First truly successful flight – a few questions

[hogster]

Hi everyone

I'm thrilled to say I've just had a set of successful test flights of my X8 platform ballasted up to about 5.6kg. This is the platform:

http://www.horizonimaging.co.uk/blog/20 ... -platform/

It seems to fly very well, however I have a few questions.

This is too much vibration isn't it

16-12-14 ACC plots

I'm really at a loss as to what I can do to reduce the vibrations. I've spun each prop up in turn to around 60% throttle and the chassis has felt very smooth and vibration free. I also spun up each pair of props and it's been similar. Here's a plot of the ACC values when I did these tests:

9-12-14 motor run-up tests

There are 4 sets of 3 tests: top prop, bottom prop, both together (again, spun up to 60% throttle each time). Since producing this plot I balanced the props still further by adding tape and checking the vibration in the frame so it should be even better now.

I still wonder whether it's the harshness of the motor demands that's causing additional vibration – here are all 8 MOT plots from my flights today:

16-12-14 MOT plots

And here's just a single MOT plot from a single hop so you can see it better:

16-12-14 single MOT plot

I've seen other people's MOT plots and they're practically slightly undulating lines – none of this harsh noise nonsense! Is there anything I can do to tame these MOT demands? I very much doubt the motors can respond that quickly anyway, and in the worst case it might even cause the ESCs to lose sync the demands are changing so rapidly ...

The AQ board is mounted on reasonably hard M3 bobbin anti-vibration mounts. Before I added these I had an incurable motor 'ringing' problem, so they do seem to be helping slightly (the vibration was just as bad without them, but they've obviously made a subtle difference to the type of vibration being picked up).


Another slight problem I noticed was that engaging alt/pos hold always made the platform sink very slowly. You can see this in this GPS HEIGHT plot – you can actually see the height decreasing each time I engage the alt/pos hold mode:

16-12-14 GPS HEIGHT plot

I tried increasing the Navigation Altitude Position P and I values but to no real affect. What parameters should I be adjusting to give the altitude hold more authority?


Other than that it seemed to fly very well, quite remarkable considering the heavy vibration present! I wasn't able to induce any oscillations in flight when giving sharp control inputs so that's good. When hovering in ground effect there was a little oscillation though, but maybe that's a good sign – not to see it in flight but to see a little oscillation in turbulent air just above the ground?


By the way, here is my EagleTree plot from the same flight:

16-12-14 EagleTree plots

Approximately 31A to hover which equates to around 12.5 minutes of flight time if using 80% of the battery's capacity – not too shabby


Many thanks for your help, it's always much appreciated.

David

[SuperSharpShooter]

Make sure you take it out on a very windy day and fly it into the wind and see what happens. With the amount of vibes you're seeing in a calm hover, I would expect you to have problems.

I have a hexa and a quad here which have very smooth ACC plots on a calm day, and flies more or less perfectly, but as soon as I fly them into any sort of wind (15km/h or above), there is some sort of vibration/resonance which saturates the ACCs and results in full power climbs while in position hold. I haven't managed to solve this problem despite balancing everything, and trying endless variations of the tilt rate and angle PIDs. Been meaning to buy a whole new set of props to try out, but don't want to waste more money as these props/motors have been well dynamically balanced already.

Have you considered just ditching that frame and going with a flat hexa running the same motors?

Alternatively for $1400USD you can get an s900 frame + motors + ESCs. I'm running a Pixhawk on mine, and get perfect smooth ACC plots and predictable and accurate performance. Flight time of around 20 minutes with a 400g camera + 3 axis gimbal and 16k battery. The pixhawk is mounted on a 3d printed . I tried mounting the AQ board on one of these mounts on my quad which frequently gets its ACCs saturated, but couldn't get it to fly acceptably at all.

[hogster]

Hi SSS, thanks for your reply.

I'm in the same position as you with your props – I've invested a large amount of money and time in this frame, and I don't want to spend any more on a completely new frame

This frame doesn't seem to have inherent problems or there wouldn't be so many videos on their site from people using it successfully:

http://www.carboncore.com/
http://www.carboncore.com/video-showcase/

If anything it seems the AQ is the weak link here as the DJI controllers don't seem to have any problems with these airframes. Is there anything that can be done in the software to cope better with strong vibrations? Either I need to continue experimenting with the settings, or I need to fit the AQ to a better vibration isolation mount ... or get a DJI controller which I really don't want to do!

Cheers!

David

[SuperSharpShooter]

Yeah it looks like some people are getting good results with your frame. I wonder how much of that good video performance is down to running a zenmuse also?

It might be a case of AQ just not being able to handle vibrations that other FCs seem to be able to cope with due to the hard mounting. I know the wkm/a2 is supposed to have some sort of dampening in the IMU box which probably helps on these sorts of frames. I have been trying to avoid buying into the DJI ecosystem too heavily, but I did buy a Naza as a backup/test to see if I could get one of my AQ craft flying better with it. All of the APMs I've ever run have had the FC on an anti-vibe plate too.

And once you've dealt with the vibes, there's still the black art of PID tuning which I've never managed to fully perfect on any of my Autoquads. I would really love to get them all flying 100% though since Autoquad has the best feel of precision control out of any of the FCs I've tried. I just can't get the occasional wobbles out during flight, or to get it to that stage of perfection I very much would like to achieve.

[kinderkram]

If there are vibrations on a craft it´s better to fight the source of the bad vibes rather than trying to cover the symptoms.

Glenn: do you have a log file of your ACC and motor outputs on the Pixhawk or APMs in flight? You say you have them all mounted on antivibe mounts so I guess they don´t work without.

[afernan]

Hi David:
yes, you do have a problem with vibrations. This is the first problem you need to solve, then all other problems will be fixed. In fact AQ controller have an important constraint with vibrations. First a few basics things:

- AQ is highly sensitive with vibrations (much more than any other board I know). This is one of the main reasons why it flies so well. Please have a look to many post on that (like viewtopic.php?f=43&t=3916) and papers like this: (http://autoquad.org/software-downloads/?did=127)
- You should be below 1.5m/s2 tipically. Now you have around 14m/s2+ !!!. You´r lucky it can fly!
- AQ don´t like vibration dampers. The will do the thing even worst, evenmore they will prevent to fly.
- Frames that can work for other controllers may NOT work for AQ with same setup (frame, motors, ESC´s and propellers). This is because (see next bullet...)
- AQ copter setup works like a "whole"; I mean: all interacts with the control: frame design, ESC´s type and calibration, motors, propellers...

I´m deep working in stablish a practical method to design a copter with AQ. For the moment the best approach you can do is to reproduce a succesfull configuration (see our wiki http://autoquad.org/wiki/wiki/building- ... ce-builds/)

To design a succesfull new frame with AQ is NOT an easy think, for sure but when you get one working it really worths the effort!

Angel

[SuperSharpShooter]

[hogster]

Many thanks for your replies everyone, and sorry for the delay in mine

I appreciate that vibrations are definitely an issue, and I'm now experimenting with mounting the AQ on vibration damping grommets. So far the test results look promising ...

On the second graph from the top on this page (the individual motor runs) the worse vibration was about ±3. With the vibration dampers, I'm now looking at ± 1, but mostly around ± 0.4 less. This was only with tests on one arm, but I expect the other arms to be similar.

I appreciate the AQ is not meant to like soft-mounting, but I figured it's worth a try – things can't exactly get worse than they are at the moment! Also, I reckon the only reason the DJI controllers don't seem to have a problem with this frame design is their IMUs are vibration isolated in their own plastic enclosures. Hopefully adding the damping grommets will replicate this performance for the AQ ... or at least improve it!

I will post my findings here ... watch this space

Thanks again for your help and I wish you all a very happy Christmas!

David

[hogster]

Good news!

Just had a quick test hover and the vibrations seem to be an order of magnitude lower ... well maybe not an order of magnitude, but it's a lot less!

23-12-14 ACC plots

The motor outputs seem much smoother too – here's a single MOT plot:

23-12-14 single MOT plot

It felt just the same to fly as before and the wind was gusting harder this time ... I need to do some longer test flights but this certainly looks very promising for now!

Cheers!

David

PS. What is it about soft-mounting that the AQ doesn't like? I can imagine if the dampers cause the AQ board to oscillate then this can certainly cause havoc with the control algorithms ... but if, as in my case, they just reduce the levels of vibration (quite considerably!), what is the problem? Thanks

[hogster]

Ok another test flight, this time longer and in more wind again.

24-12-14 ACC plots

The results are still considerably better than before (although not quite as good as my garden flight due to the additonal wind), and after making a few tweaks to the PID settings to try and iron out some little wobbles, it seems to fly very well

The GPS position hold still didn't work well with holding height – it always starts sinking when I engage the pos hold mode, despite fiddling with the Navigation Altitude Position P and I values. Are these the settings I should be changing to improve the altitude holding power? The position hold seemed to do what it was meant to though ...

Looking at the GPS_HEIGHT plot, it's correctly showing that the platform is sinking as soon as I engage the pos hold mode ... so it knows it's sinking, it's just not doing anything about it!

Many thanks for your help and I wish you all a very happy Christmas

David