https://ardupilot.org/copter/docs/tuning-process-instructions.html
During decent your aircraft is struggling with yaw and starts using 100% of your output to compensate. This will impact your available thrust. I can’t see much without the log though.
I find this a particularly interesting discussion. I’m not sure that I agree with the assertion that VRS cannot occur in multirotor aircraft. I don’t believe that due to the design of the multirotor propeller precludes them from encountering VRS. This is becoming a more prominent discussion in the technical community due to the impacts it could have on multirotor eVTOL aircraft as now there are lives at stake. Here is a link to a video that presents research on this topic directed more toward eVTOL aircraft.
VFS Lecture on Vortex Ring State for eVTOL Aircraft - YouTube
But fundamentally I don’t think multirotors are exempt from this phenomenon. It is possible that they can better deal with the situation because the propeller RPM is changing which varies the wake clears the instability in the wake. Whereas traditional helicopters maintain constant rotor speed and thus have to change the flight condition to escape from VRS.
My $0.02
you think we’re just gonna trust the guy who contributed to the success of the v-22 osprey??
I haven’t seen something like this for a while:
We have an underpowered aircraft to start with given your hover throttle appears to be 52.3% at full battery voltage and drops to something like 62.3 % near the end of this flight. So we are in that range where any large attitude control output will cause your aircraft to drop simply based on it’s mechanical design. In fact you can see thrust loss warnings in normal flight and two instances where the aircraft can’t maintain altitude before your landing.
You can see the aircraft losses yaw control and the large control output in yaw does mean you only have 50% thrust.
I would need to see the aircraft but you have some issues here.
thats neat, comparing CTUN.Alt / DAlt to get a sense of when the aircraft is underpowered. but, is it not acceptable for Alt to dip below DAlt for ‘aggressive’ maneuvers? why cant it ‘give up’ a little on yaw if it means crashing?
I always find discussions of VRS amusing, not because I don’t believe in VRS but because I never see any of the issues that are attributed to VRS. I think people just give up and believe that they can’t do anything to mitigate the behaviour they are seeing because “it’s VRS” - it’s not true, good noise reduction and tuning are your friend here.
im sure people would want to believe (at least i do) that its just a tuning problem, because it seems much harder to fix vrs. how can tuning explain how i can continue flying normally after bouncing back up after a hard landing?
Easy. You lack the excess thrust to fly out of the condition until it’s disrupted by the ground. The turbulence (whether it’s VRS, wake turbulence in the vertical, settling with power, etc) dissipates rapidly, and you can take off again.
you’re saying the initial problem wasnt tuning, but vrs et al?
No, the problem is the lack of thrust for the flight condition, regardless of how you define it.
I’m not sure how much mileage you will get out of tuning with the thrust/weight condition you have. Another way to view what @Leonardthall posted.
CCW motors only shown so it’s not so busy (CW motors were commanded low to attempt the Yaw demand):
Higher values prioritize attitude over throttle. The Acro flying FPV guys set this high.
HI @bnsgeyer,
Am I being told by a heli dev that multirotors resemble helicopters in flight? 
(in joke for those wondering)
I think this depends on how relaxed the definition of VRS is. In the presentation you link to the definition is expanded to include any time you descending such that the outflow of the propellers is re-circulated through the propellers. We have lots of names for this such as descending into your own prop wash for example. Multirotors most certainly do get reduced thrust and more notably, increased turbulence and compromised attitude control.
I think it is convenient for researchers to relax this definition to help get some funding from all these flying cars we have now…
My understanding of a fully developed VRS is where two donut shaped vortex’s form on the helicopter disk. One circulates up outside the rotor disk and the other rotates up along the fuselage. Both have down flow in the outer portion of the disk.
Multirotor’s use fixed pitch propellers with increased cord and pitch as we get closer to the hub to maintain a consistent lift profile along the blade while helicopters carry most of their lift load on the outer portion of the blade.
A helicopter has a relatively constant lift profile around the disk that facilitates the support of a full vortex ring that surrounds the disk. A normal multirotor configuration has independent, closely spaced disks. This means that any vortex that forms can’t surround a single propeller disk.
So yes a standard multirotor propeller can support at lest the outer vortex ring. The combined effect of multiple propellers (especially an octo) placed in a ring could approximate a helicopter disk and theoretically be used to demonstrate a fully developed vortex ring state (an ugly one).
However as soon as any part of the vortex starts to form attitude control start so suffer and each disk starts to vary significantly. This variation destroys an chance of anything resembling the clean and powerful outer vortex I associate with VRS.
Take Home Point for Users: You are not suffering VRS, you have a problem with your tune, low battery, not enough thrust overhead, or didn’t realise you switched to land.
I look forward to seeing that first log showing a multirotor with all motors sitting over 85%, no battery issues, with an increasing decent rate immediately after being able to hover below 75% throttle.
oh, haha! attitude not altitude! d’oh! (i thought it was altitude)
If you want to see a craft bounce on landing set it to 4 
i think the main point of “contention” in this thread, is that the lack of thrust is (not) due to vrs or something fancy like that.
It’s not. There is no contention. You simply lack thrust on that airframe. We all agree there. It’s a physical limitation and it becomes apparent when you put the craft into a demanding situation (potentially exacerbated by poor tune…but that’s neither here nor there when thrust to weight is under 2:1).
If there is a point of contention, it’s that you’d either:
A: never have enough RPM to fly (vertically) out of it
B: always be able to fly out of it given enough RPM
Edited to say RPM because thrust implies force, where RPM implies commanded thrust whether or not the force is actually produced.
But I think we are agreeing that B is true as well.
This isn’t a tuning problem as such, it is a combination of poor yaw control and very little thrust overhead.
We currently prioritise roll, pitch and yaw over thrust. This means that a full yaw output can reduce the thrust to 50%.
The ATC_THR_MIX_bla parameters impact throttle priority over roll, pitch and yaw at low throttle, not at high throttle.
When I wrote the motor mixers I thought long and hard about placing yaw priority above or below thrust. This is much more complicated than it seems because it has a significant impact on motor loss redundancy.
this would fully explain the behavior of my craft. is there a parameter to change this to 60%?
Full yaw control would mean (for example) all the CW motors at 100% throttle and all the CCW motors at 0% throttle (or whatever min throttle is set to). Thus reducing the overall thrust to 50%.
if we let the other motors stay at 20%, then it would be something like 60% thrust
