new hardware revision autoquad?

Info and discussion about the original AQ v6 flight controller

Re: new hardware revision autoquad?

Postby bn999 » Fri Feb 14, 2014 3:14 pm

JussiH wrote:But in the end the socalled "vibration sensitivity" in AQ comes down to the fact that at some certain amplitudes and frequencies the ACC data becomes corrupted by saturation and even though the UKF is very powerful, it just cant recreate data that is getting lost by saturation.


I think we need to be clear on this point so that people do not spend money on a FC that is not appropriate for their application.

The AQ platform has been specifically designed to achieve the absolute best performance possible limited only by the limitations of the low cost MEMs sensors that are available on the market currently. All testing and tuning was done with the expectations that the sensor data is as clean and noise free as possible. During development, if problems were witnessed caused by vibrations, I did not waist any time trying to figure out how to gracefully deal with them. I instead went to the root of the problem and eliminated the source of the vibration.

Other FCs do not integrate ACC data but instead put low pass filters on position and altitude reports from the GPS / barometer. This is fine and it shields you to some degree from bad data from ACC's and MAGs, but you will never achieve the level of performance of fully fused inertial navigation system such as AQ employs.

If you are experiencing vibrations strong enough to saturate the ACC's, then you should not be flying - you should be spending your time eliminating them. If you are not willing to do this, then you should not be flying an AQ at all. I suppose that sounds a bit elitist, but it is just the truth. Don't expect AQ to do something it was not designed to do.

One of the primary applications for multi-rotors is capturing video from the air. Having a vibration free platform is goal that is well aligned with this philosophy.
bn999
 
Posts: 1559
Joined: Thu Jun 21, 2012 11:40 pm

Re: new hardware revision autoquad?

Postby phynix » Fri Feb 14, 2014 6:31 pm

For me, it is ok to have something that isn't plug and play. And, also the path of eliminating the source of a problem and not the symptomps sounds reasonable and it also worked for me. Overall, I am happy with the AQ. For me, it is both: a "toy" for my hobby but also kind of a !!! SPAM !!!. But latter probably would not work without the former.

BUT: If I have a copter that is setup nicely and has a minimum amount of vibrations and so on (as I try to set up my "!!! SPAM !!!" right now) and I tune it to the best performence, then I have one big fear: If something happens in flight that causes suddenly high vibrations (broken prop, screw that gets loose, damaged esc, minor collision with a branch of a tree...), it can actually lead to a divergence of the UKF resulting in an inevitable crash. I am afraid that this risk seems to be significant higher with the AQ compared to other FC. At least one of my copters drifts in pitch or yaw (resulting in a crash) reproducible when the vibrations get a little bit higher (as e.g. with cheap Gemfan-Props). And I observed the same when I accidently mounted a bad, unbalanced prob on my octo! Twithing continuosly to one side, probably due to saturation effects....

Even worse, PID settings that support the most stable flight for a low vibration setup by trend create a higher sensitivity to vibrations (high D-values)!

I will stick with the AQ, but this is one point that might be worth to take into account. Especially as the AQ is designed for camera-copters ;). For me, this is an important safety issue. What is the use of an octocopter, that can still fly when one motor is lost, but the FC causes a crash when the vibrations exceed a relatively low level?

A machine, that is working almost perfect on a "sweet spot" as it is highly optimized for it, I love it. I am also willing to work hard to get it into this sweet spot. But if a crash is almost certain as soon as the sweet spot is left....

I would realy love to see the AQ to be a little bit more "robust". It does not has to fly well with incresed vibrations e.g., but it has to be possible to get it home safely.

I hoped that this will be less risky with the DIMU :).

This is not meant to criticize the work of the team, don't get me wrong! Overall, I am still happy with the support in forums and the progress of your "baby"! It is just meant to be an additional point of view...

Sören
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: new hardware revision autoquad?

Postby 13brv3 » Fri Feb 14, 2014 8:01 pm

I think this is a good discussion to have, and I can easily see both sides of this. I see the current AQ as a great FC board for enthusiasts who are willing to put forth the effort that's require to get the best performance. There isn't anything at all wrong with that, assuming that's the goal. I always like to have a bleeding edge project to play with, so I'll continue tinkering with it for sure.

On the other hand, if the board will ever be intended for end users, I believe it's going to have to be more forgiving of "real world" situations. At the moment, I can't really recommend it to anyone but the most dedicated users. I hope that changes, but I can accept that it may not.

Cheers,
Rusty
13brv3
 
Posts: 152
Joined: Wed Oct 24, 2012 1:10 pm

Re: new hardware revision autoquad?

Postby bn999 » Fri Feb 14, 2014 9:02 pm

phynix wrote:I will stick with the AQ, but this is one point that might be worth to take into account. Especially as the AQ is designed for camera-copters ;). For me, this is an important safety issue. What is the use of an octocopter, that can still fly when one motor is lost, but the FC causes a crash when the vibrations exceed a relatively low level?


I don't know how we got to the point of saying that "relatively low vibrations" == crash. I reject that notion completely and cannot let it go unanswered.

About the worst that I have seen is high vibrations causing the ACC Z axis output to go into a VRE regime resulting in the machine rising quite quickly since it thinks it is falling - in PH mode. That can only happen with very large vibrations, which again, you should not be flying with.

As to stability being impacted by high vibrations, I have not seen it, ever. The primary driver for attitude control is gyro rate output. The gyros can perform worse when subjected to high vibration, but not the degree that you loose your attitude estimate. You might be in for a shaky ride, but that's about the extent of it.

In fact, I have done testing where the ACC readings have been purposely turned off during flight. The machine flies perfectly well with only the GYO & GPS input in manual mode. Additionally, the further the ACC readings are from normal 1g output, the less the system trusts them for any attitude correction.

I have not read this entire thread, but if somehow the conclusion is that high vibrations caused someone's machine to crash, I'm certain there was something else going on.

Other than mechanical failures or pilot error, most crashes (with AQ) seem to be caused by improperly tuned PIDs or motor mixing tables. This is to be expected as there is no one size that fits all.

To comment on the DIMU, based on my close to 1 year experience with it, performance is much better than the original AIMU as one might expect for next generation sensors. The sensors are easier to calibrate due to the fact that GYO & ACC are on the same die. There is a 4x increase in dynamic range on both and the ACC's VRE performance is much better than the original ADXL. Given all of this, the MAGs have become much less important and it looks like it might be possible to ignore them completely during flight without compromising performance. This would completely remove any worries about magnetic interference and drastically reduce the currently stringent MAG calibration requirements.
bn999
 
Posts: 1559
Joined: Thu Jun 21, 2012 11:40 pm

Re: new hardware revision autoquad?

Postby 13brv3 » Fri Feb 14, 2014 9:51 pm

bn999 wrote:To comment on the DIMU, based on my close to 1 year experience with it, performance is much better than the original AIMU as one might expect for next generation sensors. The sensors are easier to calibrate due to the fact that GYO & ACC are on the same die. There is a 4x increase in dynamic range on both and the ACC's VRE performance is much better than the original ADXL. Given all of this, the MAGs have become much less important and it looks like it might be possible to ignore them completely during flight without compromising performance. This would completely remove any worries about magnetic interference and drastically reduce the currently stringent MAG calibration requirements.


As always, thanks for your comments Bill! This all sounds very promising 8-)

Cheers,
Rusty
13brv3
 
Posts: 152
Joined: Wed Oct 24, 2012 1:10 pm

Re: new hardware revision autoquad?

Postby phynix » Fri Feb 14, 2014 10:40 pm

Thanks bill for your explanation. As I said, it was not my intention to criticize anyone, expecially not your work.

Ok, I have only access to a limited amount of experiences that I made with three different copters. I rarely observed a sudden rolling or pitching of the copter that was not visible in the telemtry (artificial horizon in QGC stayed leveled - so the attitude estimation got probably corrupted). This "issue" was gone or at least significatly reduced after I balanced props etc, leading to my assumption, that a certain amount of vibrations can lead under certain circumstances to a divergence of the attitude estimation. But, this is my personal experience and not an extensive analysis. Especially, I still did not go through the theory of the UKF. I should have made this more clear. Sorry for that! It didn't want to spread unproven rumors. The thing is, that "you have to reduce Vibrations" seems to be one of the common answers that is given when an in-flight-problem is descriped in the forums.

If the AQ is a little bit "shaky" in the presence of vibrations, I don't care much - as long as he stays maneuvrable :).

Ok, back to the topic of this thread. The facts about the DIMU sound great, especially if the AQ firmeware can make use of the increased accuracy :). I am realy looking forward to have fun with it! A board that has directly mounted the DIMU instead of the AIMU in the near future would e great.
phynix
 
Posts: 73
Joined: Mon Feb 11, 2013 11:03 pm

Re: new hardware revision autoquad?

Postby ViennaTom » Sat Feb 15, 2014 3:26 pm

quote "i'm a bit afraid of soldering the headers to the controller board and would appreciate if i could order a controller with preassembled extension header regardless whether the DIMU board development was finished"

is this that hard to understand ? neither did i ask for a plug and play solution nor for an already finshed hw/sw development of the new DIMU solution. i just don't want to start with a board that i have to solder extension headers or to see another becoming available costing less and performing double. i thought the 'final' solution will be based on the extension header with an optional DIMU board but maybe i'm wrong with that assumption. maybe the best place for the dimu would be further away from the electronics of the board anyways ....
thx Jussi for the offer to solder the header for me - really i have done plenty of soldering in my life but baking or small size smd are beyond my skills.

@terramax:
ich hab schon eine ganze menge dinge zusammengelötet und sicher schon mehr bugs in sourcecode verpackt als du fliegen im stall hast - gerade die offene sw-architektur und die möglichkeit 'erweiterungen' dazu programmieren zu können (kamerauslöser, fallschirmauslösung, gimbal optimierung, ....) reizt mich an diesem projekt.

thx for all of our responses anyhow - i will go to cambodia now for 4 weeks and came back later :-)
ViennaTom
 
Posts: 3
Joined: Tue Jan 21, 2014 11:59 pm

Re: new hardware revision autoquad?

Postby Jdmagoo » Sun Feb 23, 2014 1:35 am

@Jussi

I reckon we must be getting close?? The Chinese New Year is over and hopefully the DIMU's are on the way.

Also, will viacopter sell CAN bus upgrade kits for the existing ESC32. Would be good to purchase all the appropriate components together in the one spot without the worry of ordering the wrong parts
Jdmagoo
 
Posts: 485
Joined: Sun Feb 17, 2013 8:35 am

Re: new hardware revision autoquad?

Postby JussiH » Sun Feb 23, 2014 12:54 pm

Yes, yes and yes....

DIMU´s are in production, I should have them soon - I will know more this coming week.

We will sell CAN upgrade kits too. I will also offer an assembly service on the boards and ESC´s
JussiH
 
Posts: 2103
Joined: Thu Jun 21, 2012 9:25 pm

Re: new hardware revision autoquad?

Postby TFSyth » Fri Mar 07, 2014 3:51 am

Will the new DIMU prevent using the VectorNav VN100 module? or should the rugged version be used?
TFSyth
 
Posts: 4
Joined: Mon Mar 03, 2014 8:50 pm

PreviousNext

Return to AutoQuad 6 Flight Controller

Who is online

Users browsing this forum: No registered users and 3 guests

cron