Just checking when 4.0 beta will be released for copters. I am asking to try to get ready with a test rig and give ton of feedback
The right word is I think is “imminent”.
As I see we can expect the first beta within a couple of weeks.
Great, looking forward to it.
Yes, we’re getting very close. “Complex Fences” (aka stay-out zones) have been completed so the Vibration failsafe PR is the last critical things we need to do before beta testing starts. I have an S&R event next week which will make support difficult so I think the beta will start in about 2 weeks (between Oct 14th and 18th).
Thank you, in time to still get some good weather, great.
I have started testing 4.0 Dev on a helicopter and passed the first stable mode test flight using MatekSYS F765 wing FC, which has just released details:ArduHeli 4.0 FW support for MatekSYS F765 wing
@rmackay9 BTW what do you mean saying beta testing? Are there people (internal devs?) who will test a part of the feature scope each? Or do you mean the beta goes public and you will listen for a random feedback?
Is there a top-level feature list of 4.0 that needs to be tested?
Most of the functions are tested in simulator and in flight before committed. Some of the hard core enthusiasts and testers are already flies with the development branch of the code.
Beta is a feature freeze branch of code, and only fixes and important backports allowed into it. The beta test relies on the community, since there are many different flight controllers and aircraft configurations, which needs to be tested. Beta test versions are often contains bugs, sometimes severe ones that can cause crash on some configurations, so they are “fly it only at your own risk”…
Beta feedback is not random, there are many people who willing to do it and giving proper feedback. (it is called the community :D)
Ok, but how does this work? What is the workflow for the community tester’s feedback not to be random?
How to see if I test something that wasn’t very tested yet, or I just retesting a thing that was already checked by 100500 other people?
If you test, you will test the functions that are relevant to you. The diversity of the tester base provides the necessary spread of testing.
Indeed, there will be rarely used functions that are less tested than common functions. But this testing method is the best one that can be implemented with volunteer testers and literally no budget. Even Microsoft is using the same method with their insiders program…
https://github.com/ArduPilot/ardupilot/pull/12270 ? Pretty please.
Last weekend I flew my S900 (Hexacopter) and my Daya H680 (Quadcopter) with the recent built of 4.0 on a 3DR Pixhawk. Alltogether around 40 flights 10-15min per flight.
The only thing I can say: FANTASTIC and BRILLIANT WORK to all developers.
I couldn´t find any issue, the flights seemed to be much more stable and smooth than with 3.6.10. I only updated it without changing any parameter. Everything which is important for me worked. Some new functions (i.e dynamic notch) I didn´t test because I was not sure how to configure.
I am very confident about 4.0. So if you want me to participate on beta tests, please aks me and tell me what specific test I should do. I have 8 different Pixhawks/Pixkacks/Pixracers…, so I could do different test if you want.
Thanks again for the fantastic work
I have one question with arducopter 4.0 passthrough telemetry. on a 3DR Pixhawk do I still need the special telemetry cable or can I now also use (analog Pixracer) a standard cable with configuring halfduplex and inverted in the parameters?
With a 3DR Pixhawk you’ll still need the cable to invert.
Anything based on a STM32F4 (regular Pixhawks, Black Cubes, etc) besides Pixracers will still need the serial inverter cable. Pixracers have the inverter onboard.
Anything sporting STM32F7 (and H7) like Pixhawk4, Kakute F7, Orange Cube, etc, are capable of inverting the signal internally.
But, to add some confusion, there are certain FrSky receivers that expect a straight uninverted serial
Without https://github.com/ArduPilot/ardupilot/pull/12270 the harmonic notch is only static, drastically reducing its value (except perhaps for helis) which is why I am so keen to get it in . If it goes in I’m ready to write some extensive wiki pages documenting its use as, granted, it’s a bit of a dark art at the moment.
@andyp1per, yes Leonard is also very keen that the final part of the notch filter go in as well so I think it’s very likely we will include it in 4.0.0. Whether it makes it into 4.0.0-rc1 or -rc2 I can’t promise immediately…
That’s great news!
I would argue that it’s the “next” part rather than the “final” part, as there is always https://github.com/ArduPilot/ardupilot/pull/11886 which I am using to drive the notch, but probably too much of a stretch for 4.0!
And then there’s dshot …
… and double notches …
To someone interested in some testing details:
We use “Continuous Integration” aka CI tests that test every single line of the source code against expected results. Right now those tests cover 50% of all source code lines against SITL based simulations. The results are here : http://firmware.ardupilot.org/coverage/
On top of that the developers and beta testers perform flight tests on real hardware.
This is just a small part of the quality measures that the ardupilot community uses
I have tested the flight：