Most stable 6.8?

[Trnquill]

What is the most stable build of 6.8 firmware? Is there some know bugs in some of the builds or should I just grab the latest nightly (or was it weekly) build?

I started playing with AutoQuad last autumn and used the latest 6.8 available at that point. I had a mid-air flip of death and was adviced to go with the stable release candidate 6.7 firmware. So, I did. However, I have never been that satisfied with performance of 6.7. When changing firmwares I switched multiple times back and forth between 6.7 and 6.8 and confirmed that:

-Altitude hold in 6.8 is rock solid at default PIDs
-I have never been able to tune altitude hold that perfect with 6.7. The hex is constantly flying up and down or even struggling to keep steady altitude at all without manual override with throttle stick.
-Attitude hold in 6.8 is rock solid at default PIDs. The hex is just super steady in the air, with no odd noises or vibrations
-With 6.7 firmware I had strange oscillation, twitching and "twittering" motor sound from the start. Reducing vibration level of the frame eliminated twitching but twittering and oscillation remained. I can not tune them out.

So, I would be very eager to try out 6.8 again! I just don't want to crash the hex only because of some minor software issue that "everybody knows"...

My hex is quite large, a tad over 90cm from motor to motor. 17" props, 380kv motors, 4S batteris, around 4,5kg AUW.

[hogster]

I am also interested to see the responses to this thread, as I'm about to put an AQ on a large X8 platform, and want to start from the best point possible (software wise).

Do you have any log files from the flight where the Hex crashed that you could post here? Trying to diagnose what happened on that flight may well be the best place to start ...

Cheers, David

[Trnquill]

[SuperSharpShooter]

I'm interested in the answer to this too.

I want to know which build of the firmware is best to use for a non-DIMU board with ESC32s running r49, and also if there's a recommended FW build for running with simonk ESCs. Is there any danger or known issues running r271 or r346 with an AIMU board with either of these setups?

[epyonxero]

It would be nice if there was an open FW discussion thread where we could read about the latest experimental versions.

[Trnquill]

[Max]

Builds with known major bugs get taken down. I think it has happened a few times, but it's rare. However, some versions might have gotten very limited testing. The longer it's up and no one reports bugs, the safer I'd consider it, but we really have no data about who has tested what or how thoroughly (which would be nice). Some code gets more testing before being committed than other code (again, no real measure of that is available). OTOH, smaller issues get fixed along the way, or are improved, so the newer versions will tend to be better in those regards (but might introduce new issues).

The latest builds of 6.8 and 6.9 branches are considered beta quality. I'd probably go to 6.9 at this point. We'd certainly like to know about people's experience with it. I'm not sure if fixing anything in 6.8 specifically would make sense at this point, but maybe. The 6.8 builds can be a reference point for troubleshooting potential problems with 6.9 (eg. "it does/doesn't do that in 6.8").

The change log you link to is the primary source of what has changed, for everyone. We try to summarize released versions in the docs, but it's hard to keep up. You can relate the change log to the builds based on the "rev" number, which appears in the .hex file names as ".rXXX". In the change details view you can also check buildnumber.h to see the corresponding build number, which is the ".bNNNN" in the .hex file names. Latest experimental (trunk) is v7.0. Previous versions are archived in http://code.google.com/p/autoquad/sourc ... 2Fbranches

HTH,
-Max

[Trnquill]

Thanks alot! So, I'll go with 6.9 then.

[hogster]

Out of interest, when is a build considered 'stable'? After it has been around X months, or after it has been tested by a number of people who have reported back no issues? I like the idea of having a set of test procedures that have to be passed before a build can be declared stable ... Although I imagine ironing out bugs like the one which caused Trnquill's crash is no easy task!

Thanks David

[Trnquill]