Thank you for the clear explanation Sam!
I didn’t notice that there has been short throttle spikes in the acceleration-phase of descending.
My first try has been with TECS_SPDWEIGHT=2, but that leads to a very bad climbing-phase.
I will have another try with soaring-switch in loiter-soaring-position when descending starts.
Cheers
Willy
Just entered a PR that implements a CROW mode switch option, three position, LOW normal setup of weights and progressive per the wiki…MID force progressive…HIGH disable aileron crow movement (ie -W1 forced to 0)…SITL tested…awaiting Tridge and Peter Hall’s review
Logging in to ardupilot.org for the first time in a year or so, and thrilled to see many familiar names active on ArduSoar! Hoping to take latest version for a spin in one of my gliders later this summer. In the meantime, I have been bitten by the slope bug, starting first with Ahi and Alula. Has anyone worked on “aerobatic scripting” in Ardupilot? Learning aerobatic moves from watching the human pilot, and smoothing out those moves into a library of loops, rolls, wing-overs, inverted flight, etc.? Or maybe a scripting language for an aerobatic routine?
Kelly
@Samuel_Tabor any thoughts on non-Pro Radian? I.e., rudder-only ship? I’ve got the standard, the Pro, the XL, but thinking the world record ArduSoar altitude should be attempted on the classic Radian. 
My concern is not too many folks have flown ArduPlane rudder-only over the years, so don’t know if steep bank angles and other aspects of smart soaring are likely with rudder-only. I see a bunch of folks are on the Phoenix 2400, but with China shipping being what it is, I’m looking at my existing hanger.
Kelly
Welcome back Kelly!
Yes the non-Pro Radian is a good ship, most of the early ArduSoar work was done on one and I still have one here. Rudder-only can be made to work well, there is a specific setting RUDDER_ONLY for it now. I’m not sure what this does in addition to setting KFF_RDDRMIX and YAW2SRV_DAMP, which I use successfully. The altitude record using ArduPlane soaring feature is currently over 11,000ft.
Sam
Thanks Sam. Re the record, it’s not official unless there is a video, and/or FAA fine, right??? I think I should be wearing ADSB out when I attempt the new record, right?
Kelly
Sam, on a related note, ATXHeli recommended (on RCG thread) the Omnibus F4 Nano V6 as a lightweight (5g) FC that can run latest ArduPlane, including ArduSoar. I have seen various posts lately on conditional compiles enabling/disabling different portions of ArduPilot, and latest release vs private releases you are compiling for various FC’s. Can you summarize latest there? I am fine using your bleeding edge stuff, in fact prefer that, especially the new polar learning code. The F4 Nano V6 I believe has 1mb RAM, is that sufficient for the latest master with soaring, for your custom compiles, etc.?
Kelly
@Samuel_Tabor
Hi Sam. Switching back to EKF2 from EKF3 solved several problems. On startup the error message on compass disappeared. I don´t have one. And it also started thermalling when sufficient updraft was found instead of RTL. As there was not much updraft I will test that later, wheather allows. I had switched to EKF3 because of AHRS arming problems, now solved.
Maarten.
Hi Kelly,
Yes the Omnibus F4 Nano V6 is supported even though it has only 1MB flash. To allow this it is EKF3 only (no EKF2) which has not had as much testing with plane, as Maarten found out, but is maturing quickly. It logs to onboard dataflash rather than SD card.
The Matek F765-Wing controllers are another good choice if you can afford a little more size and weight (26g).
Sam
Hi Maarten,
I’m glad that worked. I’m going to have some EKF experts look at your log to see why the EKF3 wasn’t behaving.
Sam
Thanks @Samuel_Tabor. I already have a couple F405-Wing’s, and a few of the older Omnibus. I’ll make a decision on FC soon and get going on the Radian assembly. Am I using latest Master, or one of your builds?
Kelly
Hi Kelly,
Another one to consider is the Matek H743-Wing - more powerful processor and CANBUS. I just found out about this one. My builds are the same as master (just a little behind) except mine have the glide polar learning. So it depends if you are interested in testing that. I have good glide parameters for the Radian so you may want to skip polar learning and use those instead.
Sam
Unfortunately, it will be probably another month on the H743 production availability (just a SWAG, not official) since they are changing to a different baro…the F405-Wing or F765 are both about the same size, with the F765 having 2Mb semi-future proofing it a bit…all of them use the BMP280 baro…might be worth the wait for the H7 but it will be more expensive…the DPS baro is supposedly better than the MS5611 that the higher end boards use…
Thank you both. I think I’l either stay with my existing F405-Wing (updated to master last night) or try the Omnibux F4 Nano V6.1 at $30 and quite small. What are the benefits to ArduSoar fans of CANBUS on board?
Kelly
I got my F765-Wing soldered up, installed Arduplane 4.0.5 firmware and Qgroundcontrol 4.0.8. I have flight modes setup, but when I went to setup a soar_enable_ch I couldn’t find it on Qgroundcontrol. As a matter of fact the only soaring option I can find is soar_enable. Is the problem in Qgroundcontrol, or do I need a more recent release of Arduplane?
set SOAR_ENABLE, reboot
That worked a treat. Thanks for the quick reply.
Hey Mogli! i’ve been having the same issue with the controller attempting to nose up, gain altitude and microstalling. How’d you fix it?
I already had TECS_SPDWEIGHT at 2.0. Did you just upload the new firmware Sam made?
Also, @Samuel_Tabor, I have problems updating the cubeOrange from MissionPlanner, i usually have to download .apj’s from firmware.ardupilot.com. This isnt in the latest 4.0.5 stable release right? Is there a .apj i can grab from somewhere for the cubepilot orange so i can try all the new learning stuff out? I’m flying my own design so itt’l be interesting to see what aero data it coughs up and how it compares to what CFD gives me.
Hi ohitstarik,
I have replied in the other thread regarding the glide control, in summary it’s fixed in latest/master.
Correct, it’s not in the 4.0.5 release. Most new stuff is available in latest/master, except the drag parameter learning that is only available from source or in my builds here
Sam
1 Like
Hi Sam,
After many very windy days (perfect for the slope ;o) , today in the morning we had silent conditions. So I had a second try with your polar-learning-feature. And it seems that it worked properly, because _B is changed from 0,031 to 0,093 and _CD0 is also changed al little bit.
I’m using TECS_SYNAIRSPEED = 1, but TECS_OPTIONS = 0, because I just read about that parameter in the other thread.
You can find my log here https://www.dropbox.com/sh/z7pfgihoft435ne/AADy8WcR8s5EKIGVxuzl2UWka?dl=0
I hope that helps you to verify your feature.
Willy