Please select Into the mobile phone version | Continue to access the computer ver.
Zenmuse X5 Jello/Rolling Shutter (98% SOLVED) M600,D-RTK,X5
1234Next >
16359 127 2017-1-3
Uploading and Loding Picture ...(0/1)
o(^-^)o
Dmitry_R
lvl.1

Russia
Offline

Hi LT, Ricci.
No flights for long time but have an idea for LT
LT can you take the upper plastic cap off and make a photo of the connections, especially gps connector

MIND THE FAN CABLE!!! I broke it once pulling too hard
2017-2-2
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

IMG_1209.jpg IMG_1207.jpg IMG_1194.jpg
IMG_1192.jpg
IMG_1192.jpg
IMG_1207.jpg
2017-2-2
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Greetings chaps,
I mentioned before  about the Rolling  shutter having a reduced effect when the craft was flying backwards. I have studied the data for the flights and the acceleration  was about 12m/s in both fwd, and rearward direction and the shutter speed was the same so why a marked difference? My current thinking is aerodynamic oscillation.  I wont go into technical details due to overly complicating the observations however some of the dedicated testing LT has done has shown improvements in the reduction of "r/s" and for him the addition of 85g old lead strip acros the first mounting plate (previous page pictures) has yealded the best results. I have done a few alternative mods with reducing the rubber isolation movement and bracing the camera plate support arms, but I still get "R/S" when the craft transitions from one state to another, ie pitch and yaw, acceleration and desecration. The discovery that traversing backwards has less of an effect on R/S no matter how the camera is pointing  made me suspect the aerodynamics on the chassis might be causing some of the issues. LT, stiffened the top plate which is essentially a wing when traveling through the air and subject to dynamic forces such as lift,  bufferting, turbulence and not forgetting oscillation.  The lead strip could act as a damper for oscillations of movement as well as something called a airflow spoiler. As can be imagined the complexities of vibration damping on a dynamic platform are complex and the only true result is compromise because the camera design is the main cause due to the way a CMOS sensor works.
LT or Dmitry, if your flying, fancy doing a confirmation flight to see if its just my M600 which has less R/S in reversed flying directions.
The saga continues.......

Any input form the other readers very welcome, as I am sure we are not the only ones to be in this boat and maybe someone has a fix or other ideas???????

Ricci
2017-2-2
Use props
Dmitry_R
lvl.1

Russia
Offline

HI
Ricci, I see that you have a pro version, i use standard one and connections differ.

i had a problem after assembling the X5R gimbal to matrice. i followed exactly as it was on online manual
but there were no video link. few days of pain, and i took it to the dealer. they fix it.
as on the video  i plugged 7 pin cable from the gimbal to 3 pin cable on CAN port. but it is wrong. Screen Shot 2017-02-03 at 17.46.34.png

the repair guys plugged the cable to the main A3 and it worked. i paid 150 euro for the job.

i thought this info might help LT in his connection problems some how...        

D.
2017-2-3
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Dmitry_R Posted at 2017-2-3 07:14
HI
Ricci, I see that you have a pro version, i use standard one and connections differ.

Greetings Dmitry,
strangely I had no issues with the X5 because it was installed at the factory, I had the first unit to ship to Europe so I guess they have changed how they ship them now.
Its really great that you share makes my feet tingle with joy, thats how we keep the community together.
I was hoping to get the new X5s camera but I have now learnt it will only fit the inspire, bo-hoo!!
when are you flying next?

Ricci
2017-2-3
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hi chaps,
I hope to fly today to do some testing and would appreciate it if someone could verify and do the same test.
I have been working on some active dampers, I think LT was talking about them earlier, well I need to confirm how the vibrations are propagated to the cmos sensor.. If you can test by pointing the lens down to the ground  from the pointing fwd direction and noting how much R/S there is and then when the camera is pointing down rotate the gimbal yaw 90º to right or left still with lens pointing down, and note the R/S. I am trying to see if the camera sensor is less effected when 90º aposed to the crafts heading, ie fwd.  plus do the traveling rear wards with camera in both directions of travel.
hope that made sense.

Ricci
2017-2-3
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

Hi there guys. DJI is finally offering to analyse the craft. Just sorting out the details because being in South Africa poses a bit of a problem with the shipment and I'm trying to find out wether they can send me just a gimbal connector as a final test before returning the unit.

Dmitry: Thanks for pointing out the cables issue you have had. I played around with the concept in the beginning by shifting the CAN cable around between the available positions on the A3. The RTK in our installation also takes up a CAN port. How are your efforts coming on with the issue you're having?

Ricci: Great feedback. Also experienced the same with going forward and backwards. I will revert on the 90 degrees yaw and tilt combo for jello. As for the disconnect, I currently suspect that it is a tolerance issue in the gimbal connector onto which the camera connects. As for the analysis and troubleshooting DJI wants to offer, I first want to ask them to send me a new gimbal connector. If this is not the problem, I will send it in.

Dmitry & Ricci: Thank you for the support and motivation to keep going. You guys are really showing the right spirit and power of a forum community by helping others! As mentioned previously the issue is far beyond the technical scope and understanding of many people and I am really grateful that you guys have chimed in.

DJI Support: I really know you guys are doing your best and I think DJI-Jamie is one of the heroes from team DJI on the forum. Just a suggestion: If you don't know and refer something please keep the customer updated and in the loop even if it just a quick pm or update saying you're still waiting for feedback. It will save a many people a lot of grey hairs.

I'm planning a dual battery set mapping mission today and weather permitting, hopefully provide feedback later. Just started raining!

Thanks and regards all!
LT
2017-2-4
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

I've just had another detailed look at the GoPro footage of the gimbal movement. You can clearly see the rear arms dancing around which cause the gimbal plate and camera (Sprung mass) to shake a little too much which might be the cause of the jello effect. I'm still noticing this albeit a lot less than initially. Think a solution would be to link the rear and front arms together with an extremely thin plate.

Ricci: As per your comment about the aerodynamics, I think at greater speed we have a buffeting effect with oscillating eddies around the camera because of the shape (this is not really a solvable problem apart from changing the shape of the camera body) and the only solution I can think of would be to either increase inertia (adding mass somewhere) or rigidity to reduce the effect. Alternatively adding small elements to the leading edge (front of the CF plate) to increase a turbulent flow instead of laminar flow across the plate (similar to what they are doing on the leading edge of aircraft wings). This should have the effect of reducing laminar flow forwards. My suspicion is that we have a lot more turbulent flow over the plate when flying backwards because of the location of the plate with respect to CoG. Only other option I can think of is to replace the plate with a honeycomb structure to increase the stiffness.
As for the disconnects: As mentioned I feel that after adding the felt to take up the slack (which I removed after one flight) the mating tolerances have changed (might be um's) in my case which is why I'm no longer getting disconnects in flight (This is a great relief but still annoyingly irritating when you have a customer standing next to you and asking a lot of questions why you are restarting the craft so many times). They only happen with the M600 stationary with 1 exception 2 days ago.

I haven't tried  your test today because we had around 50mm of rain in the area. Will give it a try tomorrow and report back on my findings.

I'm also considering locating the RTK and Airlink to the CF plate below the M600 to shift the CoG downwards however this would require me deviating from the default antenna positions which I'm reluctant to do.

Regards
LT
2017-2-4
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Nice feedback LT,Fantastic news about DJI offering to take back the M600 and I hope they send you a test part first.
I too have been grounded by the weather and today is going to be as bad too so back to theorising, my favourite pastime. I am pleased you have noted flying backwards produces less R/S, now we just have to figure out why......
LT & Dmitry, have you tried changing the sensitivity of the gimbal? I have put the damping controls to custom and turned everything up to full, not sure if the damping is only for the operator input or maybe it is a global function applied to the gimbal motors and the operator input is damped. I will test fly to see where this adjustment fits in when the dam rain stops.I am going to strap an old iphone 4 to the mid camera plate and run a vibration app because I am strugling with reading the acceloremeters using the "Data viewer", I dont know the frequency of the sensors, there bandwith and sensertivity so it is efectivaly like looking at a a bubch of wiggles with only a Delta diferance and a trend to show the way, I guess I am spoilt with the datalogging I use. Iphone app is very good for repeatability tests.

catch you later and report back findings.


Ricci
2017-2-4
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

LT,
What are your thoughts on mounting the camera attachment plate on some closed cell foam? I am desperate to find the direction of the vibrations on the sensor and work back to the plate anchor points because the vibration are being amplified at the camera. I am also thinking of changing the rubber dampers for springs so they can be measured and modified mathematically to isolate the frequency. Main problem is shutter speed and aliasing.....tests have to be done under strict conditions if we are to find a fix. Shutter speed the same and catalog changes.
DJI you should sponsor our research???????

Ricci
2017-2-4
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

ricci2 Posted at 2017-2-4 19:27
LT,
What are your thoughts on mounting the camera attachment plate on some closed cell foam? I am desperate to find the direction of the vibrations on the sensor and work back to the plate anchor points because the vibration are being amplified at the camera. I am also thinking of changing the rubber dampers for springs so they can be measured and modified mathematically to isolate the frequency. Main problem is shutter speed and aliasing.....tests have to be done under strict conditions if we are to find a fix. Shutter speed the same and catalog changes.
DJI you should sponsor our research???????

Hi Ricci

I'm also considering a honeycomb plate instead of the carbon fiber plate. Will try to obtain a piece this week and test.

With regards to changing the sensitivity of the gimbal: I did try various settings but it only seems to change the way the gimbal responds to input from the user.

I did a Map Pilot mission again on Sunday with the GoPro focused on the X5 and bracket and again the only thing that I noticed is that the rear downward arms are vibrating more than the front ones. The vibration seems to be in the X as well as Y direction which can only be explained by the fact that the Carbon Fiber plate buckles in Z under resonance and translates this to the down-arm movements.

I also did a test as mentioned in #86 with the camera tilted 90 deg down and yaw at 90 deg - this seems to produce no jello/R/S but I only had 30 seconds to do this test before the batteries were empty.

I flew a mapping mission with Map Pilot today: First set of batteries (around 21 minutes) with absolutely no problem with camera/gimbal disconnect, the M600 then returned for a battery change and I sent it on its way again. It then completed another 10-12 minutes before the first disconnect. After this, I returned home and put the batteries on charge. After charging went out to complete the mission but had a disconnect just 3-4 minutes into the flight and gave up on the day's mapping. The mission was programmed to fly at 12m/s at 120meters and it was quite windy since after evaluating the log files on Healthydrones, the M600 was using around 15.7 amps on the up-wind runs and around 9.7 on the down-wind runs. Both disconnects happened on a down-wind run. Maybe I'm reading too much into this but I'm just trying to isolate the cause of these disconnects.

After the mapping sessions, I went on to shoot some video and I completed 2 sets of batteries without a problem (manual flying).

To conclude, I think the Jello and R/S is caused by resonance on the carbon fiber plate of the gimbal bracket. My Gimbal/Camera disconnects are caused by some tolerances being out of spec on the Gimbal Connector.

Regards
LT

2017-2-6
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hi all,
Experiment with going really soft on all the mounts and foam isolate the fixing screws too,  the reverse of the current direction because isolating the full bandwidth of frequencies is proving to be a troublesome thing to do.

I shall keep you updated.Going
ricci
2017-2-7
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Good something chaps,
I have some interesting news. I have been researching the reason why adding sticky lead segments to the first attachment board under the M600 was so successful for LT with reducing rolling shutter or Jello/aliasing.
I have just stripped all the lower gimbal plates off and the sliding guide tubes and what I discovered is a potential cause of amplified vibration. Let me be more defined here, the top gimbal plate with the rubber grommet for the gimbal and camera cables is a potential sound board, similar to the wooden board under a grand piano or the back panel on an upright, the shell of an acoustic guitar, ect, ect. How is this possible? By chance the plate is approximately 2mm thick, (weight issues) its fastened at the corners onto the tube carriers at four small points of contact which are then connected to the airframe by hollow tubes. The tubes are also forms of soundboards (listen to the blue-man group). The best way I can explain this is to liken it to tapping a guitar and hearing the sound amplified. Here's the interesting bit, the thin top plate and tubes don't produce a vibration of their own really but amplify the motor and airframe frequencies. There is a chance the plate might vibrate at its natural frequency being thin and extremely rigid which will be transmitted directly to the camera however foremost it's an amplifier.I have glossed over the subject rather and also neglected to account for the rubber isolator balls between the gimbal standoff's and the bottom plate. You might naturally assume the frequencies are high given the sound of the motors but the frequency could be low similar to a sub woofer where you feel them rather than hear them. The rubber is designed to damp certain frequencies but the science of rubber and collapsable spheroid damping vessels is another level of dedication I am not sure DJI would venture into.
So LT, added the lead strips to this potential sound board and changing the way the oscillations pass through it, effectively damping the amplification. My evidence is based on this experiment: I placed an oscillator on the craft and a spectrum analyser sensor on the top plate, I moved the frequencies up and down to see if the top plate effected the amplitude. It did and I now know the range I need to look at but I am still in the dark as to the exact frequencies that effect the cmos camera sensor, so what I have done is add sticky felt pads to see if this damps the amplified signals to the sensor, like filling the guitar with a blanket or adding lead segments.
Although the science is fascinating it will still be down to trial and error to see if this works, so more testing now.
Please feel free to chip in at any stage as this is an ongoing project and the subject will be better for it.
Onwards and Upwards!!!!
Ricci
2017-2-7
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hello fellow wobblers, I have a cure!!!!!!!

Found the cause!!!

ricci
2017-2-8
Use props
Dmitry_R
lvl.1

Russia
Offline

Hei All,

Ricci. if it is truth, next time i am in France i will get you a huge bottle of calvados!!!! (or you will be in moscow).
i had no time for flights recently but can do any tweaks or tests this weekend if needed.


2017-2-8
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

Hi Ricci

Thanks all for your efforts in trying to find a solution. Please enlighten us on your discovery.

Had a Mapping mission yesterday with around 6-7 disconnects. A 17 minute mission took me 2 sets of batteries(40 minutes) with all the RTH & restarts...

I did the trial of flying in all the directions taking footage, will process today and upload.

Regards
LT
2017-2-8
Use props
fans86368c91
Second Officer

Australia
Offline

ricci2 Posted at 2017-2-7 05:52
Hi all,
Experiment with going really soft on all the mounts and foam isolate the fixing screws too,  the reverse of the current direction because isolating the full bandwidth of frequencies is proving to be a troublesome thing to do.

G'day people.

One material worth investigation is the dynamic properties of Blu Tack.

I am NOT pulling your chain. Some machinists have found that a blob of Blu Tack on a boring bar is useful to dampen mechanical vibrations (chatter) of the tool and music lovers use it to couple tone arms to the pick-up cartridge.

Following on from this have made a .22 rifle tuner with a few ounces of this stuff enclosed in a tube. It works well.

Why it works so well is because it (the original Bostic Blu Tack)  is a negative thixotropic material - it is malleable under slowly applied pressure but becomes stiffer as the frequency increases.

It is absorbing vibrations by re-arranging its structure.

Make a 25mm cube from Blu Tack, pull the 8 corners to extend them and and throw the Blu Tack against a smooth, hard surface as hard as you can. What you think might happen and what actually happens is amazing.

Anyway, that's my 5¢ worth - I do not fly the same bird as you (I have a Tarot 680 with dji flight electronics) but I follow the thread in case I decide to purchase a Zenmuse.

All the best people.

* doghunter *
2017-2-8
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

Hi there fellow freelance troubleshooters and solution finders.

DJI: some input on the matter would be appreciated
DJI-Tahoe Ed: Still waiting for the torque settings on the propellers.

doghunter: thanks for the advice, I will surely give it a try.

Ricci: I finally managed to compile the promised video test flying in various directions filming at different angles. It is currently busy uploading to youtube. You can have a look at
Most noticeable is when the camera is pointed towards the rear.

Dmitry: How are your troubleshooting efforts coming on.

Let's keep moving until we find the solution!

Regards
LT


2017-2-9
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hello Chaps, and welcome Doghunter,
Great suggestion Doghunter, Non Newtonian viscoelastic sheer thickening fluid (Non-thixotropic) offers some great abilities in the world of dynamic damping but tuning the stuff is very precise, so you need to know the frequencies; however there is a fundamental downside, the properties of the substance cause the material to stiffen with shock loading, thus becoming less absorbent and better capable of transferring the load, also the rate of change is logarithmic.  "blu tack" has all the properties of the above however when it is subject to external forces the damping rate will change and become like a reversed shock absorber so it will only be usable as a non dynamic absorber or sticky mass, its non-Newtonian aspects will not be usable, which is why Lt's very dense lead on adhesive pads work so well.
I do see a use for Non Newtonian properties within the drone business but haven't sorted how to implement them yet, input welcome as this is a power free resource that changes state with force input.
The cure for the Rolling shutter/jello is not complex but will require some disassembly and relocating of the gimbal support. I have taken some pictures for people to follow.
I will also do a video comparison to justify the mods and you could do one mod at a time to see if it alone is the fix or a combination of the two fixing it.The first thing to understand is the variable nature of each craft so what we are looking to do is blanket cover any area that propagates vibration to the cmos sensor.
The discovery that the top plate is a vibration amplifier should be the first place we look at. We are trying to stop it resonating with the simplest of measures. LT,  adhesive lead pads will do the job but so will sticky felt pads, I use ones to put on chair feet, covered the carbon plate but left spaces between pads to decouple any low frequency transmissions. Like the old fashioned wall of egg boxes to absorb multiple frequencies.  The important factor is to stop the carbon plate from ringing, it should be a dull thud.  If you just do this damping then have a flight, it could be the only task you require. Another fix would be to change the material of the plate for something thicker and more dense, or pepper pot it with large holes.

If you need to go to the next step you have to remove all the camera attachment gear including the rubber dampers, and angled standoff arms and cable connectors so be gentle. You have to swap the left and right rear arms to face the opposite direction on the right & left side,  same for the front ones too. This reversing allows you to get the gimbal connector in the middle of the craft where you can fine tune its location.
You are looking to align the pitch gimbal motor to the centre of the airframe and get the COG running through the cameras COG. This is only possible in the X, direction as viewed from the side. Note the X5R will be heavier and might have a cog different to the X5 so do your best to locate the cameras cog and get it inline with the airframes Y, axis centre as viewed from the side. I could not physically move the gimbal plate back enough to get it to the centre without reversing the arms, so don't compromise here just undo the eight hex heads on the arms and remembering to use loctite 222 or similar on the threads.
Moving the camera back will mean you can see the props if you pitch above the horizon by a few degrees or accelerate 100% and that could be a compromise you might have to except.
Why did DJI not mount the camera there in the first place? Why is this going to help with reducing rolling shutter/jello? Because the airframe is a rotating mass with six potential vibration creating machines on the periphery, the place to be is equal distance from where all the sources meet and potentially nullify one another. I am sure this is a great factor in reducing the dynamic movement of the camera when manoeuvring, especially when the camera is pointing to the ground. You must take in mind that when the camera is pointing down it offers the greatest air resistance due to its profile so with this knowledge high speed pictures or video shooting will have high dynamic force issues fighting the gimbal motors which are constantly stabilising the platform so consider much slower track speed if this is an issue.

I have some before and after footage I will upload to show you the results I have, and any questions You might have please feel free to fire them at me.


Happy wobble free flying.

I cant take credit for the fix if it works, so LT who spends hours testing new things is an equal partner and in fact started the entire thread.

Ricci
move gimbal to cog.jpg felt pads.jpg
2017-2-9
Use props
fans86368c91
Second Officer

Australia
Offline

One of the reason that I suggested this low-tech solution was that when I added some Blu Tack to my dji H3-3D gimbal to correct an out of balance GoPro (it was tilted about 10° off horizontal and I wanted to reduce the load on the gimbal motors) I noticed that the camera images were sharper. I deduced that the high frequency vibrations from the (balanced) motors and props were reduced. I already knew about the Blu Tack properties so maybe this influenced my choice.

I no longer have access to powerful FEA software but it could be interesting to analyse the structure if the material mechanical properties were available.  

One place that I think may be worth investigating is to use such a material to pack out the gap in the tang.

The sad thing is that many people are doing the hard yards to correct/improve dji's products with no recognition or rewards for their efforts but I am sure that they (dji) read/discuss these threads.

Regards from Australia  * doghunter *     
2017-2-9
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

LT, what can I say.........

Job well done!  From what it was to what you have now its a world of difference.......Now we know what is causing R/s its a simple job to refine the cure.
The vid is well edited, informative and precise, great for demo purposes, so just the disconnect to sort for you then. We should set up a drone diagnostic site and actually get paid for doing this stuff.

Doghunter,
Really good idea about the Blu-tac on the tang, its perfect for the non-thixotropic material, because as soon as the connector block starts to move the B-T will Harden in sheer locking the tang together.
I wonder if M3 looked at this pile of goop in the corner and thought arrr! lets hang pictures with it? I wonder if it was a post-it note thing where marketing pulled it from the bin. As an adhesive its great at keeping the surface to surface bond together because when the mass pulls it hardens locking it in place.
Great work people.

Ricci
2017-2-10
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hi,
Update to cog on the x5, if you have the x5 with an Olympus 12-42 lens the cog is in the centre of the top yaw gimbal motor. Align this point to an imaginary line between the landing legs, the craft Yaxis..

If you have  an  X5R you could let us know where the cog is too.

Ricci
2017-2-10
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Greetings LT & Dmitry,
Just read this quote from DJI-Jamie, "The X5 and the X5R has been on the market for more than a year, it has had a prettly long run." I think we have the answer to some of the questions in that statement.
I have to say as an early adopter of the M600 and related gear I knew this day would be here when the bugs would not be rectified in the MK1 but be rectified  and incorporated  into a MK2.
I know the spares will be available for some while, I just hope the improvements are retro fixable.   The M600 is fundamentally a well designed film platform that will take some beating and in my opinion will only be succeeded as motor and battery's  evolve. If the s900 / s1000 are an example to the owners of the M600 series many parts are interchangeable and upgradable.  I picked up and used my old IPhone 4s the other day, it still made calls and joined the network as did the 7s+ beside it, so the difference was I changed my requirements. I bet the s900 is still a good platform to film from and it will upgrade to the A3,Lb2,  which only leaves the extended flight time the six batteries offer. The M600 days are not done yet. I will consider getting the X5s when an adaptor is available just because it offers better filming abilities and unless DJI have gone into the camera electronics business along with Sony and alike they too are playing catch-up. Just my thoughts to keep my chin up about my purchasing choices.
The X, series cameras were and are a great compromise to putting a red and a £5000 bit of glass on the bottom of a flying fan and as long as cmos technology keeps getting better I see no reason to jeopardise my personal mega expensive ground based tech unless a job absolutely required it. Who has an 8k monitor to see the benefits when 1080p is the norm? I might argue differently if I were in the big screen industry.
Please feel free to input if you have other ideas, this is a forum after all and not just about complaining..
Happy wobble free flying.

Ricci
2017-2-10
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hello all,
Small technical update:
I am lucky enough to have some calibrated test equipment available and decided to place small accelerometers on the pitch and yaw gimbal motors. I wanted to understand the implications of acceleration and deceleration on the camera gimbal combination, the results apply to all flying film platforms that have the ability to hover and manoeuvre in any direction. A key factor in understanding if these forces will effect you're filming is to look at how you film. If you are static in a hover and pan and tilt your craft, the effect of acceleration will have a small to no effect; however if you are doing a ground survey on the move taking shots every few metres then dynamic forces are going to be your enemy. In theory a brushless gimbal should detect the out of balance forces and apply an opposite force to counteract,  it should do this wherever it is placed on the craft. In practice the further the gimbal/camera is placed from the crafts centre the more effect the forces will have, the distance from the gimbal cog and the crafts cog make an imaginary leaver which is a force multiplier and increases the load on all the gimbal motors by many factors. I don't want this to be a school physics lesion as I am sure many of us fully comprehend the dynamics of flight and the  physics involved. For the briefest of explanations I measured the craft at 12m/s in four opposite directions 90º apposed, I placed the camera/gimbal in the factory recommended position and also in the centre of the craft, using the camera/gimbal cog as the datum.
The results: At as little as 3m/s the camera was unable to resist the load due to acceleration and an increasing differential was present between a point I called centre and the camera,  at 12m/s the forces quadrupled. This phenomenon where the craft was in reverse and there was less rolling shutter turns out to be the gimbal design resisting the dynamic forces better. I then centred the gimbal/camera on the crafts Z, centre line and repeated the same tests. The net result was 87% less dynamic forces were felt by the camera/gimbal under the same bunch of tests.  
Conclusion:  for any mass/camera combo hanging under the craft,  spend time finding the cog of the mass and align it with the crafts centre Z, line as viewed from the side to reduce the dynamic forces felt on the gimbal sensors. Also recalibrate the gimbal when it is in the centre.  I can only assume when DJI designed the gimbal mount they had no idea how the users would be pushing the envelope of the vehicle capabilities.  On that thought I can only assume again the development of the drone has progressed because the users keep pushing for more, so maybe we have to except model changes as our own desire to get more out of the technology and we should count ourselves lucky DJI have the resources to expand and develop the range.  Still hurts when your dream craft is redundant after only one year though!!!!!! bugger.
I suppose the old adage "It ain't broke don't fix it" should now say " if we broke it , how you gonna fixed it better"???

Ricci
2017-2-12
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

Hi Ricci, Dmitry & fans86368c91

With the Jello now "sort of" resolved, I've made another modification in light of the continuing gimbal/camera disconnects I'm experiencing. My thoughts around this was to lower the CoG by shifting the D-RTK receiver to the carbon fiber plate of the Gimbal adaptor bracket thus eliminating the whole expansion bracket on top of the M600 and this also has the added benefit of the craft weighing around 220 grams less. This seems to increase stiffness/rigidity to the plate as well and so far it looks promising after a few flights. I will post photo's as soon as I have everything mounted in final positions.

As for the gimbal/camera showing N/A in the Go app and me having to cancel autopilot mode missions when mapping, I'm beginning to suspect that it might actually be something to do with Lightbridge 2 overheating rather than related to any kind of vibration (I've shaken the M600 in various directions with no success in causing a disconnect). I just had an infrared thermometer pointed at the LB2 and temperatures are upwards of 53+ degrees Celsius. The craft is stationary and switched on and I have to restart it every 2-5 minutes because the camera keeps disconnecting. By disconnect, I reiterate that the camera settings disappear in the app. I'm still receiving the video feed and can control the gimbal on the R/C but cannot take any photo's or video.

Beginning to think that I should start a new thread as this issue is not related to the Jello/Rolling shutter topic.

Also thanks for all your inputs so far. It really shows that there is a community that is more dedicated to help fellow owners than the creators of the equipment. True custodianship of a product would require much more input from DJI than what we are currently experiencing.

Regards
LT

also to add fuel to the fire, I've been referred to a "respectable" DJI distributor in South Africa by DJI support to assist me in sorting problems out as a last line of troubleshooting before I return the M600 for analysis and it seems that they do not carry any stock of either the X5, X3, Z3 or gimbal adaptors. Very shocking and disturbing!
2017-2-13
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

Just saw in another post by Hawk62_Engineer where he also makes mention of the overheating LB2... Different issue but same cause. possibly.
http://forum.dji.com/thread-84828-1-1.html

What a coincidence
2017-2-13
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Lt,
I have just read another thread from someone also in SA, Hawk62_Engineer,  he said he has relocated the LB2 topside to get cooling as it was overheating. You could either do this to get it some airflow or make a ducted tube to bring cool ram-air into the general `LB zone, remember it needs an exit too. Maybe you could spray the top with a white paint. I use a windscreen reflective cover to protect the M600 when in the back of the truck.
If you want to start another thread count me in although I suspect Dmitry would dream of 53º.

ricci
2017-2-13
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hi LT,
We must of written at the same time. Thought could you run with the top cover off, it would mean no fan but it might help with radiated heat loss especially if you had an ally heatsink poking into the airstream.
look at an amplifier output stage cooling fin.
Thanks for adding to the "What I would like thread, I bet myself only three people would chip in and so far its correct alas.........  I helped develop this thermal paint for manned rocket launchers and it stopped the poor driver being boiled in his own juices. Pm if you want the stuff we made.

Ricci

2017-2-13
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

no such thing as coincidence.;
2017-2-13
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

ricci2 Posted at 2017-2-13 13:23
no such thing as coincidence.;

Agree 100%, just cause and effect.

Also had the same thought about flying without the top cover. It all makes more sense as our most successful mapping mission yet was done early morning with temperatures around 14-16˚C. All other missions were done in 25-40˚C and all of them had disconnects. I'm doing a demo for a contractor tomorrow and will revert with the outcome. If this is the cause, we shift out attention thermal management...

Regards
LT
2017-2-13
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Love thermal, new thread or keep this going?

ricci
2017-2-13
Use props
lt
lvl.4
Flight distance : 2516778 ft
South Africa
Offline

I'll post the D-RTK modification here because it is still related to stabilizing and lowering the CoG of the M600 thus improving or shall I say reduce Jello/Rolling shutter effect of the X5 in flight.

This modification results in a saving of around 200-220 grams and also shifting the Center of Gravity to a point lower than the standard M600. Firstly I removed the 85 grams lead weight on the Carbon fiber plate of the gimbal bracket. Then proceeded in mounting the D-RTK received and Datalink Pro onto the plate. I then removed the expansion bracket from the top of the M600 and mounted the antenna brackets as shown in the photo. This layout also allows the D-RTK Antenna pucks to be lowered when transporting and I suspect it would fit inside the case made for the M600.


D-RTK Received & Datalink Pro Placement

D-RTK Received & Datalink Pro Placement

Antenna Configuration

Antenna Configuration
2017-2-14
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Greetings wobbly viewers,
I am just putting a vid together that proves conclusively the fix works. It took the footage from just usable to near perfect even with the Olympus 12-45 zoom zoomed in to the max, including pointed to the ground which was unusable; however is 99% improved. Shame the camera has now been taken off the DJI store and only the Z3 can now go on the M600 pro. I wonder if it had a problem????
Thanks to LT for all the diagnostic testing and analysis that highlighted the issues. Another positive aspect is the community cohesion across many continents.

ricci
2017-2-16
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Dmitry_R Posted at 2017-2-8 19:45
Hei All,

Ricci. if it is truth, next time i am in France i will get you a huge bottle of calvados!!!! (or you will be in moscow).

Hello Dmitry,

Just wondered if you have had time to try the fixes?

Ricci
2017-2-17
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Heres the video with the results of Before and after the fix.

2017-2-17
Use props
Dmitry_R
lvl.1

Russia
Offline

Hey LT, Ricci and new wobble friends.
Sorry for not getting back for long, had lot of job to do (i was using phantom for aerials shots )   as i see you have a solutions and it is wonderful ! cant wait till i will tweak my M600 after your amazing research work.
not sure when but will update as i will have any results
Thanks!!!
D.
2017-2-19
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Hi Dmitry,
I really hope the tweaks work, As LT shows anything that damps the top plate is a good thing even blu-tack, double sided tape with foam. sticky sweets!!!!
happy flights

Ricci
2017-2-20
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

Greetings all,
Had a chance to test the Groundststion pro app today with the modified Fix and after setting a 3D map with 97 waypoints the footage looks amazing. I.e. No rolling shutter despite the craft start-stopping for many pictures. I also ran the 3D with video running and that was fine too, In the past this action would have desasterious effect as the X5 is pointed directly down, not now! so if you use the X5 for geomapping this fix could save the day.

Ricci
2017-2-22
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

More testing today of the Ground station pro app and I used the geofence. Some very erratic behaviour from the controls, very jerky movements of the M600 near the fence boundary but the footage was rolling shutter free.  I think this is as a severe test as I can push it through and the footage is fine.

all the best ex, wobblers......

Ricci
2017-2-24
Use props
ricci2
Captain
Flight distance : 61158 ft
  • >>>
France
Offline

You know when you should put a project to bed? when its sorted? Well I think I can now do this with the X5, the last test today gave me the information I have been looking for.
I wanted to know what direction the sensor travels in to generate the worst rolling shutter, now I know how to produce rolling shutter if anybody wants it as a special effect!!!!!
I was looking to do vertical testing only where the craft was lifted to the sky and then the ground at the highest accelerations possible, I changed the expo on the M600 to be very aggressive and took the craft to the top of the shop (max ceiling DJI allow) then plummeted the beast to the ground, repeat until board and do it once for luck or the batteries expire, depending on your attention span.
Results: the fix allowed me to isolate the severely impacted sensor and so on the vibration fixed craft I managed to induced some rolling shutter with extrem vertical movement, the horizontal movement did not induce R/S so the answer was the sensor is susceptible to vertical load transits or delta change velocity movement. This also answered the question as to why the R/s was bad when the camera is pointed down to the ground and the craft travels in a horizontal direction, the sensor rotates 90 degrees and its vonrable Y, axis now becomes a virtual X, axis and that's how the R/s takes hold, in theory the camera in the horizontal should have the least R/S under movement but if you add any inclination below or above 180 degrees it induces  a component of the Y, axis to be added and in my case that's what happened, rolling shutter increased as the craft was pitched in the vertical axis. Case closed I think.
Conclusion:  do the mods to the balance  the camera, locate the gimbal on the cog, and damp the top plate and that will reduce the effects of the R/S to a minimum, don't fly aggressively with the camera pointed to the ground ( diferent if taking stills as this is a snapshot in time, freeze frame.) don't use GSP if you are using a video pointing to the ground as it is very aggressive when it stops to take snapshots, 98% better with mods. Be realistic about the footage as you are hanging a camera in a floating fan in an unpredictable weather environment. Job Done.So far there have been only three people to acknowledge the rolling shutter on the X5, and IT, and myself have made huge improvements to our situation however if you have made any mods it could be good to get feedback.
Happy flying and a huge thank you to LT, and all who came together to fix this Non issue DJI.......

Ricci
2017-2-26
Use props
1234Next >
Advanced
You need to log in before you can reply Login | Register now

Credit Rules