Little note about the FM from the devs

[exil]

With danger of going off topic... In fairness I think most of my issues with the DCS Huey are an artifact of using a Hotas vs an actual cyclic collective setup. Cant afford one :( but I don't suffer the tendency for pilot induced oscillations in the sim Gazelle as I do the sim Huey. I do like and enjoy the Huey and its FM is great when not hovering, but I just feel the Gazelle's is better. The one thing I wonder about the real world Huey to DCS is the auto? I've never autoed a real one and suspect it doesn't have such an aggressive tendency to drop the nose as in the sim.

 

Totally agree, especially to the last sentence!

[Jester986]

I'll try all that and report back when I get home, right now I'm sitting in the hangar with nothing to do but post on forums...

[exil]

I'll try all that and report back when I get home, right now I'm sitting in the hangar with nothing to do but post on forums...

 

Thanks! Curious about your findings!

[FragBum]

With danger of going off topic... In fairness I think most of my issues with the DCS Huey are an artifact of using a Hotas vs an actual cyclic collective setup. Cant afford one :( but I don't suffer the tendency for pilot induced oscillations in the sim Gazelle as I do the sim Huey. I do like and enjoy the Huey and its FM is great when not hovering, but I just feel the Gazelle's is better. The one thing I wonder about the real world Huey to DCS is the auto? I've never autoed a real one and suspect it doesn't have such an aggressive tendency to drop the nose as in the sim.

 

My personal experience is even a home brew cyclic and collective makes DCS helicopter flying so much better. Also I think the center detent/springs in a joystick hides so much of the fine control especially in the Gazelle.

 

Oh and a cyclic and collective doesn't have to cost a fortune I'm sure there good, but.

 

My home brew WIP

Edited September 22, 2017 by FragBum
<typo>

[Jester986]

Ok, I hope you guys appreciate this because it was a lot of work. Start with some disclaimers. I do not have or care to look up the Gazelle performance charts, so I'm not going by is this 100% accurate but does it feel right and does it seem reasonable. Im doing this on an Oculus rift with a cheap TM Hotas, both will have effects on my opnion but what you are getting is does an experineced helicopter pilot with a marginal amount of experience in a similar (Astar/ Ecureuil 24 hours flight time) BUT DIFFERENT airframe feel like this module represents a real helicopter? If someone wants to link some documents though I'll compare against the real numbers. After going through this though I commend Polychop. There is some room for improvement, howver this module consistently "feels" right even if the numbers are a little off. Also I need to B!** that I still have no internet at home thanks to hurricane Irma, so I have to type this on my computer and work all kinds of stress inducing voodoo to make it talk to my phone and then copy and paste it to the forum over the cell network. Ok, lets grab a beer and get started.

 

All tests performed in 1.5.x in and aropund Sochi in the SA342M with 70% or 642 lbs of fuel, 3704 lbs empty weight. 4345 loaded weight, or 94% max gross weight as per DCS. Weather conditions were set at zero wind 15 degrees C and pressure of 29.92, except where noted. This establishes what is known as a stanard day and means any indicated altitude will match denisty altitude. Denisty altitude being the conversion that accounts for heat and pressure changes that affect aircraft performance. I do not know though if DCS follows a standard lapse rate. Temperature is supposed to decrease by 2 degrees C per 1000 feet altitude. So if DCS doesnt abide by that it'll skew performance a little.

 

Test 1: Exil's suggested test of taking off with no cylic or pedal inputs. Both with and without SAS on I get the same results. The aircraft lifts to a hover at about 65% torque and begins a left rotation, but holds a hover. Im able to land it by reducing collective and making no other inputs. This is not how a real aircraft would respond. I have a Hotas with a spring holding it centered so I'll Imagine this as a pilot taking off holding the cyclic neutral. The rotation from lack of yaw control and helicopter inherent lack of stability (I don't care enough to look up positive and negative dynamic and static stabilities right now but you can if you want, but you never use it after flight school) mean its going to start diverging in both the pitch and roll axis. Most likely resulting immediately in a skid caught on the ground followed by dynamic rollover.

My opinion: although this is technicly incorrect in a previous post I stated I appreciated how this module doesn't tend towards pilot induced oscilastions at a hover like the huey does. Dollars to donuts this is why. So ask yourself do you prefer a more realistic hover when you are trying to be in control, or when you let go of the controls? I suggest they leave this aspect as is. Moving on...

 

Update*** I tried this test again in 2.x on the Nevada map and the helicopter took off left skid low, caught, and rolled over just like it should.

 

Test 2: Three foot hover check. In ground effect takes 65% torque which sounds about right and 450 degrees T4, which seems low, but I'm not used to French helicopters and their crazy T4 gauges vs. American TOT gauges. It's probably what it should be.

 

Test 3: Take off, set 65 % Tq and begin take off roll from 3 foot hover. Slight dip with forward cyclic, shudder when effective translational lift begins at 14 knots and definitively in ETL somewhere between 24 and 30 knots. Was hard to say the exact moment it occured. Vy seemed to occur somewhere around 60 knots, again hard to say the exact moment, but most helicopters are between 52 and 60 knots. So far everything seems spot on.

 

Test 4: Set 80% Tq and 60 knots gives about 490 T4 and a rate of climb of 500 meters a minute which tanslate to Moon units of 1640 feet per minute. Sounds about right.

 

Test 5: Quick stops. Scale became a big issue here. I think the world feels slightly smaller and slower than real life. In flight school we would do the quick stop manuever at 45 knots and 45 feet agl to be both above etl and ground effect. I had a very difficult time not crashing with those numbers. After some practice I was able to pull the manuever off but it felt wrong. I felt much too slow and much too low. There was also an inordinate amount of sink when initiating the manuever. However when I raised the alittude and airspeed to 70 and 70 everthing felt right again and the manuever was easily completed. So I think its a DCS world sclae issue not a module issue.

 

Test 6: Autorotation. I have a few minor quibles with this one. I initiated the manuever like we did in flight school at 1000 feet agl and 70 knots. First time it didn't feel right. I left the collective bottomed out and NR (rotor RPM) was stable but didn't climb. What I did should have resulted in a rotor overspeed, also the glide path was much too steep. Second attempt after entering the auto I added a hair of up collective and I got the results I should have the first time. A build of NR and a glide path I would expect from a three bladed bird. This auto profile felt right. I did a few more to experiment though and came up with my biggest complaint. Polychop this is the one of two things I would ask you to change. Please consider it. When in an autorotation pitching the nose down reduces the flow of air through the rotor blades and dramaticly decreases NR. If your at a low speed when you enter an auto and have to build speed for a flare by lowering the nose it takes a dramatic reduction in collective to maintain NR. Conversely if your at a high speed when you enter the auto and have to slow down pitching up dramaticlt increases NR and you have to add collective to prevent a rotor over speed. Experimenting with these two situations yieled only very slight changes in RPM. It should be more exagerated. Also it hard to put specific numbers on this part but the amount of energy after flaring at the bottom of the auto felt much to forgiving. I think NR should probably deteriorate faster neccesitating a less forgiving timing of pulling collective to cushion landing.

 

Test 7: Exil's suggestion of a 20 degree nose down dive. I was unable to duplicate what he described. At 90 knots and 1100 feet msl I initiated a 20 degree nose down dive without changing power. At reaching 120 knots I pull aft cyclic and the aircraft immediately slows and climbs. Not at all what Exil describes. Perhaps you accelerated to a much higher speed? Settling with power and vortex ring state are two terms that get used interchangably however they are not always the same. All VRS is SWP but not all SWP is VRS. VRS is a specific aerodynamic condition while SWP is anytime you have power pulled in and are unable to arrest you decent. My guess is you had a high airspeed but not enough power to overcome you downward inertia.

 

Test 8: Is ground effect modeled? To test this I took off from CVN 70 carl vinson which was stationary. On the deck I was at 66 feet msl and pulled up to an IGE hover at 69 feet msl 3 feet above the deck, power required 65% tq. I then hovered at 4 knots off the edge of the deck and began a decent into the water. I performed the test again adding power to maintain an oge hover at 69 feet msl and power required was about 78% tq. Looks like its modeled and spot on.

 

Test 9: Heres one where we run into more problems... I attempted to and was unable to induce vortex ring state. First attempt 15 degrees c 29.92 pressure vertical decent from 2000 feet and I wasn't able to. Second try 10,000 feet 40 degrees c pressure 29.92 and unable. Third try temp maxed at 50 degrees c pressure bottomed out at 28.35 and weight increased to 4947 107% of max and unable. I had a few flickers of the mushy control response you would get in real life and increased rate of decent but they would quickly disapear. I think I just confused the FM. 20 knots Tailwind approach with same conditions and unable to induce VRS. Sometimes pilots will think they're in VRS when there not, how you 100% know you are is if you add power your rate of decent will increase. If you lower power it will decrease so your collective response inverts, this never happended. Too bad I was looking forward to trying the Vouchaird manuever in the sim (google it). So devs you can correct me if I'm wrong but it doesn't appear VRS is modeled at all. It's over modeled in the MI-8 and UH-1 but it would be a nice addition to this module. Please consider adding.

 

Test 10: Oh shit I forgot Exil's suggestion about the turn. No 2 minute turn indicator so 90 knots drop the 0 plus 5 equal 14 degrees of bank for a two minute turn. I used about 15 degrees on the AI and 90 knots. Without making any pitch or power adjustments I get about 600 foot per minute decent. Well I just traded some of my lift vector from the vertical to the horizontal to make me turn. Sound about right. Maybe a tad high on the decent. I do it again and a slight nose up and 5% more Tq and everthing behaves as it should.

 

Test 11: Various aerobatic manuevers. I do some that would be a very very bad idea but should technicly be possible and it is able to do them. Rolls and loops and such. And I do a few others it should not be able to do and the results are rapid unplanned disassembly. I was not able to do sustained inverted flight and an unloaded rotor disc with exagerated cyclic movements resulted in a tail chop just like it would IRL.

 

Verdict. Polychop... A job well done. Please consider making adjustments to the issues I point out in tests 6 and 9, however overwhelmingly it all feels right. I love this module a whole heartedly reccomend it. I am very much impressed by its fidelity. Further disclaimer I only had one beer during testing but several while writing, so I apolgize for grammar and spelling etc etc.

P.S. If any devs reading this are interested I think it would be pretty cool to help test modules for you after doing this. I'll even forgoe the beer if neccesary.

Edited September 23, 2017 by Jester986
Different results for 2.x

[FragBum]

@Jester986. Sure do thanks. :thumbup:

[Holbeach]

@ Jester986.

 

Very comprehensive test. Thank you.

 

Did you turn the gyro off?

 

..

[borchi_2b]

@Jester986:

 

Thanks a lot for your detailed description. I will take this and sve it to my document of flightmodel descriptions and issues, so I can have it reiewed by one of the french guys soon.

 

The more detailed descriptions I can have from real pilots the etter it is in total.

[Jester986]

@Holbeach, yes I did.

[sc_neo]

I'd be especially interested concerning the first test made concerning ''positive and negative dynamic and static stabilities'' when lifting off and hovering, specifically what the french pilots say about it. And its true, there is a small amount of positiv cyclic pitch needed to go to a straight hover, but no countering roll cyclic input. I don't know, up to now i maybe put it down to the SAS helping me out on this.

 

@Jester986 Could you explain a bit more in detail what you meant by ...''So ask yourself do you prefer a more realistic hover when you are trying to be in control, or when you let go of the controls?''?

 

Why would this (necessarily) mutually be exclusive? Ideally, we needed to counter as realistically as possible all static and dynamic forces induced by the two rotors and still be (able with sufficient practice and gear) to achieve a realistically stable hover. The PC simulation should help us out only in sofar as we do not sit in the real thing with all the sensory inputs that this provides that we must forgo when sitting infront of a monitor and in a non moving chair.

[FragBum]

 

...

 

Why would this (necessarily) mutually be exclusive? Ideally, we needed to counter as realistically as possible all static and dynamic forces induced by the two rotors and still be (able with sufficient practice and gear) to achieve a realistically stable hover. The PC simulation should help us out only in sofar as we do not sit in the real thing with all the sensory inputs that this provides that we must forgo when sitting infront of a monitor and in a non moving chair.

 

This is a good point, I'd be concerned if any changes favoured so called "joystick"control. Pickup is exactly that countering all static and dynamic forces, cyclic may not and probably should not be centered it should reflect the balance point of the aircraft. Am I missing something here?

[Jester986]

At the moment it takes me about the same level of effort to hover the Gazelle module as it does for me to hover a real helicopter, if I hop over to the Huey module it takes much more effort for me to hover. Hovering a real helicopter is extremely difficult when you first start. It took me ten hours of practice to be able to and for awhile after that if I had to hover for awhile I'd be soaked in sweat and physically exhausted when I was done. By the end of my private licenses however it was second nature.

 

The nature of simulations for some reason make it extremely difficult to model hovering correctly. When I went to flight safety to fly a bell 206 sim for flying to oil rigs, and that was an actual aircraft cockpit with a half dome screen around the front, they told me don't try to hover just immediately take off. Well I had to be a cocky pilot and try to hover it... I had a real cyclic and collective and was sitting in easily a million dollar sim made from a real aircraft and I couldn't come close to hovering it. I don't know if its a lack of physical feedback or a lag in control responses from the computer but sims cause pilot induced oscillations when hovering.

 

I suspect Polychop got around this by adding some extra stability to the aircraft. So I believe they are mutually exclusive. You could make the argument that I should relearn how to fly for the sim but I've been doing this for years and have never relearned how to hover. Its iniatially harder in the real thing but it gets easier in a way it doesn't in a sim.

 

So the way I see it we have a flight model that feels right with hands on the controls where we spend all of our time, and wrong when we take our hands off at a hoverwhich we shouldn't be doing anyway. Im happy with it now.

[FragBum]

With good input controls the Huey is a dream to hover so to the Mi-8 and Gazelle, user interface inputs will have some effect on user experience. But we shouldn't dumb it down.

 

Agree IRL it is a different feel, Robinson is easier and more lucid experience than DCS but this is VR I get that. ;)

[Hawkeye60]

Thanks a lot for your detailed description. I will take this and sve it to my document of flightmodel descriptions and issues, so I can have it reiewed by one of the french guys soon.

 

The more detailed descriptions I can have from real pilots the etter it is in total.

This is very welcome news!

Thank you borchi 2b!

 

The willingness to explore well presented information and experience bodes well for the Gazelle's future!

My Favorite Chopper by far!

 

(Send my best wishes to Blinky please.......):D

[Jester986]

And a little update, I feel like I was unjustly harsh on the Huey module. I just went back to it and if you try the same take off hover test it was much more stable then a real helicopter would be as well. It ends up transitioning into forward flight with a right bank. Hovering it was easier than I remembered though I think because last time I was playing with it I was just trying to do sling loading by hanging my head out the door and looking down at the cargo which took away a lot of my reference for the horizon. My apologies to Belsimtek.

[FragBum]

And a little update, I feel like I was unjustly harsh on the Huey module. I just went back to it and if you try the same take off hover test it was much more stable then a real helicopter would be as well. It ends up transitioning into forward flight with a right bank. Hovering it was easier than I remembered though I think because last time I was playing with it I was just trying to do sling loading by hanging my head out the door and looking down at the cargo which took away a lot of my reference for the horizon. My apologies to Belsimtek.

 

Mmm the reference for the horizon against the air-frame, this is what my instructor pushed on to me. <World==You> that's your juxtaposition. :thumbup:

[sc_neo]

In the end it comes down to whether the french pilots or a real life civilian Gazelle pilot can confirm whether this aircraft (like all other helos apparently) has a tendency to roll over to one side (left i think with the Gazelle?) because of gyroscopic precession when lifting off to a hover. I have just tried it again in 1.5 and there is no roll instability at all. If you happen to get the perfect minute backwards cyclic pitch and right yaw you are golden. The MI-8 really needs a good deal of right and backwards cyclic ptich in addtition to right yaw. Quite frankly, i am fairly new to the MI-8 although i love the Gazelle, having to counter that instability on all three axis is way more fun.

 

Actually, the point can be made that properly hovering the Gazelle is slightly more difficult. If you have a joystick with even a slight center detent/friction point (i hope you get what i mean) which you have to overcome each time when i need to make minute adjustments on pitch and roll, you are bound to oversteer and start a nice dance. If on the other hand, you are out of the joystick centre by default you don not have to fight that centre friction many joysticks have.

 

So my test in the Gazelle showed no influence of the gyro or the three SAS flips (bottom right of pilots console) on the aforementioned lift off characteristic. Maybe the Gazelle has some additional stabilization system built in which counters the roll tendency all by itself. Only real life Gazelle pilots can enlighten us on this one.

 

But if this is a liberty taken by the devs to make our life a bit easier, i would humbly ask to rather go the Belsimtek way and let us counter all forces by hand. Cutting out (if that was in fact so, again we don't know at this point) a complete axis of control to make it easier for us is called arcady, i reckon.

 

With that said, after you have put all real life ingredients in, you can of course implement it in a way that accounts for having not the same sensory inputs as if sitting in the real thing.

 

Anyway, interested which style of implementation you guys prefer.

[SnowTiger]

"rapid unplanned disassembly"

 

Hilarious > I love it.

I have had many of those !

[Jester986]

#1: Is it arcady to fudge something if it results in an overall more realistic experience? Not trying to be a jerk, its going to come down to personal opinion here. I'll be interested to see what the pilots with actual Gazelle experience and the Devs think.

 

#2: Gyroscopic precession is not whats going on here. And even where it is used its not actually correct, its an approximation that we use IRL flying helicopters to make it easier to understand something much more complicated. Ok, lets say where going to take off holding the cyclic comepletely neutral IRL, where going to run into alot of problems. First one that comes to mind is in the Astar (the closest to the gazelle I've flown) we would have a center of gravity slightly behind the rotor mast which means the helicopter is going to want to drift rearwards. So you would add a little forward cyclic to stop it. Second is translating tendency. This is the tendency of a helicopter to drift in the direction of tail rotor thrust. In an American helicopter the torque is trying to turn the nose to the right so the tail rotor develops thrust to the right to counter the torque. That right thrust cause the helicopter to drift to the right, Bell hangs a slight left cant in the rotor mast to counter this and Robinson simply requires the pilot to make a left cyclic input. That left ward vector from the rotor disc causes the left skid to hang a little than the right. Flying an MD 500 though it was much more pronounced. With a Eurocopter everthing is reversed because the blades turn the other way, resulting in a right rolling moment on take off and a right low skid. Note extreme left or right center of gravity can have an effect on this. So how does all this come together? Well an increase in power increases torque, which takes more tail rotor thrust, which increases translating tendency, which takes more cyclic to counter it, which robs some of the lift vector away from the vertical which requires more collective to maintain altitude and the cycle goes on and on... in the end though without any inputs you would end up catching a skid and rolling over most likely.

 

3# Ok jerk what is gyroscopic precession then? Well if you have a spinning gyroscope any input made to it is felt 90 degrees later in the plane of rotation. So if you think about our rotor systems... according to this if I want to go forward I push the cyclic forward and remove some pitch from the advancing blade (american helicopters) when its 90 degrees off my right and add some to the retreating blade when its 90 degrees off my left full downward deflection is reached dead ahead of me by the advancing blade and full upward reached behind me with the retreating blade, tilting the rotor disc forward, trading some lift vector for thrust and the helicopter moves forward. Except all that is wrong. Our rotor system isn't a true gyroscope because the majority of the weight isn't concentrated to the outside and blades can lead lag and flap. So what is going on then? Something called phase lag which is a combination of some gyroscopic affects and the actual time it takes for the blade to react to a change in lift. It varies from aircraft to aircraft. on the R44 I think it was about 72 degrees not 90. None of it is really relevant to pilots, more to the people who design and work on helicopters.

 

EDIT*** After playing around with it some more I got a much better onset of retreating blade stall. The initial problem I had with it was my mistake.

Edited September 23, 2017 by Jester986
I bad incorrect information.

[Holton181]

...gyroscopic precession...if you think about our rotor systems...all that is wrong...

 

Good thing to point out!

(but as you say, not really affecting the pilot)

To often confused, even among experienced pilots and even worse, in helicopter theory literature!

[FragBum]

Some help here please, I'm not seeing translating tendency in Gazelle, Huey or Mi-8 now I think I understand what it is and that certain design considerations help augment the flight characteristics.

 

At take off I apply collective slowly and balance toque and attempt to take the weight of the aircraft off the skids (wheels*) using the cyclic so that the thrust is lifting the aircraft vertically and indeed if I get it right I lift off vertically which includes bringing the aircraft up off a non flat surface. I also don't expect the aircraft to be perfectly level it will hang at it's balance point depending on it's centre of gravity.

 

My thoughts here is that I'm just automatically compensating for the effect??

 

 

* Fairly new to the Mi-8 and still coming to terms with the pogo stick landing gear and wheels. :cry: :D

[Jester986]

I just did a quick test and all three demonstrated translating tendency in 2.0 while only the Huey and the mi-8 did in 1.5. if you want to test it yourself take off and apply pedal to hold the nose straight and only make pitch adjustments with the cyclic do not make a left or right roll adjustment. You will notice that the gazelle and the mi-8 both drift left while the Huey drifts right in 2.0. And 1.5 only the Huey and the m-i8 demonstrate it. Also if you missed the edit I did on my big write up I tried doing the hands off take off in the gazelle in 2.0 and got Dynamic rollover as it should most of the time, but a few it did not.

[sc_neo]

This thread is very informative...i learned quite a bit since yesterday. I retested for translating tendency in the Gazelle and i see it in both 1.5 and 2.0. If i compensate for torque and forward pitch without any roll input, i find the Gazelle to be drifting to the left, but completely level, no tilt in the roll axis at all. I cannot produce a rollover in either Dcs version when lifting off. The Gazelle is very skiddy on tarmac or sand in both versions. Maybe that thing is just so light that or something that it lifts off before translating tendency can push it to a degree which would make it roll even slightly.

 

And i must say; i always wondered why helos move to one side, but i could not come up with a likely explation. I did not see anything about the main rotor which would explain it. I just thought the tail rotor thrust was completely ''used up'' to counter torque. When i look at pertinent documentaion its clearly explained...huh.

 

Still, i struggle with the idea of a one sided rolling tendency or tendency to tilt in one direction in the roll axis. If a helo has static or dynamic instability in the roll axis, shouldn't it go in both directions?

 

So i think we have established, that this in not happening in either game version in the Gazelle. The only time the Gazelle tends to rollover is after a hard lading and it either gets stuck with one skid (bug?) or something else in the rotor systems gets damaged. But on a clean start i cannot make it roll over.

 

@Jester986

 

Its weird that we have different findings on translating tendency in 1.5 and rollover behaviour in 2.0. No idea whats going on there. Anyway, the Gazelle has an autohover feature which obvioulsy means it can compensation for roll by itself. Whether this explains that perfect roll axis control when lifting off i don't know. In autohover mode, its not as perfect and the Gazelle swings slowly from side to like a pendulum.

 

So what force in a helicopter would explain that fairly strong tendency in the MI-8 to tilt over to the left, for instance? I do get that rolling over when lifting off can be due to the skid or wheel catching on the ground because of translating tendency. But why this tilt in the roll axis after lift off? If i look at the MI-8 i don't see anything that would explain a left sided centre of gravity. Is there some force (be it with us) i am oblivous to? Or might a helicopter be designed i a way to be slightly unstable in one direction, so that the pilot knows which direction tilt most of the time happens and thus can reliably counter instability in that one direction?

[Jester986]

I really, don't want to go off topic on the MI-8, but I'll reintegrate that you have to put cyclic in the opposite direction of tail rotor thrust to counter translating tendency and that cyclic input makes the skid/wheel on the side of the input hang lower and it will be proportional to total power. More power means more torque, which needs more tail rotor thrust, which needs more cyclic to counter translating tendency.

 

 

Edit: and I just realized I answered a question you didn't ask... You get a left roll because the translating tendency is pushing it in that direction, and if a tail rotor is mounted above the center of gravity it also introduces a rolling moment in the direction of thrust. The higher it is above the center of gravity the longer the arm and the larger the moment. Now do we have any helicopters with a tail rotor mounted rather high...

Edited September 25, 2017 by Jester986
I'm stupid

[exil]

Okay, tested the Gaz again tonight.

 

I think i found out what bothers me about the FM.

In the real world (asuming no wind, no gusts, standard day) when i initiate a turn the following basically happens:

Pushing my cyclic to the left/right for a certain amount results in a certain amount of bank angle. E.g. pushing cyclic to the left by 2cm results in a 15° bank angle if i hold the cyclic steady at this position (i know this is very simplified but basically what happens when you break the physics down).

 

Now i tried the same thing in the Gaz. The Gaz starts turning at a certain amount of cyclic input but doesn't stop rolling at a certain amount of bank angle. It just keeps on rolling. I activly have to counter it with applying cyclic to the opposite direction resulting in a back and forth (or left and right) with my cyclic. So it's not possible to maintain a steady 15° bank angle.

In order to make sure it's not me overcompensating my controls i put my saturation back to 10% in pitch and roll axis. I was pretty stunned by what happend.

 

I applied left cylclic and a long time nothing happend. Still level flight. Then i just applied a tiny bit more cyclic and held it there when the gazelle startet to turn. This beast almost flipped over!

 

I made a little video in order to demonstrate what i just encountered. Watch the red diamond in the red cross at the bottom left side closely.

 

 

For me, this really doesn't feel like a real one. It's like if there is an internal deadzone programmed in it. If you overcome the deadzone, you will turn no matter how little or big the rest of the input is. It's totally digital...0 or 1.