Beta Testers requested to fix Compass Issue

[BigBadFun]

Hi Rob - yes absolutely.

Rotate compass to factory position.
Plug in. Power on. Applied B2.
Power off/on.
Advanced Calibration.
Calibrated controller sticks.
Set gains to 120-120-120-180 / 180-180
Power off. Go outside. Power on.
Calibrate compass (dance). Go.

 

[BigBadFun]

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

Hi Ian - need to do further testing.... but I did notice subsequent flights after brief power/on/off cycles using same battery (not removed) that initial hook behaviour seemed different. In fact I think it over-compensated a touch and hooked the other way ever so slightly?! Charging now and will try after battery removed for a few hours.... hmmmm... :geek:

 

[John Shaw]

I have reviewed my videos from today and still have the same impression as I stated earlier today - maybe even a shade better. Even if the "learning" is a short term memory thing it would be great. Long term memory - even better. I may have to raise my "satisfaction number" from 80% to 90%. I just hope I haven't fooled myself somehow. "Eagerly awaiting confirmation or rebuttal.

 

[John Shaw]

John,
What battery levels do you have set up for your 1st and 2nd. Level warnings?
What was the battery level when it went into a RTH mode?

Testers:
What was the reason and outcome for boosting the Altitude Gain to 200%?

I had to go check
Low Battery Level Warning is 20% - it had been 30% and I didn't change it, it must have changed with the load
Critical Battery Level Warning is 15% - as it has always been.
I can't remember exactly when it RTHed. I do know that I did note that I was expecting the audible alarm when we went under 30% and that didn't happen - we were even checking we had the volume turned up on the tablet. So we were well under 30% when it RTHed but I can't say if it was at 20% or 15%

After I finished my test runs with 3.05b2 today I decided to video while climbing to 400 feet. About half way up I thought to check battery percentage and it showed 22%. No warnings at this time but I figured time to head for the deck. As I was coming down the audible alarm went off on the phone and when I checked it was at 19%. It was getting ready to head to the takeoff point and I checked to verify I had no control. I flipped the right switch to ATTI and back to GPS. I had regained control and I continued on down and got into a low hover and waited for it to start down on it's own. When this happened I caught it with left stick up and noted the battery was at 14%. I held it off until 7% and landed.
So that matches the setting I saw in assistant, 20% and 15%.

 

[ianwood]

Yup. Wanted to check with you guys before suggesting it but it appears to remember the previous adaption. That seems to be new in beta 2.

 

[KassioLM]

What is the max descent speed on this beta firmware?

 

[damoncooper]

What is the max descent speed on this beta firmware?

I didn't try to push it (VRS is not my friend) but it but still appears to be 3m/s.

 

[sar104]

Just to repeat an observation that I found when testing b1 response as a function of available satellites - the initial error and rate of learning did appear to be quite sensitive to that parameter. So in trying to detect whether the FC is remembering what it learns between power cycles (or even just assessing the performance of the beta) by observing initial power-on behavior, I suspect that this should be taken into account and I think it may be responsible for some of the variation in performance that we see.

 

[Xrover]

The test results of 3.05b2 seem to be getting a bit confusing (to me)

1. Some folks are flying the Vision and some non-Vision. Signatures and posts might not include which P2 you are testing on and how the results might differ between P2 Models.

2. Many testers are using different gains settings. Defaults could be different between Vision and non Vision.
I can see some of the non Vision testers are using different gains and not the 3.05b defaults. Wouldn't this skew the results?

 

[droneranger]

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

I did some additional testing near sunset yesterday and I can confirm that the J-hook returns when you power cycle the aircraft.

I also noticed a marked altitude decrease upon "release sticks" after a forward run. This is consistent during the entire flight time, well after the J-hook has damped. It is usually an altitude drop of 1-2 meters, but it was definitely not there before the update to 3.05b2. I should note however, that I was running very old firmware (1.08) before the update to the beta, so this behavior might have been present before 3.05b2.

I intend to do more thorough testing this weekend and try to confirm (or not) these initial results. Should we be doing a basic or advanced IMU cal prior to testing? My IMU readings are nominal and according to DJI a cal is not needed. Also, any recommendation on gain settings for testing?

 

[macheung]

That 1-2m of altitude drop can be real bad for those that do high speed low altitude runs, especially across water.
Was there options to adjust max tilt angle and descend rates in the assist for v3.05b?

 

[BigBadFun]

I also noticed a marked altitude decrease upon "release sticks" after a forward run. This is consistent during the entire flight time, well after the J-hook has damped. It is usually an altitude drop of 1-2 meters, but it was definitely not there before the update to 3.05b2.

Exactly what I reported too. I have same setup as you. I christen it the 'ramp' and wouldn't be too good over water indeed! Mine is less than a metre though. Didn't use default BETA gains as I've heard they are too high. I use 120-120-120-180 / 180-180 Works great.

I tried different combinations on that Vertical Gain setting. 220, 120, 100 - couldn't see any noticable difference so set it back to 180.

 

[BigBadFun]

Was there options to adjust max tilt angle and descend rates in the assist for v3.05b?

No. We apply the BETA using same v3 software everyone has so unless they release new software, no change to options.

 

[Xrover]

I had reported the Altitude drop as well as inconsitent altitude readings back on test of the 1st version of 3.05 Beta. This bothers me as i fly fpv and relied on using the verticle rate reading on the the osd. I could look at the P2 and could visually see it was dropping but the osd would frequently show it was not changing Altitude even though it was.

 

[Andy T]

Phantom specs for hover accuracy are 0.8 meters vertical and 2.5 meters horizontal. With aggressive driving I imagine those accuracy numbers might be a bit greater?? It all sounds great to me. Release the software this week. Stomp on the brakes in your car next time and the nose will dive down and it might skid sideways but that doesn't mean there is anything wrong with your car.

 

[ianwood]

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.

 

[damoncooper]

I also noticed a marked altitude decrease upon "release sticks" after a forward run. This is consistent during the entire flight time, well after the J-hook has damped. It is usually an altitude drop of 1-2 meters, but it was definitely not there before the update to 3.05b2.

Exactly what I reported too. I have same setup as you. I christen it the 'ramp' and wouldn't be too good over water indeed! Mine is less than a metre though. Didn't use default BETA gains as I've heard they are too high. I use 120-120-120-180 / 180-180 Works great.

I tried different combinations on that Vertical Gain setting. 220, 120, 100 - couldn't see any noticable difference so set it back to 180.

Observed as well.

 

[robby666111]

Finally tested 305b2 tonight.
First impression (having not been a b1 tester) is WOW, this is such a huge improvement on 3.04! No TBE. Got a slight J-Hook on my first forward run. 2nd run a bit less, then completely gone on the 3rd run!
I also noticed the ramp-effect when moving to a hover.
I powered off and on again, and the J-Hook was still gone. Repeated the power off/on a couple more times, and it appears that the J-Hook did not return.
I will try again in the morning, and see if it is still gone. I guess there is a chance that with different satellite orbit arrangements it may need to re-learn. But at least for a flying session across 2-3 batteries it could be that we only see J-Hook on the first couple of runs then we are good.
Bit more info from DJI on this and the ramp would be good. But I for one am pretty stoked with what I've seen.
Next thing for me to try is a GS mission. I use the datalink (not the new app) based groundstation, and with pre 3.05 the recorded video was always off center as the copter "crabs" along the course (ie. it wants to fly crocked). So I'm hoping that now the copter knows how to fly straight this will also be fixed! That will be a major win for me if this is fixed.

Cheers and good night.
Rob.

 

[wkf94025]

I am very curious about the altitude variations, particularly if it is confirmed to be new since 3.04 and before. I was under the impression altitude was managed barometrically. While I can easily imagine how fixing one thing can break another, this one is a bit of a head-scratcher for me.

Origin aside, I will take vertical wandering over lateral wandering and/or yaw errors almost all day long. The one exception being when filming surfing up close, when a reliable floor is critical. Dang.

Kelly

 

[sar104]

I am very curious about the altitude variations, particularly if it is confirmed to be new since 3.04 and before. I was under the impression altitude was managed barometrically. While I can easily imagine how fixing one thing can break another, this one is a bit of a head-scratcher for me.

Origin aside, I will take vertical wandering over lateral wandering and/or yaw errors almost all day long. The one exception being when filming surfing up close, when a reliable floor is critical. Dang.

Kelly

It is barometric, but complicated by the pressure at the sensor being somewhat affected by airspeed, which has to be corrected for. I'm pretty sure that is why the altitude instability is most noticeable during rapid changes in speed.