S800 Arm's carbon fiber wrapped

K

ksrimoungchanh

Member
Well, with just 2 6000mah packs about 200ft up, I accidentally took throttle below 10% and it killed all props.. it was sideways, when I started the motor and was about 40% throttle and it recovered. I am sure there were more then normal G's put on those arms. It did not break.. now I want to support the arms going forward. I am afraid of micro fractures that are not visible to me or detectable by eyes only.

Last thing I need is to create enough heat on the ABS plastic arms with micro fractures while the Zenmuse on it.

KS


Acmach5 said:
Have S800 users had arms break in flight? I feel that this video above makes the vibes look worse than they actually are. I don't care how well balanced your props are. When they are spinning opposite directions and constantly changing speed for flight you are going to get some shakes. I gotta say if the arms are of poor quality and breaking thats one thing. If they are designed to have some give I'm ok with that. It's been working well for Boeing for quite some time now in their wing design. If you consider any successful modern aircraft or building, they generally have some flexibility designed into them. What DJI needs to do is some stress testing and post the videos showing at how many pounds of pressure the arms break.
Click to expand...
 

D

DennyR

Active Member
To give a better understanding of how much vibration can be attenuated with electronics consider a simple loud speaker. It has a voice coil which is basically a solenoid. It vibrates a disc against a magnet which creates sound waves which your ear drum detects and a vast range of frequencies can be heard. It is the same principle that is used in space stations to eliminate hull vibrations on telescopes that sit on a hexapod. It is the same thing inside a Cineflex and most other high end camera mounts. the voice coils also create a few degrees of inner axis movement. You can watch a Zen going through its initial first time calibration. it will vibrate one axis at a time whilst measuring the frequency response with the other two. A noise canceling headset works by sampling a background noise and then creating a reverse waveform that effectively cancels the noise by turning that wave into a straight line.:tennis:

How many Zen videos have you seen with any sign of vibration artifacts. That is the bottom line. Does it work or does it not? The only way that I could deliberately induce a slight vibration was to stick it in the direct 120 mph slipstream from a full size helicopter. You cannot hand hold any camera in that environment without vibration.

The standard S800 works just fine. forget about flex in the arms just get out there and learn how to use it properly.
 
Last edited by a moderator:

D

DennyR

Active Member
jes1111 said:
For sure the requirements for a gimbal controller exceed those required for MR stabilisation. Gyros are accurate but they drift - so you have to add at least 3-axis accelerometer and some clever maths to keep the gyro drift down. But then you introduce the dreaded horizon drift because the linear acceleration (from sudden and high-g movements) biases the estimation... and so on, round and round in circles ;) - so you introduce more sensors (mag, barometer, etc.) and fuse all the data together - but up goes the cost and complexity. And/or you use more expensive sensors that have better characteristics - and then the complexity of the maths goes ballistic. :)

There's no doubt IMO that DJI either farmed out the development of the Zenmuse or acquired some fresh (and serious) talent internally - the electronics side of the Zenmuse is significantly higher level than their (current) FC.
Click to expand...

As the Zen uses a standard WKM IMU for its basic stabilization what you say is totally incorrect. It has a second high sensitivity IMU that is placed on the final axis for the final part. The major part of the pointing is done with the same IMU as the model is using. That is easily proved by the fact that you can direct the heading of the camera in all three axis by moving the WKM IMU. The tiny High sensitivity IMU is constructed in the same way as the FC IMU in a folding box configuration. It was designed and made by the same people without any doubt. Whereas my own mounts had an inner axis that moved only a few degrees DJI used that inner axis IMU idea to modify and refine the output to the 3 main axis. Made possible by the precision engineering.

I spent hours trying to convince Geo Mamo that he did not need to make an AHRS as all MR's already had a perfectly good one otherwise it could not fly. All he needed to do was make a simple 3 axis IMU with high sensitivity gyros and accels. This would then be used to modify the final output to the servos. Just like a tail rotor gyro works. I used Colin Mills 720 tail rotor gyros modified with 50deg/sec gyros. they worked like a Zen. I tried several other IMUs including a CC but the registry could not be modified to access the high sensitivity outputs for some strange reason.

Why would DJI farm out any design, they have plenty of highly qualified EE's and Mech. E's. Take a look at the job situations vacant there to see what the minimum is.
 
Last edited by a moderator:

J

jes1111

Active Member
It should be perfectly obvious, Denny, that I'm poking and prodding at the Zenmuse and how it works precisely because I don't know. Not having one in front of me (and not being rich enough to buy one just to look) I can only rely on information gathered here from owners, combining that with what I've learned in the wider sense. What I said above is not "totally incorrect" - the general paragraph about sensor fusion is a simplification, for sure, but it's more useful in carrying the conversation forward than a breakdown of Extended Kalman Filtering versus Complimentary Filtering (which I could only cut and paste anyway). As I've said elsewhere, the ability to move the axes by tiny, accurate amounts is not the limiting factor - being able to measure with sufficient speed and accuracy, free from noise and bias - that's the really tricky part. I'm still trying to discover how DJI does that without using a very expensive mil-spec gyro. I've already speculated that the mother IMU is supplying a "gross" attitude estimation - only now have you confirmed that by (reluctantly?) revealing one more detail. I do wish you'd stop interjecting only to deny - it's counter-productive.

It's easy to observe the noise levels in the gyro and the accels on a CC/CC3D - just bring up the raw scopes in the GCS. In both cases the noise floor is way above what would be needed to meaningfully correct the rotational errors at "sub-pixel level" (quoting from the DJI website) whether you use direct drive brushless or voice coils or whatever. Combining the sensor data with some clever maths can produce a reasonably accurate attitude estimation, I'd like to discover how DJI do it - most especially, what components are employed.

I also stand by what I said about DJI's engineering expertise. The DJI WKM is nothing revolutionary - in fact, it's an average performer in this field, indicating that the engineering involved is "off the shelf" stuff, pretty much available from the app notes published by the sensor manufacturers. The Zenmuse is in a different league. Job adverts are always an interesting piece of intelligence about a company - they show you what expertise that company doesn't have ;)
 
Last edited by a moderator:


D

DennyR

Active Member
I also stand by what I said about DJI's engineering expertise. The DJI WKM is nothing revolutionary - in fact, it's an average performer in this field, indicating that the engineering involved is "off the shelf" stuff, pretty much available from the app notes published by the sensor manufacturers. The Zenmuse is in a different league. Job adverts are always an interesting piece of intelligence about a company - they show you what expertise that company doesn't have ;)[/QUOTE]

There are a few people out there, who like me knew how to do this a few years ago. The fact is that nobody would risk the investment and the time to do it, and then offer it at an affordable price for the average modeler. DJI took the bull by the horns and did it and made a great job of it. For around 20 K mil.spec. stuff has been around and probably no more expensive to produce than the Zen. I am frankly surprised that it took them so long. I was banging on about gyro FSR for years before anybody took any notice because they all wanted to build hooligan IMU's that would perform aeros. Your CC was one such case. all fearing the saturation point, despite the fact that industry professionals were saying that camera movements of even 5 deg/sec were to fast. Inner axis sensitivity is in the order of 50x greater than a ADX620 from AD. and is not expensive. You guys were given all of the right clues.

Yet we still have people wanting to swing gimbals about with 360 deg. rotation so that they can perform what they still believe will be great footage.

As far as manufacturing a Zen competitor goes, the profit on each unit would not get me out of bed in the morning nor any sensible European manufacturer. If you had ALL of the manufacturing info how could you set up CNC machines and perform the bulk buying etc. and come out with a profit. They will be on a new and better model before you get the first one out of the door.

More importantly Mclaren locked out the front row at Monza tomorrow... That is what I call real state of the art engineering.
 
Last edited by a moderator:

P

Pedro nunez

Member
Acmach5 said:
Have S800 users had arms break in flight? I feel that this video above makes the vibes look worse than they actually are. I don't care how well balanced your props are. When they are spinning opposite directions and constantly changing speed for flight you are going to get some shakes. I gotta say if the arms are of poor quality and breaking thats one thing. If they are designed to have some give I'm ok with that. It's been working well for Boeing for quite some time now in their wing design. If you consider any successful modern aircraft or building, they generally have some flexibility designed into them. What DJI needs to do is some stress testing and post the videos showing at how many pounds of pressure the arms break.
Click to expand...

One of our arms broke last November mid flight.
We have video and photos of the incident. Kept quiet because DJI promised all sort of great customer attention. They did not offered that, not even close.
We got it replaced and the arms material seems different now. We are still considering ways of reinforcing them. We would not buy an S800 again given the chance.
 



You must log in or register to reply here.
Top