Beta Testers requested to fix Compass Issue

[Captain Salty]

Prior to 3.05, I don't ever recall seeing my OSD reading anything over 1M Altitude on startup. It has always bounced between 0 and 1M. There is no reason for it not to read that. The iosd mini displays your current height above the ground (After it get's HL)

I also don't ever recall it reading anything over 2M after landing. It has also in the past constantly updated itself as I am ascending or descending. What I am now seeing with 3.05 is that I can visually see the Phantom (non Vision) Ascend or Descend by itself or with stick input, change Altitude by more then 3 Meters and the iosd mini H reading will remain unchanged.

These are my observations with my P2 (non vision) and 3.05 Beta.
If you operate something long enough then you notice slight changes in it's behavior.

I'm not the only Beta Tester to comment on the Altitude fluctuations with 3.05B
Today i paid more attention to what the iosd mini was telling me and what was actually occurring.

Only just installed my iosd mini, read 5m whilst on the ground consistently yesterday, I'm on 3.04

 

[ianwood]

DJI has taken on our observations from the first beta and are preparing a new version for us to test that they believe will give us better results. Stay tuned...

 

[KassioLM]

DJI has taken on our observations from the first beta and are preparing a new version for us to test that they believe will give us better results. Stay tuned...

Thanks for update!!!

 

[wkf94025]

I have relied on the OSD to give me pertinent info like Climb or Descent Rates and Altitudes.
I have flown very low and fast at times ... using the Verticle and Altitude info on the OSD as my reference.

Same here, including over water. Now questioning the sanity of that practice. Dang.

Kelly

 

[xtonex]

If youve never seen the iosd give the wrong altitudes then you havnt spent much time looking for it... its happened all the way since 1.08 and every firm after that. not every time but ive seen it plenty. alot of my flytrex data is wrong because of it also. its the naza giving the wrong info out

 

[SilentAV8R]

Nothing unique here about the altimeter. It is a simple barometric pressure device. Any full size pilot is well acquainted with resetting altimeters as an everyday thing when flying. The altimeter on my Futaba radio allows you to reset to zero before flying. It would be nice if that was possible with the Phantoms.

 

[sar104]

DJI has taken on our observations from the first beta and are preparing a new version for us to test that they believe will give us better results. Stay tuned...

So was the consensus from the first round that altitude is less stable in flight, yaw is occurring without stick input, and CL is still off by the same as before (2 x declination)?

I mentioned after my 3.05 tests that I felt that the approach being used for the fix is more aggressive learning rather than looking for and correcting the error(s) in how the declination is being used. I don't know if you had time to look at any of the video of those tests but, from my perspective, they are barking up the wrong tree.

The double declination offset was, I believe, the main reason for our assumption that the Naza does have built in declination tables (or a global model) but that it was using those data incorrectly. In all your correspondence with DJI, has there ever been confirmation that it does include stored declination data rather than simply learning, on each flight, by comparing compass heading and track? That question is fundamental because if there is declination correction then there is no need for learned correction except to account for crosswinds etc., so why are they still supplying fixes that obviously attempt learned corrections to the measured orientation error?

 

[sar104]

Nothing unique here about the altimeter. It is a simple barometric pressure device. Any full size pilot is well acquainted with resetting altimeters as an everyday thing when flying. The altimeter on my Futaba radio allows you to reset to zero before flying. It would be nice if that was possible with the Phantoms.

Does the Naza altimeter actually calculate or care about absolute altitude? I've always assumed that it just works in differential pressure/altitude mode.

 

[SilentAV8R]

Nothing unique here about the altimeter. It is a simple barometric pressure device. Any full size pilot is well acquainted with resetting altimeters as an everyday thing when flying. The altimeter on my Futaba radio allows you to reset to zero before flying. It would be nice if that was possible with the Phantoms.

Does the Naza altimeter actually calculate or care about absolute altitude? I've always assumed that it just works in differential pressure/altitude mode.

I think it is a simple differential pressure device. Which is why it will always be a bit off and why it would be nice to be able to re-zero it before flying. Mine usually reads between 5 and 15 feet. So I need to note that value and then subtract it from the displayed altitude if I want to know how high above the starting point the Phantom actually is at the moment.

 

[xtonex]

DJI has taken on our observations from the first beta and are preparing a new version for us to test that they believe will give us better results. Stay tuned...

So was the consensus from the first round that altitude is less stable in flight, yaw is occurring without stick input, and CL is still off by the same as before (2 x declination)?

I mentioned after my 3.05 tests that I felt that the approach being used for the fix is more aggressive learning rather than looking for and correcting the error(s) in how the declination is being used. I don't know if you had time to look at any of the video of those tests but, from my perspective, they are barking up the wrong tree.

The double declination offset was, I believe, the main reason for our assumption that the Naza does have built in declination tables (or a global model) but that it was using those data incorrectly. In all your correspondence with DJI, has there ever been confirmation that it does include stored declination data rather than simply learning, on each flight, by comparing compass heading and track? That question is fundamental because if there is declination correction then there is no need for learned correction except to account for crosswinds etc., so why are they still supplying fixes that obviously attempt learned corrections to the measured orientation error?

Are you saying CL is still off after the jhook cleans itself up? are you resetting the course after the clean up?

 

[ianwood]

Does the Naza altimeter actually calculate or care about absolute altitude? I've always assumed that it just works in differential pressure/altitude mode.

The sensor data is absolute but I am sure the Naza only cares about relative. And remember there's no way to calibrate it so it's set for standard pressure 29.92 and whatever fluctuation there is day to day just kinda happens.

I also think that the relative pressure can get a large error at power up simply because the flow of air from the blades is going to change the pressure surrounding the Phantom once you start the motors. This may account for OSDs saying your 20ft up when first take off even though you're only 3ft above the ground. Once in the air though, there shouldn't much if any significant deviations.

I agree with xrover, there is an increased vertical instability but only when you are flying it hard and making quick changes with the sticks. But it is an increased instability and not something entirely new.

Are you saying CL is still off after the jhook cleans itself up? are you resetting the course after the clean up?

CL is set with the first set of fast green lights. Whatever the compass thinks is the heading at that time will be the CL heading until you reset it.

 

[RemE]

Does the Naza altimeter actually calculate or care about absolute altitude? I've always assumed that it just works in differential pressure/altitude mode.

The sensor data is absolute but I am sure the Naza only cares about relative. And remember there's no way to calibrate it so it's set for standard pressure 29.92 and whatever fluctuation there is day to day just kinda happens.

I also think that the relative pressure can get a large error at power up simply because the flow of air from the blades is going to change the pressure surrounding the Phantom once you start the motors. This may account for OSDs saying your 20ft up when first take off even though you're only 3ft above the ground. Once in the air though, there shouldn't much if any significant deviations.

I agree with xrover, there is an increased vertical instability but only when you are flying it hard and making quick changes with the sticks. But it is an increased instability and not something entirely new.

Sounds good but I get altitude errors plus and minus altitudes, on 3 Phantoms and 1 450 at power up, no props spinning. You would think a cold boot up would yield a relative zero altitude, but it's just not the case. It is a purely relative barometer, it is not set to anything, just starts from "zero" at boot and goes from there. I do wonder if it gets biased from any vibration at power up??

Flying a model low by the OSD reported altitudes is playing with fire. I will add, that if you note the current reported altitude at takeoff, it "should" be close to the same number when you land, so just keep that bias number in mind. If it says 10ft, just remember 10ft is ground level for that flight.

Regarding vertical stability, that can be optimized with vertical gains, that function works well I think.

That could be your next crusade, fixing or adding an altitude reset!

 

[xtonex]

Are you saying CL is still off after the jhook cleans itself up? are you resetting the course after the clean up?

CL is set with the first set of fast green lights. Whatever the compass thinks is the heading at that time will be the CL heading until you reset it.

yea thats why i was asking him

 

[sar104]

Are you saying CL is still off after the jhook cleans itself up? are you resetting the course after the clean up?

No - for those tests I did not reset CL during the flight. What I meant was that the initial recorded course lock is still off by twice the declination.

 

[xtonex]

Are you saying CL is still off after the jhook cleans itself up? are you resetting the course after the clean up?

No - for those tests I did not reset CL during the flight. What I meant was that the initial recorded course lock is still off by twice the declination.

I guess what i meant was were you expecting something different?

 

[sar104]

Are you saying CL is still off after the jhook cleans itself up? are you resetting the course after the clean up?

No - for those tests I did not reset CL during the flight. What I meant was that the initial recorded course lock is still off by twice the declination.

I guess what i meant was were you expecting something different?

Yes. I was expecting that if DJI had correctly implemented declination correction using stored or calculated values then there would be little or no course lock error. If, on the other hand, they do not use stored or calculated declination values, then I would expect it to be trivially correct, since it would just maintain its magnetic heading recorded at startup. The only reason that I can see for a large CL error on startup is that the Naza is attempting to correct the magnetic heading based on stored declination values, but getting it wrong.

 

[tom3holer]

After exhaustive research and testing I have concluded how the alt readout on the OSD works I think.

Turned on the monitor in my office on the main floor about 12m above msl. Turned on the P2 and it read 0m.

Shut the P2 off and went up stairs a vertical distance of about 3 m. Turned the P2 on and it read 0m.

Shut it off and went back down and turned it on again and it read 0m.

Walked up stairs with it on and it climbed to a finish of 3-4m.

At no time was there any gps signal.

This sorta proves that on initialization the alt is set to zero as read from the baro from the OSD.

Tom

 

[ianwood]

RemE, could be the barometer is not being allowed to initialize and settle down before readings are taken. That said, it should be a fairly simple pressure transducer. So, dunno why the initial readings aren't more accurate for relative position.

Sar104, agree with your sentiment and testing method. The real fix should have the initial compass readings right on target without any adaption and testing the original CL is the best way to see that. I will say though that your results are different to mine. I need to go out again this weekend and do some more thorough CL tests.

Back to the gains, I will note what mine are when I am back home as I haven't even done that yet but I probably will avoid changing them until after I've done some testing on the new beta that comes from DJI. We all need to be sure we aren't introducing new variables in between tests that can influence the results. If you are changing your gains, please note that in your results.

 

[sar104]

After exhaustive research and testing I have concluded how the alt readout on the OSD works I think.

Turned on the monitor in my office on the main floor about 12m above msl. Turned on the P2 and it read 0m.

Shut the P2 off and went up stairs a vertical distance of about 3 m. Turned the P2 on and it read 0m.

Shut it off and went back down and turned it on again and it read 0m.

Walked up stairs with it on and it climbed to a finish of 3-4m.

At no time was there any gps signal.

This sorta proves that on initialization the alt is set to zero as read from the baro from the OSD.

Tom

I don't think there is any question about that. As Ian mentioned, it measures absolute pressure but does not try to convert that to an absolute (relative to MSL) altitude, because that is not useful information. It uses relative altitude for just two purposes: (1) to enable auto-throttle for maintaining altitude (or controlling rate of change of altitude) and (2) to relay altitude relative to the home point.

 

[Dadcat]

Does the Naza altimeter actually calculate or care about absolute altitude? I've always assumed that it just works in differential pressure/altitude mode.

The sensor data is absolute but I am sure the Naza only cares about relative. And remember there's no way to calibrate it so it's set for standard pressure 29.92 and whatever fluctuation there is day to day just kinda happens.

I also think that the relative pressure can get a large error at power up simply because the flow of air from the blades is going to change the pressure surrounding the Phantom once you start the motors. This may account for OSDs saying your 20ft up when first take off even though you're only 3ft above the ground. Once in the air though, there shouldn't much if any significant deviations.

These simple pressure sensors are also subject to error from temperature change. The barometer is located somewhere on the main board so it gets heated up quite a bit. It sets a zero altitude point at power-up. As it warms up, it takes that as a slight pressure drop, meaning altitude increase. Mine shows increasing altitude just sitting on the ground.

My Vision+ seems to end it's first battery showing an altitude of about 15 feet or more on the ground. Second battery always seems better, but I've only flown a second battery 9 or 10 times and haven't paid that close attention.