine learning or 3.05, its just you are getting to be a better pilot. Learning to fly & be responsive to the aircraft