This is on a 75mm quad (BetaFPV frame, 0720 brushed motors) and Beecore V2 F3 FC, 550mah 1s. Flying in rate mode (Betaflight).

Have a problem that on tight turns, sometimes rolls (inwards) out of control and crashes. Watching the FPV recording and OSD info, typically pitch and roll are both at ~30-35 degrees when this happens. It is an instant completely out of control roll. If you aren't really pushing it round the corner, doesn't happen. Exactly the same thing is happening on a friend's 85mm (same setup but 8520 motors). Flying outdoors in a wood, but no RSSI problems.

Initially thought this was only at low battery voltage, but actually it happens with a full battery too.

Then I discovered that turning on air mode (not the permanent switch in configuration, but set to enable air mode in modes when armed) makes the problem go away. Friend tried it on the 85mm, and it didn't. Or at least, it sort of did but it could still tip backwards out of control in a turn, but actually, now doesn't do it as fast, and if you react quickly, you can catch it.

The difference was however, he had air mode permanently on in the configuration page, was not set in modes.

I tried the same (removed the mode, switched it on permanently). Now the same thing as him, seems to tip backwards, but not as fast. Seems air mode (permanent) and air mode (mode) are not quite the same.

So, very strange:
 1. Why is it happening at all.
 2. Why does air mode affect it.
 3. Why does air mode affect it differently depending on how it is switched on.
 4. Has anyone else seen this.

Airmode is the same whether permanently switch on or toggled through a switch. If you fly without air mode, and you go to zero throttle, you will loose control. I wonder if this is happening to you?

Air Mode is the same whether permanently on, or on a switch.

Basically Air Mode allows you to keep control even when at 0 throttle. With Air Mode: OFF, when you go to 0 throttle your motors/propellers stop spinning, which means you have no control over the quad. With Air Mode: ON, when you go to 0 throttle your propellers keep spinning a certain percentage which allows you to still control the quad. They don't spin enough to take off, or even maintain altitude, they spin just enough to provide some control of your pitch and roll. For example, if you are coming down in a dive you will be at 0 throttle, but still be able to control the quad because the props are putting out that small amount of lift due to Air Mode. Turn it off, and when you are at 0 throttle you will not have that smooth control of the quad, it will simply fall according air resistance, and it's momentum.

I fly all my quads with Air Mode = ON, all the time. From my 600g+ 5" Freestyle quad, to my 59g Micros. I like ALWAYS being in control of the quad's attitude, regardless of my throttle position.

If you are losing control when at 0 throttle, try with Air Mode ON. You also may want to try raising your idle speed a little to keep more control of the quad when at 0 throttle. If this is happening even when you are on throttle then I'm not quite sure what would cause a quad to lose control like that. (I'll let a person more experienced with Whoops answer)

Are you flying in Acro mode? Because Horizon mode will keep level flight turning normal maneuvers, but if you move the stick all the way over to one side it will allow the quad to flip/roll, so if you are trying to make a very tight turn it might be thinking you are trying to roll.

(19-Jul-2021, 05:10 PM)MrSolo Wrote: Basically Air Mode allows you to keep control even when at 0 throttle.  With Air Mode: OFF, when you go to 0 throttle your motors/propellers stop spinning, which means you have no control over the quad.  With Air Mode: ON, when you go to 0 throttle your propellers keep spinning a certain percentage which allows you to still control the quad. They don't spin enough to take off, or even maintain altitude, they spin just enough to provide some control of your pitch and roll.  For example, if you are coming down in a dive you will be at 0 throttle, but still be able to control the quad because the props are putting out that small amount of lift due to Air Mode.  Turn it off, and when you are at 0 throttle you will not have that smooth control of the quad, it will simply fall according air resistance, and it's momentum.

Just one slight addition to this for clarity. If you have Air Mode switched OFF but you also have MOTOR_STOP switched OFF, even at zero throttle the motors will keep spinning at idle speed rather than stalling and stopping completely, but the difference is that the FC will have no authority over the motors, so they are just spin at the set idle speed. With Air Mode switched ON and the throttle dropped to zero, the motors will not only continue spinning at idle speed, but the FC still has authority over the motors so if any need to be spun up or down slightly to adjust / correct the attitude of the quad, the FC can still do that with Air Mode switched on. It can't do that if Air Mode is switched OFF.

Thanks for all the replies. Where to start?

Yes, all acro.
I know what air mode is supposed to do, but it definitely seemed to me to behave differently as a mode, than if permanently enabled, but I won't be dogmatic.
This wasn't on low throttle, far from it, it happens on moderately fast corners (outdoors) taken as fast as possible. Typically pitch and roll were both 30 to 35 degrees.
Air mode (as a mode in my tests) did seem to solve it, BUT at a cost. It makes the quad feel completely under powered, it is night and day different. It is more that just needing more pitch/roll, it is more difficult to keep it flying - it just won't keep flying with as much pitch or roll. Doesn't do this with air mode off.
The betaflight was 3.4 (with some mods), but swapped it to an unmodified 4.0.6, being the latest for F3 boards. No difference.

Not at all clear what is going on, but it is basically useless as is. Is it firmware or hardware? I'd be happy to try a different FC (FrSky) if I could find one that had a buzzer output (there is a BetaFPV one but it needs a separate receiver), but they are harder to find now - most seem to have gone brushless.

airmode does give much more boise to the system, do you have a capacitator attached? is it still well connected?


bf3.4 is pretty old, what parts are you using? single escs? 30x30mm hole diameter?
there are a lot good fcs out there, dont hide solder the receiver :-) frsky might be expensive or overpriced choice (not sure?), betafpv is a quality level i wouldnt suggest. i believe ghe betafpv product you saw might be an all-in-one; fc and esc on one board? i dont know an single fc from them.

(26-Jul-2021, 12:46 AM)hugnosed_bat Wrote: airmode does give much more boise to the system, do you have a capacitator attached? is it still well connected?


bf3.4 is pretty old, what parts are you using? single escs? 30x30mm hole diameter?
there are a lot good fcs out there, dont hide solder the receiver :-) frsky might be expensive or overpriced choice (not sure?), betafpv is a quality level i wouldnt suggest. i believe ghe betafpv product you saw might be an all-in-one; fc and esc on one board? i dont know an single fc from them.

I used Betaflight 3.4 and 4.0.6 (last available for F3) - no difference at all.

No ESCs - brushed.

Since loaded Emuflight 0.1.0 (also last for F3). Problem so far has not reappeared with or without air mode (all other settings as for Betaflight). However, due to inclement weather, not so far been able to test it over the same course, so it is not completely conclusive. Hoping to do so in the next few days.

FC choice for brushed seems very limited nowadays, and some of them only run Silverware. (Not a bad thing as such as far as I read, but again more stuff to learn, so I never tried).
I have no real advice to offer, but wanted to confirm that I have seen similar behaviour on my Beecore V2 75 mm whoop. It is pretty irregular, though, as in "I cannot really reproduce it but then it gets me by surprise again". Mostly in acro mode WITH airmode on, sometimes I can recover before I crash, sometimes not.
I always thought this was due to my PID tune being wonky or filters going crazy, given my non-extant tuning skills... Played around some with those, with little success.

PS: no ill performance effect of airmode for me. BF 4.0.6

(01-Aug-2021, 06:43 AM)robbie Wrote: FC choice for brushed seems very limited nowadays, and some of them only run Silverware. (Not a bad thing as such as far as I read, but again more stuff to learn, so I never tried).
I have no real advice to offer, but wanted to confirm that I have seen similar behaviour on my Beecore V2 75 mm whoop. It is pretty irregular, though, as in "I cannot really reproduce it but then it gets me by surprise again". Mostly in acro mode WITH airmode on, sometimes I can recover before I crash, sometimes not.
I always thought this was due to my PID tune being wonky or filters going crazy, given my non-extant tuning skills... Played around some with those, with little success.

PS: no ill performance effect of airmode for me. BF 4.0.6

PPS: 550 mah batteries on this setup sounds heavy! I fly mine on 230 mah HV stick batteries and get 2:50 to 3:30 or so flight time. Maybe you are getting yaw washout because the quad cannot catch its own weight coming out of the turns?

(31-Jul-2021, 03:25 PM)nmw01223 Wrote: I used Betaflight 3.4 and 4.0.6 (last available for F3) - no difference at all.

No ESCs - brushed.

Since loaded Emuflight 0.1.0 (also last for F3). Problem so far has not reappeared with or without air mode (all other settings as for Betaflight). However, due to inclement weather, not so far been able to test it over the same course, so it is not completely conclusive. Hoping to do so in the next few days.

there would also be the bf 4.1 performance edition for f3.

did you enable a whoop preset?

as robbie wrote above, what happens could also be a yaw washout, get better batteries, increase d-term values on pitch and roll and p-term for yaw can help there. yaw washout is mostly a power to weight related issue.

Whoop preset?

Don't think it's yaw washout, at least by my understanding of what it is. Been all over the place on PID values, had no effect either way, making me think it wasn't PID related. Yes 550mah gives me 4-5m flight time. Could try smaller ones if I find Emuflight hasn't solved it. Really need a logger attached to analyse what's going on.

Additional:

Emuflight helped but only partially. The 75mm occasionally does it with air mode off, seems OK with air mode on. The 85mm still did it a lot with air mode off, occasionally with air mode on.

Air mode does make both feel slightly underpowered.

if the power to weight ratio is to low, you wont be able to solve it by pids.

there are tuning presets in emuflight

After some looking into code (emuflight and betaflight) and making mods, I think I now know the issue - but there is no real solution. My theory follows, for those interested.

It starts with how the system works.

Basically there are 4 motors and each has a thrust/power range of off to full on, say 0 to 100 for convenience. There cannot be less thrust than 0 (ignore 3D) or more than 100.

Then there are the drivers: four - throttle, and three control loop demand outputs, pitch, roll and yaw. Throttle let's say can go 0 to 100 (no thrust to full thrust), whereas demand can go either way, so say -50 to +50, although actually there are no real limits, just some arbitrary clip limits.

In between the drivers and motors is the mixer algorithm, whose job is to map the drivers on to the motors. This is quite simple in concept - it adds them up, multiplying each for each motor by a factor (+1 or -1 for a quad) related to motor position. So for instance for pitch motors 1 and 3 (back) have factors of +1, 2 and 4, -1. For roll 3 and 4 (left), +1, 1 and 2, -1, yaw motors 2 and 3, +1, 1 and 4, -1. Throttle factor is always +1 for each motor.

With a bigger quad, throttle is typically ~40-50 plus or minus a lot of the time, and owing to the prop area the demand range span (highest motor output down to lowest motor output, ignoring throttle) is typically ranging between 5 and 20 (roughly balanced around 0) most of the time with occasional bigger spikes. One can see therefore that the sum of throttle and demands doesn't exceed the 0 to 100 allowed range and it all works fine.

Very different on a small 1s brushed quad. If it is flown in open space (a fast circuit for instance, which is what we do), throttle spends a lot of time close to 100, and on fast banked turns that demand range can get up to 70, 80, 90 or more, and the sum no longer fits in the 0 to 100 range. Something has to give.

Simplified, what betaflight does is (1) scale back demands so they alone always fit into the overall range span of 100, then (2) moves throttle so when added, the result is always within the 0 to 100 range. Unfortunately it gets it wrong, it assumes the max and min demands are always balanced around 0, and they aren't. The result is sometimes that it then clips one extreme more than the other resulting in an uncontrolled flip - the first symptom I saw. (This is on betaflight 3.4 - might have been fixed later, haven't looked).

In airmode the algorithm is slightly different, and asymmetric clipping (sudden flip) is less likely. However substantial reduction of high throttle is likely to occur when demands are high, giving that 'loss of power' feeling I noticed. Also, if demands have to be scaled back too far, control gets very wobbly. But anyway the fundamental problem still remains - the demands plus throttle need to fit inside 0 to 100, and - little quad in big space - sometimes they don't. That is a physical limitation, basically one needs more power than is useable to give that headroom. I managed to make the same thing happen to a bigger quad when flown very fast in a big space - uses up headroom again.

Emuflight does a bit better. The mixer code is rewritten and looking at it simplified, it among other things assumes yaw torque is proportional to throttle, so less yaw demand is needed at higher throttle. So now demands take less of the motor output space. But nonetheless, the fundamental physical limit remains and it can still run out of headroom.

I wrote my own mixer for betaflight which takes a slightly different approach, similar to emuflight in some ways but different in others, and it helps a bit - but in the end this is not completely solveable - one needs to match the quad to the space and use.

One thing that is a small help is 'props out', this is to do with how the demands add up, it gives a bit more headroom at high throttle - worse at low throttle, but that doesn't matter. Higher KV motors will also help - if they have enough torque, as would making sure CofG/P is on the prop plane (less pitch/roll demands needed), which it isn't with the battery at the bottom.

So, that is it really, it is a limitation I cannot see how to fully get round due to physical limits, though improvements can be made.

Hope this is of interest to someone.

Thanks for the little treatise, I found it very illuminating.

My impression still is that what you describe in really satisfying depth ultimately IS what many people refer to as 'yaw washout'. Your explanation is of course beautifully generic, whereas the 'yaw washout' term often refers to particular flight situations and the typical reactions that underpowered quads display in such situations. The main relevant property that people are talking about when discussing quads being more or less prone to yaw washout is the power to weight ratio, which I think is what you circumscribe by your 'little quad in big space' - when motors are saturated already, there is no room left for conducting controlled manouvres. Which, again, implies that 'changing the physics' of your quad, by making it lighter, should very likely reduce your problem. Back to that smaller battery thing :-).

a brushed quad does need escs too, they are just a different.
it is a brushed quad?

how much flighttime did you on the particular motors? good motors last about 1h of flighttime and need to be replaced than. if your motors need to be replaced, a low power to weight is one of the signs and following things like yaw washout.
you might just get a set of new motors.