Beta Testers requested to fix Compass Issue

[sar104]

So, it seems we are at about 80% of a solution with some caveats. Higher declination areas are still seeing more than others. Sar104's results still need to be factored in but they seem to be much lower than 80%. I need to retest course lock. The one time I tried it, it worked pretty much fine. And I'm having yaw issues.

On my yaw issues, I just started editing some of my footage from the weekend. I am definitely seeing some new un-commanded yaw that's not just occurring during hover. It seems to be coming from the FC as the gimbal is not compensating for it. This is a bit of a concern as it is like having the limitation of a 2D gimbal all over again just in a new way.

I re-ran my tests again this evening with a more careful setup - specifically preparing a completely level surface for startup and also with a magnetometer survey to check for local magnetic field anomalies. I also made reasonably precise bearing measurements in both GPS and CL mode, and I think I have reasonable video, although I'll have to work on that tomorrow morning. The bottom line, however, is that I can confirm my observations from yesterday, including the uncommanded yaw at takeoff, which appears to be equal to the local declination to the nearest degree or so, the somewhat random reappearance of the J-hook after the initial learning phase cleared it up, and a persistent course lock error of twice the declination.

I also noticed the distinct lack of altitude control in forward (or backward) flight that was mentioned by a couple of others.

I'll reiterate that I do not believe that this is anything other than a patch to improve (speed up) the AP learning algorithms, and the continual recurrence of the double declination offset seems to indicate that DJI have been unable to locate the fundamental source of the problem, which is really rather mystifying. I simply cannot imagine how this can be so hard to find and fix.

I'll post some video tomorrow morning that hopefully will illustrate these effects. On the plus side I did get to play with the beta version of the Lightbridge iOS app, which seems more responsive than the Android version that I'm running on a Nexus 7.

 

[Xrover]

I hope DJI is still following this thread. And i also hope they are working on a real solution to the problems. I have yet to see a firmware update from DJI that didn't create more problems then it claimed to fix.
This whole thing is outrageous and reminds me of something out of the Keystone Cops!

:x

 

[nzvideoguys]

Sorry to distract the issue, but for the last two days, I've been getting the message below many times when I try to log in to a thread on Phantom Pilots. Anyone else getting this?

General Error
SQL ERROR [ mysql4 ]

User dbo445254094 already has more than 'max_user_connections' active connections [1203]

An sql error occurred while fetching this page. Please contact an administrator if this problem persists.

 

[spudgun]

Yes I am getting those errors as well.

 

[ianwood]

Site issues: viewtopic.php?f=4&t=17561

It's a problem with the hosting company 1&1 which is a discount provider. A whole bunch of their customers have been getting shafted for the past several days (look at their FB page). I've offered to pay for a month's hosting if we move the site to a better host. Suggest others consider donations as well.

DJI is reading this thread (if they CAN read it). That's been part of the plan that the testers report back here and DJI engineers will review the observations and videos, etc.

And general reminder that this is a beta or test firmware. That there are issues is disappointing but that is also why we are testing it in the first place. Remember that DJI has no way to test it locally as their local declination is almost zero in Shenzhen.

Back to the matter at hand, I'd like to know who else is seeing un-commanded yaw. Sar104 is seeing it. I am seeing it.

 

[wkf94025]

I also find sar104's report very discouraging, and agree with xrover's last post re DJI incompetence in identifying a clear and consistent bug handed to them on a silver platter months ago. Thanks sar104 for your rigorous methodology and reporting.

Kelly

 

[damoncooper]

Also concerned with the issue reported several times about the inability to maintain altitude on its own. I experience this on 3.0.4 as well until it "cleans itself up" 7-8 minutes into flight.

 

[SilentAV8R]

I am still interested in the gain settings everyone is using. No idea if it means anything, but perhaps it does, especially the yaw and vertical gains.

 

[Andy T]

I simply cannot imagine how this can be so hard to find and fix.

The guy that is working on this probably has about 10 other projects on the go so he gets to look at this for an hour or so a week.

 

[ianwood]

Editorial:
This thread needs to stay focused on observations of issues specific to 3.05 and compass declination fix. Legacy issues from earlier releases are only relevant to this in the context of how they interplay with the declination issue fix in 3.05.

Also, I get the disappointment people, but this is why it is called "beta testing". In a clean solution with no legacy issues, this should be a simple fix. For DJI, they may have implemented things in their FC in such a way that makes it less straightforward. We simply do not know. Everything looks easier from the outside. Please also remember this is emerging technology. None of this stuff has been around long enough to be foolproof. Welcome to the bleeding edge of technology.

Which takes me to the next point: As long as DJI continues to work on fixing this issue, complaints about them not doing it quickly enough need to stop. If they show us they are making a real effort, I am not going to complain and neither should you. If you must complain, post it in another thread. This thread is focused on finding the solution only. Capiche?

Back to the working on the solution:
It is worth reiterating that my test flight was encouraging. By the 2nd pass, it was clean and straight. Compare just the first passes on my flight without the fix to the one with the fix. It's a big difference. It's not just the speed of the learning. Something else has been adjusted. It isn't 100% right as there is still a small adaption taking place but it is very quickly achieved (by the 2nd pass).

Issues:

• Compass still slightly off at first, reported by all. I think this is minor compared to before but it would be better if it was 100% correct instead of 80% correct.
• Un-commanded yaw, reported by ianwood, sar104, xrover (anyone else?). I didn't notice it at first but my video from this weekend has a fair bit of it. That needs to be corrected before this is a viable solution for me. It appears to be an FC issue as opposed to a gimbal issue as the gimbal is not compensating for the yaw.
• Altitude fluctuation, reported by xrover. I will contend that this has always been there to some degree. Maybe it is more of an issue now but I've always seen a drop or rise when doing hard forward/backward and letting go.
• CL still off. This appears to be inconsistent, reported by sar104, nzvideoguys (anyone else?). Almost as if some of the compass samples are being declination corrected and some aren't and depending on which one gets used for setting CL, you're flying straight or at an offset. I am going to do more controlled testing this weekend and report back the results.

My other concern is why sar104's results seem less promising than the rest. Altitude? Who else is at or near 8,000ft? Hopefully the video will shed more light.

As for gain settings, I haven't touched mine, ever.

Original beta testers with the OSDII loggers (riblit, onfourblades, etc.), can you send me a DAT file with a straight forward, pause, straight back, pause from the very beginning of the flight.

For direct comparison:

 

[Dadcat]

As for gain settings, I haven't touched mine, ever.

Ian, I believe the question is about what the gain settings are now.

It was reported some time ago that the default gains were changed in 3.04. Were they changed again in 3.05? Were the gains changed for the 2nd set of 3.05 beta testers?

Remember, DJI asked Onfourblades to experiment with different gain settings. He reported the results in this thread, I think.

The uncommanded yaw could just be the result of a gain set too high or low. Not necessarily a new algorithm. Doing things like adding gimbals, go pros and iOSDs and such are reasons for adjusting gains. Maybe you always had a slight yaw issue from not touching the gains when adding all the cool stuff to your Phantom 2 and didn't notice it until now that you are analyzing the data so closely.

Only way to know any of this is if people report what the gains are.

 

[RemE]

One comment on gains. When you apply a new firmware update, often DJI does change the gains. What I've seen is that the model's gains do not automatically change to the new gain settings. You must go to the gains page and click on "defaults" which then changes the gains to the current firmware defaults.

My P2 went from stock;
120, 120, 120, 200
220, 220

To current defaults (v3.04)
125, 125, 120, 140
260, 260

It probably is a good idea to report gains, and high altitude may need a little tweaking.

 

[SilentAV8R]

As for gain settings, I haven't touched mine, ever.

Ian, I believe the question is about what the gain settings are now.

It was reported some time ago that the default gains were changed in 3.04. Were they changed again in 3.05? Were the gains changed for the 2nd set of 3.05 beta testers?

Remember, DJI asked Onfourblades to experiment with different gain settings. He reported the results in this thread, I think.

The uncommanded yaw could just be the result of a gain set too high or low. Not necessarily a new algorithm. Doing things like adding gimbals, go pros and iOSDs and such are reasons for adjusting gains. Maybe you always had a slight yaw issue from not touching the gains when adding all the cool stuff to your Phantom 2 and didn't notice it until now that you are analyzing the data so closely.

Only way to know any of this is if people report what the gains are.

Precisely why I was asking.

 

[ianwood]

I was naïvely thinking I would never have to get involved with the dark art of adjusting gains. Looks like that is about to change.

From all that I've read about people adjusting their gains, it seems like a lot of guesswork as opposed to actual tuning based on knowing what each setting does. It reminds me adjustable suspensions on cars. 9 times out of 10, when people monkey with them, what was perfectly fine becomes horrible and the person who did it convinces themselves it's somehow better.

All that said, can someone give me a condensed education on gains? How do they work? What's their purpose? Why do end uses even have the option to change them vs. being hardcoded?

 

[John Shaw]

Hover Test
I am slightly southeast of Seattle Washington USA
The local declination is 17 degrees.
I have a Phantom 2 Vision in NAZA mode
Currently using 3.05
I have never changed gains

After reading some of the discussion I decided to run a quick test to contribute some data and take a look for myself.
The weather today is "ominous". Rain clouds are looming and it is a light but gusty breeze. I just went out on my deck with a fully charged battery. My deck is a wood structure and composite deck surface, the railing is also composite, the only metal is the normal nails and screws. It is surrounded by the house and assorted deadly killer trees probably varying between 10 and 30 feet tall. I took off from the 5.5 inch deck rail that is about 8 feet off the ground (I have become comfortable doing that) and hovered at about 12 feet. I flew it out over the lawn strictly using pitch and roll (no yaw) and just let it sit there until the battery was down to about 33% which would have been about 16 or 17 minutes.

Observations
1. It moved around in about a 2 to 3 foot cube (referencing the ends of the arms) and I could correlate movement to the gusts/turbulence I was feeling. I could also hear it working to counteract the gusts. All movements were very slow. Nothing sudden or scary. It was only about 15 feet to trees on either side so I was paying close attention.
2. I never saw any yaw movement.
3. I never saw any altitude movement that seemed out of place. It only seemed to be responding to gusts. It is a barometric altimeter and I expect there are some pressure changes in gusty conditions - but I never saw anything that seemed unexpected.
4. Consider that I wasn't doing any maneuvering. I just wanted to see it hover without control inputs. This obviously doesn't apply to things that might happen while maneuvering.
5. Landing on the deck in fairly small area between tables was smooth. I can't launch from the deck because I never get enough satellites (that's why I use the rail which is further from the house) so I suspect it dropped in to Atti mode as it seemed to drifted a bit when just above the deck. The landing was not part of the test - but seemed like a good thing to do at the time.

Additional observations
1. I have never changed gains and never felt the need to change gains.
2. I was worried about the speed limits that were discussed due to a decreased ability to counter winds. In my testing with 3.05 it seems as fast and responsive as ever. I probably don't have an aggressive control style but I make fast runs at the soccer field. Anecdote: while flying at the soccer field I heard the passing observers comment "boy that thing is fast". I have never measured the speed so this comment is just my impression.
3. My main negative is that initial Course Lock is wrong. It is easy to reset shortly after takeoff - but it is currently my top short coming.

 

[SilentAV8R]

There are two types of gains on the Phantom:

Basic and Attitude

Here is what the Phantom Assisstant App has to say:

What is the Gain

If Basic Gain is too large, you will find the aircraft oscillating in the corresponding direction. If too small, the operation and stability will not be so good or control could even be lost.

Attitude Gain determines the reaction speed of attitude with command stick input, the larger the value the quicker the reaction. Increase it for sharper and quicker leveling action.

Why Gain Adjustment is Necessary
Different aircraft have different gains, so that you can adjust the Gain manually according to your aircraft to have a precise flying experience.

Tips
If the aircraft parts are not compatible, you cannot get good flight experience by adjusting the Gains.

So think of it this way, the basic gain is what works to keep the Phantom level and at the desired altitude. Increase the weight and you will likely need to increase the vertical gain. So when I see the reports of difficulty holding altitude I wondered about the vertical gain. Wandering around in hover can be due to too low of basic gain on pitch and roll. Failure to hold yaw, well, too low on the yaw gain.

So the way to tweak is to start with the defaults and then increase each one (I do the pitch and roll together) until it gets "twitchy". Then back off a few percent.

Vertical gain is easy. If it holds altitude, then it is likely close to perfect. Same with yaw.

The Attitude gains are sort of like dual rates on a regular radio, or maybe expo is a better analogy. Jack up the attitude gains and the Phantom will react more quickly to stick input. Lower the attitude gains and it will feel "softer" on the controls.

Anyone who has played with flybarless helis will know instantly what I am talking about. Most Phantom folks seem to forget or ignore that it has an advanced 3-axis gyro system in it which is what keeps it level and in place.

Make sense??

BTW, mine is currently running 125, 125, 120 and 160 on the pitch/roll/yaw/vertical Basic gains. I am using the default 260/260 on the attitude gains. Feels nice a crisp and holds position, etc. very nicely.

 

[Dadcat]

I was naïvely thinking I would never have to get involved with the dark art of adjusting gains. Looks like that is about to change.

From all that I've read about people adjusting their gains, it seems like a lot of guesswork as opposed to actual tuning based on knowing what each setting does. It reminds me adjustable suspensions on cars. 9 times out of 10, when people monkey with them, what was perfectly fine becomes horrible and the person who did it convinces themselves it's somehow better.

All that said, can someone give me a condensed education on gains? How do they work? What's their purpose? Why do end uses even have the option to change them vs. being hardcoded?

I would love to get the same education. Maybe in a thread for that purpose with a link to it here.

What I think you should do now is ask all the beta testers to record what their gains currently are. Then puch the "Default" button and see if they change. And ask them to report here. I'd be surprised if the gains are not in the iOSD data the first group of beta testers sent in to DJI.

 

[Xrover]

I am going to do more testing today on 3.05. Specifically playing with the Altitude and Yaw Gains.
The Default 3.05 Gains Are:
125 125 120 140
260 260

I will report back when I am finished.

 

[John Shaw]

Per request/suggestion in this thread...
I have never updated my gains. (at least nothing I am aware of)
Stock P2V
I just checked and...

...............Pitch....Roll....Yaw....Vertical
Basic........120......120.....120....120
Attitude....200......200

I have never seen anything that motivated me to want to change them and I am happy with the current response.

 

[AnselA]

Per request/suggestion in this thread...
I have never updated my gains. (at least nothing I am aware of)
Stock P2V
I just checked and...

...............Pitch....Roll....Yaw....Vertical
Basic........120......120.....120....120
Attitude....200......200

I have never seen anything that motivated me to want to change them and I am happy with the current response.

Have you ever updated the firmware? Which version you have now?