ArduCopter 2.4 released

MyLand February 23, 2012 at 1:38pm

Hi everybody,

Small video, test V 2.4.1.

https://www.youtube.com/watch?v=A0A_E5G4No4

My config:

Frame: Silver H, from Beedrones, Hexa frame.

Motor: A2836/9

Props: 10 x 3.8

Esc.: Jdrones 30 A, reflashed with SimonK firmware

APM1

Battery: 4S, 3000 mHh, 40C

Weight: with camera mount plus 1 servo: 1920 gr.

Wind speed: 6 Km/hr.

Mode: Stabilizer

Next : Alt_hold and Loiter.

Best regards,

MyLand (Rolando Martins)

Hexa_setup.jpg

Developer

Marco Robustini February 23, 2012 at 1:30pm

ArduCopter v2.4.1 and latest fix on GIT:

I read with pleasure that your flights are still without any problem, well, I'm happy, even if we do not recommend to fly with the version currently in the planner but only the GIT version, where the problems seem to DCM and "throttle_cruise" are solved, at least from my local tests.
The evolution of the code I remind you that you can see here:

http://code.google.com/p/ardupilot-mega/source/list

-= Marco =-
(AC Dev Team)

Ihab Ali February 23, 2012 at 1:15pm

Hey Guys,

I figured out my problem...had a couple of bad servo cables that were causing some sort of a short, as soon as I replaced them everything worked fine...sorry about the earlier post.

Now on to flying 2.4.1: All I can say is OMG!!!! I didn't even have to touch the PIDs at all...it flew perfectly stable and responded exactly how I like it to...not even a single parameter tuned!!!!!! This is absolutely amazing!! Dev team, thank you very much for the great effort!!!!!

I have a minor problem with Yaw and I'm not sure if this is something unique to my setup or if it's an issue already being looked at: basically the copter always wants to yaw clock wise; I'm not talking about the yaw on take-off...basically just a very slow yaw as long as its hovering or flying . I've been using the exact same frame as I have in my 2.3 tests and I haven't touched it since then...i didn't notice any frame issues that would cause this. The only thing I changed is that I mounted the board directly to the frame using a couple of layers of double sided foam tape vs the gyro foam that I was using before(which was too flimsy). Is this possibly a vibration related issue? Compass is on and declination is set to 1.45 as I always do...radio calibrated a couple of times already...esc's claibrated...copter leveled and autotrimmed..no mixes or anything else in the Tx.

Moderator

Graham Dyer February 23, 2012 at 11:37am

Will try a few more things to get rid of my "leans" problem otherwise I'm going to shelve APM2 for a while and go back to APM1 until this gets sorted. I've just installed my 6th anti-vibe foam type (2 types together now actually) and will try again in the morning.

Ihab Ali February 23, 2012 at 11:09am

Hey Guys,

just loaded 2.4.1 from MP(twice actually)...something strange is going on: It does not see my receiver anymore(the blue light on the main board does not blink) and the ESCs keep beeping like there is no signal. Any idea what's going on?

Using APM1 with 2560 and MP 1.1.4

Bill Jowitt February 23, 2012 at 6:44am

Some more observations on flying 2.4.1 and Loiter...

Stab P, Nav P and Loiter P all affect Loiter and appear to be related e.g. if you increase Stab P to get stability and smoothness in windy conditions, you need to increase Nav P as well. The relationship between Nav P and Loiter P is harder to determine, but I think that Nav P gets the copter back to the region of the Hold Waypoint and Loiter tries to keep it there. What is unclear is to what extent other Nav parameters e.g. WP_LOITER_RADIUS, have an effect.

The code does not seem to cope well with varying and/or strong wind conditions. This morning I had wind varying from almost still to strong and gusty all within a 9 minute flight - a nightmare to try and tune. My conclusions so far are that as the wind increases, you need to increase Nav P to enable the copter to get back to the Hold Waypoint (as invariably it gets blown downwind when you release the sticks prior to invoking Loiter). Ideally of course, we should be able to have one set of parameters for all (reasonable) conditions. My feeling is that those who have obtained tight and/or smooth loiters have done so in low and/or steady wind conditions and by tuning down the parameters - if those who have succeeded in more difficult conditions can give me some pointers on what I am doing wrong, I would be grateful.

The upper limit of wind that it seems to be able to cope with is approaching 15mph. On my last flight the wind was around that, which gave me enough problems when flying manually, but to enable the copter to get anywhere near holding a loiter (even a big one), I had to increase Nav P to 5 or 6, with not very satisfactory results. As well as the guide and/or documentation that everybody is asking for, some outline specs of what a stock 3DR/ArduCopter drone should be capable of would be helpful (to prevent wasting time in out of scope conditions). By contrast, I have seen video of a DJI WKM coping very well with gusty winds of around 20mph (admittedly it was a large hexa or octo).

It seems to me that the algorithms work in 3-5 second pulses - especially in the initial phase of trying to regain the Hold Waypoint. I can watch the copter lean over and start moving, and then a few seconds later it "gives up", gets blown back and then starts again. Is there an I term somewhere that is building up to stop the achievement of loiter? Has anybody else observed this behaviour?

I am going to raise a suggestion to modify the entry conditions for loiter - it seems nuts to me that in wind we have to release the sticks as that inevitably produces a distance error that needs correction, making the whole process harder.

As well as a guide, documentation on all the parameters would be a great help. A lot of the above is my speculation based on observation and is not informed. The more documentation we can have on how the algorithms work, what the parameters do, and a practical guide, the happier I am sure we will all be.

If the above sounds negative, it is not meant to be. I am flying my copter in stronger winds than previously as I have greater confidence and whereas previously I was impressed by 15x15m boxes in wind, now I am expecting to get under 7x7 and on occasions am getting much less. It is getting there, but with a few changes I am sure it will be improved further. The overwhelming need now is for the documentation to be complete and to keep pace with changes to the software. Regards, Bill

Moderator

Graham Dyer February 23, 2012 at 6:26am

Graph of my leans problem. Log attached.

During this last flight I had to hold in more and more roll and pitch to offset the lean, it's like the trims were moving by themselves.

At the end I landed and took off again and the effect was diminished but still present. This is different to the sudden death lean that I've had (and posted about) previously, that was rapid, happening in a less than a second, whereas this is gradual increase.

I've removed the motor mounts and mounted the motors directly onto the aluminium arms but with 1mm rubber strip between motor and arm and between arm and bolts, so the motors and their bolts are 100% insulated from the arms. I tried adding 10g of lead the the APM2 board but that made flying worse as predicted.

Next to remove the soft foam and put something firmer in place.

Lean_prob1.log

https://storage.ning.com/topology/rest/1.0/file/get/3692362271?profile=original

Lloyd Breckenridge February 23, 2012 at 5:21am

While I think the new motors CLI test is useful is there a way of checking the pitch and roll inputs like the previous version?

Thanks

Moderator

Graham Dyer February 23, 2012 at 3:12am

My current APM2 anti-vibration mount (re: trying to solve the "leans" problem):

Maybe the soft foam is too soft rather than too firm?

Bill Jowitt February 23, 2012 at 3:03am

Quick observations on flying 2.4.1 and trying to get it to Loiter, in reasonable and variable winds.

Despite others observations, in windy conditions it appears that Nav P is needed, at the moment I am back to the default of 3, otherwise the copter does not get back to the origin.

From this I deduce (in the absence of up to date and comprehensive documentation), that the algorithms use Navigation to get to the Hold Waypoint, and then Loiter to keep it there. Is that correct?

Question - the current WP_RADIUS default is 100 - it used to be 2. I presume that there has been a change and that means 1m - am I correct? For windy conditions would it be good to set it to 200, i.e. 2m, so that Loiter takes over when it is back in the vicinity of the Hold position, rather than it charging past and having Nav and Loiter switching in and out as it were?

I will try in the meantime, but if anybody can actually post how Loiter is achieved, and what the Loiter and Nav P and I terms actually mean in terms of trying to tune against observed performance, I would be most grateful. Cheers, Bill

Discussions

ArduCopter 2.4 released

Replies

Activity