Ok update guys.  The last two releases....  whats up?!?!?

Ive now updated firmware twice and both times I regret it.

From the factory my APM was rock solid!  Rock Solid....  not Obama rock solid.

Today was my second firmware update and its as satisfactory as the first.  The first upgrade caused wandering and subsequent crashes with version 3.1.2.

Saw 3.1.3 was available... so I update in hopes the issues were fixed with 3.1.2

Now if something has a problem you make a fix... or "update". Right?

I mean the end objective is for the thing being updated to work better than the previous right?

Today was my first toilet bowl experience. A little un-nerving since the crashes with the last update.

Yes yes calibrate calibrate....  but lately all I do is calibrate instead of flying confidently.

If people have to be wary that its a coin-toss whether the updates actually improves or degrades the APM performance. They will all be flying NAZA instead of 3DR.

One more bad update and I will no longer be an advocate for the APM and the community.

Views: 1862

Reply to This

Replies to This Discussion

Nothing is wrong with recent updates.

Regarding that link, people who fly cheap HK clone boards on exotic-geometry homemade aircraft (Y6) are likely to have a few issues, wouldn't you think? And as can be seen from the last post by the OP in this thread, it appears that his trouble was as suspected calibration, not software/firmware.

In nearly 4 decades in a rural fire/rescue service I responded to a total of ten full-size airplane crashes. Every single one turned out to be due to pilot error.

Nothing is wrong with recent updates.

I would love to share my experiences. I have an SK-450 with an APM 2.5+ and the 3DR UBLOX+compas module. It has over 300+ hours of flight time, over 160+ of them autonomous, on firmwares from 2.9.1-3.1.2. Of all of the firmwares, 3.1.2 is the best performing one I have EVER used. I thought that 2.9.1b was good, then I updated to 3.0.2. I thought that was an improvement, but 2 weeks ago, after 92 hours of autonomous flight time on 3.0.2, I upgraded to 3.1.2, and the results were remarkable. The control loops are so perfect that I can now let my quad stay in absolutely one spot, with no fingers on the sticks, in Stabilize mode alone. This is with no vibration dampening on my APM, no expensive high quality parts, and a $250, middle of the road quadcopter with an APM on it.

Keep in mind, it takes a month, at LEAST, to set the APM up absolutely perfectly. Even with past experience, I expect to take 3 weeks or more to tune my new hexacopter to absolutely PERFECT autonomous and manual flight. Even though it might fly well with one or two flights of tuning, it will take a lot more time and fine tuning to get something absolutely rock solid in every possible way, which is more effort than some people are willing to put into their system. For example, note that there is a slight change between the 2.X.X and the 3.X.X in your flight tunings, and 3.1.X is slightly more sensitive than 3.0.2 to magnetic interference. If you're coming from 2.X.X, 3.X.X is a lot more sensitive than 2.X.X to magnetic interference, and you will need to compensate with your hardware positioning accordingly.

With every update, expect to have to re-tune a bit and adjust some settings. The updates generally change the way control loops react, and as they become more powerful, they become more susceptible to very minute hardware and software configuration issues that weren't prevalent or visible before. I had to move my compass up by 3/4 of an inch when moving from 2.9.1b to 3.0.2, That little movement took my interference from 103 to 23, and it made a big difference in keeping the copter level. Keep trying, check your vibrations and compassmot, and then mess with your PID tuning. In particular, people usually do not mess with yaw tuning as much as they should, which makes a massive difference, especially when experiencing "toiletbowling." Your tuning will make all the difference. Best of luck, keep us updated!

1) This stuff is not certified-grade aviation, where *everything* is tested ad nauseam, and certified mechanics are maintaining those systems.  Apples and oranges.  

2) Saying "Nothing is wrong with recent updates" is a blanket statement, which given that the topic is a complex system, is inappropriate, even if you're right.  

It was a design fault in placing the compass inside the APM. Thats why 3DR now make the compass as part of the GPS and they are both external and intended to be placed away from all EMI.

Its your fault in the sense, you and only you can fix this, by buying an external compass and put it in the right place away from EMI. You cant expect the software upgrades to fix this physical problem.

There are a few guides here about how to cut the trace on the board of the internal compass and how to hook up an external one. Its not terribly difficult, but does take a bit of time.

The quad I mention on the next page is an APM 2.5, and it had problems going to the 3.X.X series of code because that series uses the compass a lot more for nav and stabilization than the 2.X.X code did. That's the reason 2.8 worked for you, while the newer versions didn't. I had some hesitation moving to 3.X.X at first, but after changing to an external compass, I've found it to be absolutely incredible, and well worth the extra 15 minutes of work to install an external compass. That's all it took to install it, including disabling the internal compass by cutting the trace.

That issue was hardware, and they made amends by coming out with a revised version within a month to fix the problem (less than a week after the issues experienced were specifically attributed to the relative proximity of the compass to the PDB and batteries) and making sure the word got out about the need for an external compass. Give it a try, I'm sure it will work well! And when paired with a UBLOX GPS.... less than 3 feet of drift on average, even in 26MPH winds. Good luck!

Y'all really are going to talk me into it....

You should try it! I cannot even describe how incredibly fun and valuable this experience has been to me. It's landed me a research lab as a freshman in college, and plenty of job offerings, as well as my quad being one of the most fun things I've ever owned :) If you have a mediatrek GPS module, I would recommend upgrading to the UBLOX + Compass module. I just upgraded a month ago, and it's significantly more accurate. I mean, the difference between 10-20 feet of drift and 3 feet of drift. If you already have a UBLOX, get a standalone compass. I'm actually using the standalone compass with the UBLOX GPS lol. I already had the compass from when I used a mediatrek GPS, and just never switched over. Just make sure that if you get the standalone, you order a 4-pin df13 connector to cut and solder to your compass.

Just a note for installing them: The 2 biggest things to keep in mind are A) ensuring that you connect the correct wires to the correct pins and B) Making sure you have the correct orientation; both physically and selected in the mission planner. Make sure autolearning is enabled, and auto-declension,  then fly around for a few minutes. After your first few flights, it will be locked in :)

And if you want a cheap, incredibly robust, 10 minute fight time trainer frame for $250, PM me. I have a hobbyking setup list that I can send you :) Never had to replace a part, even with tumbles from 50 feet and 300+ flight hours on it. So much for cheap Chinese stuff not being durable! Keep us posted, hopefully you'll be flying soon!

+1 totally agree with you..

Nothing is wrong with recent updates.

If nothing is wrong could you please take a look into logs and tell me why ArduCopter did its best to perfectly stabilize INVERTED flight?
It did it fine but something is definetely not right. 

Well. Internal compass is a less than optimal solution.
By placing power away from APM, and running compassmot, is usually no problem making it work fine.
Start by making it fly well in stabilize.
Do autotune once you are sure you don't have motor sync problems.
Then do experiment with loiter.
Let loiter stop a running speed flight with heading North, then do the same while heading north-west and so on. There should not be no toilet bowl.
Post .log of you are unsure of what's going on.
Finally: usually the releases are fine, when we see bugs, it's mostly non critical features, not something that causes a sudden crash . As always, before using advanced, new, features - test.

Yaw often is a trouble.

is there a scientific approach posted on-line which can help to understand how to tune the yaw PIDs?

The absolute best I've seen for overall tuning is here, and I've always used it. It makes life a lot easier :) And if you have a smallish quad that could fit on a suspended stand, or even between two beds in a dorm room, it makes visibly seeing the best PID's incredibly easy and incredibly precise. It's how I tune my stabilize control loops, and I can easily get fairly solid PID's tuned out in an afternoon and 4 batteries. It's so effective I've actually completely scrapped working PID's to retune with that method, and being able to see exactly where the breaking point between the right amount of a value is and too much is pretty amazing. It's like a sheer cliff. The wiki pages will be a lot more helpful for tuning the autopilot side than that guide, but Dave's guide is the best place to start.

As for yaw, it basically follows the behavioral effects of tuning roll and pitch PID's, but it's a little harder to see. You can attempt to use the test stand, but it doesn't quite give you the same response as the roll or pitch does. My recommendation is to go out flying and have channel 7 set up to your yaw PID'S, and keep the quad low to the ground, then slowly turn up the rates in the same manner described for the pitch/roll in Dave's guide. That's how I personally tuned mine out, and while it took a bit more time, it was easy to correct any problems by staying low so you can drop the quad in an instant, and by keeping a sharp eye on how well it held its direction. See what other methods you can find! Good luck!

Reply to Discussion


© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service