P4 Oscillations In Flight - Cannot Find Cause

[Hands0n]

@Hands0n
6. I tried, but was unable to downgrade the firmware (no change)

Use independent app Dumldore to perform the firmware down-version change. There are several YouTube HowTo with links for the relevant firmware. Once the process is kicked off using Dumldore you can monitor progress using DJI Assistant (simply start it and then select your P4, where it will display the progress). HTH

Something else to maybe try, this came up in some chit-chat, is to re-apply the firmware V02.00.0106 using DJI Assistant - a few folk have said that has solved problems updating as well as down-versioning.

Another lad has suggested that after V02.00.0106 the VPS should be re-calibrated (again using DJI Assistant), something that I did not think to do. I noticed in my flight log output that the IMU (Barometer) and VPS were reporting different altitudes, at one point where the P4 was above 30ft altitude the VPS was reporting 0 altitude!

At this stage, I am going to fly the down-version v01.02.0602 as soon as I can and assuming all is okay I may just go back up to V02.00.0106 and this time recalibrate the VPS before flying again. It all depends on how much time I can grab to do all this work.

 

[Hands0n]

Yep same here... Spent an hour or so last evening (time that I'll never get back - lol) on the dji forums reading about the issues folks are having with their P4's post the latest update. I unboxed my P4P yesterday and after a half dozen flights one of the first things I noticed was the lack of 'rock solid' stability in hover. My only other dji is a MP and perhaps I'm too used to how it locks in. My P4P is not horrible and hopefully not as bad as folks have it in the dji thread they dont feel safe to fly but I guess I always assumed the P4P would actually be more stable due to it's size.

I have the MP as well as the P4 and before these issues my P4 flew exactly as stable as the MP. I often described it to others as if the P4/MP were stuck on top of a pole in the ground. I've built my own using various FC for several years and pretty much nothing comes as close to the stability I've experienced in these aircraft. And I include 3DR's APM and Pixhawk in that statement, although they do come damned close If you've instability and you're on V02.00.0106 it may be an idea to go through all of the Calibrations (IMU, Compas, Gimbal) and also VPS [using DJI Assistant for the latter]. Some lad on YT has said that sorted his instability in his P4 with .0106.

 

[David Cooke]

I have a mavic as well so have not flown the P4P+ much since the upgrade . . when I did the .106 last week, I carefully re-calibrated EVERYTHING. Seemed to fly ok . . did not notice anything at the time. Yesterday took it out to a working shoot trying to use tripod for a smooth dolly shot and could not get a stable speed . . in fact it seemed odd that FULL defection was not at all smooth and slow as the Mavic. Must have tried the shot 4 times to get it smooth . . .(I put it down to a bit gusty wind and turbulent with trees cars and buildings around) . . so I thought it was just me. Having read this thread now and the DJI forum notes that is exactly the case with me. . .

Let me ADD something else to the mix.

That same shoot I had DJIGO4 (latest vers) crash in-flight. It rebooted and recovered so I landed and rebooted everything. Also, I was right on the boundary of a local airport No-Fly . . not a problem, I have clearance and an UNLOCK. After unlocking and flying back and forth across the boundary it constantly (maybe 6 times) reminded me MID-SCREEN that I was in or near a no-fly zone and I needed to tap the 2 check marks and "Confirm" . . needless to say it's impossible to shoot steady 2 min takes with that popping up in your face ever 30 sec.

So I'm going to do a full test series today and see what's up . . I think I'm screwed using the P4P+ till this gets fixed
NEED A ROLL BACK . . . FAST!

 

[Hands0n]

So I'm going to do a full test series today and see what's up . . I think I'm screwed using the P4P+ till this gets fixed
NEED A ROLL BACK . . . FAST!

Using Dumldore the roll-back took just about 15 minutes. I had a strange behaviour from the Battery pack when powering down. Someone suggested re-doing the roll-back to fix it as they'd had the same issue fixed that way. So I re-ran the roll-back and did nothing else, it fixed the power down behavior.

 

[RémyHeli]

Thanks for sharing experience also thru these feedbacks. Much appreciated and valuable to this great forum.

 

[sar104]

Having looked in more detail at the complete DAT record, there are several good periods of full forward stick that are not noisy. 909 - 918 seconds, relative to record start, is a suitable example. All the diagnostics (accelerometers, gyros, magnetometers) show the oscillations, which have a period that varies in the range 0.5 - 1 second. For example:

The oscillations are in phase on the pitch and roll, and appear to be directly caused by motor speed variations. Looking at pitch vs. back/front motor speed difference (since that cause pitch):

It's apparent that the motor control is causing pitch overshoot, as evidenced by minima in the forward pitch angle being followed by increases in the motor speed differential (as expected) resulting in maxima in the pitch angle followed by reductions in the motor speed differential. This does look like a PID-type problem with incorrect tuning leading to oscillatory instability.

Exactly the same behavior is visible on the roll vs. left/right motor speed differential:

My interpretation of these data, which could be completely wrong since I don't know the intricacies of the FC control algorithms, is that it is an FC tuning problem rather than an IMU calibration problem.

 

[RémyHeli]

Makes really sense sar104. I have a TBS Vendetta and one of the basic rules with Betaflight configuration software is to reduce increased pids or loop frequencies when resulting/reaching to over-resonating situations, but generally under specific flight conditions only until not exagerated value.

Any change in this area to be tested by pilots pushing to the limits with real flight and not just just basis theoritical assumption. It's like service repair to test flight and not just send back after replacing parts. I know I can dream..

Typically over correction to level horizon will result to bouncing back and forth as too strong correction auto-input vs physical craft dynamics. Also resulting to increased esc controller load .. motors .. gimbal ..

So what? Well only DJI can answer properly to this as they own the product and I would love see them fixing the problem and not abandon P4 owners!"..

 

[Hands0n]

Thank you again @sar104 for the extraordinary depth of analysis of the logs. Considering that this firmware came out during July 2017 DJI have been noteably silent about what they appear to have done in .0106. What I do know, from actual in-field experience, is that there is no Gains tuning that I can make to combat the effect.

However, I do have a question for you. One of the chaps on YouTube had this same issue, plus an issue with 4K video on NTSC. He claims to have solved the instability issue by recalibrating the VPS, which is rather interesting and not something that I had even considered. But, looking at my flight log here DJI Flight Log Viewer - PhantomHelp.com I have noticed that there is a disparity between IMU Altitude and VPS Altitude, also I've noticed where at some point VPS says that the aircraft is at zero (0) ft altitude where the IMU is reporting above 32ft.

Could that be a hint/lead as to how to resolve? Could recalibrating the VPS really solve this? Guesswork perhaps, I've not seen anyone else offer that suggestion yet ... Not even DJI who appear to be in denial about .0106 to this day! Not helpful at all.

 

[BudWalker]

....
But, looking at my flight log here DJI Flight Log Viewer - PhantomHelp.com I have noticed that there is a disparity between IMU Altitude and VPS Altitude, also I've noticed where at some point VPS says that the aircraft is at zero (0) ft altitude where the IMU is reporting above 32ft.

Could that be a hint/lead as to how to resolve? Could recalibrating the VPS really solve this? Guesswork perhaps, I've not seen anyone else offer that suggestion yet ... Not even DJI who appear to be in denial about .0106 to this day! Not helpful at all.

With the .txt file vpsHeight unknown is presented as 0.0. The .DAT does it right by having vpsHeight unknown be "blank"

 

[sar104]

Thank you again @sar104 for the extraordinary depth of analysis of the logs. Considering that this firmware came out during July 2017 DJI have been noteably silent about what they appear to have done in .0106. What I do know, from actual in-field experience, is that there is no Gains tuning that I can make to combat the effect.

However, I do have a question for you. One of the chaps on YouTube had this same issue, plus an issue with 4K video on NTSC. He claims to have solved the instability issue by recalibrating the VPS, which is rather interesting and not something that I had even considered. But, looking at my flight log here DJI Flight Log Viewer - PhantomHelp.com I have noticed that there is a disparity between IMU Altitude and VPS Altitude, also I've noticed where at some point VPS says that the aircraft is at zero (0) ft altitude where the IMU is reporting above 32ft.

Could that be a hint/lead as to how to resolve? Could recalibrating the VPS really solve this? Guesswork perhaps, I've not seen anyone else offer that suggestion yet ... Not even DJI who appear to be in denial about .0106 to this day! Not helpful at all.

As @BudWalker commented, the zero VPS readings are an artifact of interpretation by the Log Viewer that you used. It seems unlikely that VPS has anything to do with the problem but, since we are not privy to the inner workings of the flight controller, who knows. It might be worth trying.

 

[David Cooke]

I just sent DJI two messages regarding discrepancies on my last flight near a NOFLY. Primarily complaining about the poor smoothness in TRIPOD . . . and general wabbly approaches I experienced as it fluttered and bucked for no reason or control input. Second though, it threw up the "please check these 2 boxes and CONFIRM you are authorized" dialog 5 times in the middle of a slow pass making it impossible to video anything. Then DJIGO4 app crashed and restarted spontaneously! . . there are lots of problems with 0106 . . your insights are excellent Sar104 . . thanks . . why can't they roll back . . just this once.

 

[Hands0n]

I was going to try a control test of firmware versions V1 and V2 but one lad on here and the official DJI forum (HCADAMS3) has beat me to it. There is absolutely no point me simply duplicating what he's done!

I've been posting my experiences to this and the PhantomPilots user group. It's interesting to me that it seems that many people don't have an issue with the new firmware. I wonder what the difference is between their set-up and mine.
So Far:
1) did update using Assistant and got shaky flights and other weird behavior.
2) using , I downgraded the copter to v01.02.0602 and got flight stability and other weirdness stopped.
THEN
3) I re-upgraded to v02.00.0106 using app and got the same issues as #1 above so it's definitely the firmware on the copter.
4) I am re-downgrading doing #2 again and will try to get DJI to send me the beta mentioned in post 1298# .
Here are the files I use to update my P4
and firmware link
and firmware link
https://drive.google.com/open?id=0Bwdvxz6eJBrzWmJ1dnlBMzlLNGc<br />​

Thinking that I'll just stay on .0602 until DJI issue an update to follow V02.00.0106.

What a bloody palaver.

 

[Hands0n]

As @BudWalker commented, the zero VPS readings are an artifact of interpretation by the Log Viewer that you used. It seems unlikely that VPS has anything to do with the problem but, since we are not privy to the inner workings of the flight controller, who knows. It might be worth trying.

I will check out the downgrade this weekend and if it is okay I'll take a backup of it for safety. Then I may just go give it a try to update again and then try a VPS calibration. I've not performed VPS calibration before so can spend the time between now and then studying the writings and videos on the topic.

Its also rather amazing that there are reports of P4 and other affected aircraft users not impacted by the new firmware. Which is really odd. I work with software faults (on an industrial scale) and never seen anything like this kind of variability before .... Generally it all works or it all doesn't work. DJI's randomness seems to have hit new ground ...

 

[sar104]

I will check out the downgrade this weekend and if it is okay I'll take a backup of it for safety. Then I may just go give it a try to update again and then try a VPS calibration. I've not performed VPS calibration before so can spend the time between now and then studying the writings and videos on the topic.

Its also rather amazing that there are reports of P4 and other affected aircraft users not impacted by the new firmware. Which is really odd. I work with software faults (on an industrial scale) and never seen anything like this kind of variability before .... Generally it all works or it all doesn't work. DJI's randomness seems to have hit new ground ...

It's possible that variation in components (perhaps different suppliers) has effectively resulted in more combinations of aircraft setups than DJI can test with new firmware - hence some aircraft flying fine and others dysfunctional.

 

[hunterxero]

Downgraded to v01.02.0602. Initial flights testings shows that the oscillation is gone. Range slightly increased, but that could be related to lower interference, need more tests.

Guess I didn't have to buy those motors after all...

 

[Hands0n]

Downgraded to v01.02.0602. Initial flights testings shows that the oscillation is gone. Range slightly increased, but that could be related to lower interference, need more tests.

Guess I didn't have to buy those motors after all...

Thanks @hunterxero - it is good to get more confirmation that the down-versioning to v01.02.0602 actually works.

While flying with the current problematic version I did not actually observe signal issues but when looking back at the flight log I have seen that there were in fact many recorded at one specific part of the flight ... I kept the P4 in the field just in case the worst happened ... The attached picture shows where the signal errors inexplicably occurred, although I did not seem to actually lose control.

I'm hoping for the same result as others who've down-versioned to .0602. It will be nice to get my "old" Phantom 4 back.

 

[GMack]

This is interesting.

Has anyone tried it while the P4 is hovering and bouncing around, and how the before and after look? I'd guess if it is not smooth in the data line during flight, it should also show the deviation in hover as well between the old and new firmware versions?

 

[sar104]

This is interesting.

Has anyone tried it while the P4 is hovering and bouncing around, and how the before and after look? I'd guess if it is not smooth in the data line during flight, it should also show the deviation in hover as well between the old and new firmware versions?

I did check a couple of hover segments from the flight record that @Hands0n posted - those had no obvious oscillatory behavior comparable to what was visible during forward flight. I can post the details later if anyone is interested.

 

[BudWalker]

It might also be interesting if @Hands0n could perform an experiment with inputs full right aileron, then full left aileron, and then full back elevator. Would there be the same oscillatory behavior, and would it be along the same axes.

 

[David Cooke]

It might also be interesting if @Hands0n could perform an experiment with inputs full right aileron, then full left aileron, and then full back elevator. Would there be the same oscillatory behavior, and would it be along the same axes.

I'm interested in that too . . .Thanks everyone for the really excellent feedback . . . and I'm out the door right now to do a full functional test on the flightware 01.04.0602 . . (which it says is "current") before I downgrade to v01.02.0602 as above . . see how that behaves . . . then if no change maybe back to 01.04.602 and see if anything is fixed. Will post later today.