rotor collision... what went wrong

Bushmanni · May 7, 2010

Having dead zone increases your reaction times when responding to aircrafts movements. Basically it makes you prone to over correct and produce pilot induced oscillations. Having dead zone just makes flying very very hard compared to not having it. If you really need to have dead zone I would do it with the user curve. The basic dead zone adjustment basically eliminates possibility to use small and precise control inputs. The basic curve feature curves the curve way too much in the extremes and not enough near the center position. The only way to get a mostly linear (and hence predictable) curve with smoothed out center position is with user curve.

EtherealN · May 7, 2010

Any aircraft movement small enough to need you to move only within the dead zone is most likely a movement that you simply should ignore, caring about those tiny movements instead of just letting the aircraft settle itself is what'll get you into PIO. (That's something I struggled with when I was learning to fly.)

Deadzone and curves or completely linear - it's all a matter of getting used to those specific controls. But one question: why do you need precise control at the extreme ends? I don't think I've ever used the extreme inputs except when I've done it on purpose to just have fun - not even when performing split-S, barrelrolls and loopings in the shark...

Bushmanni · May 8, 2010

I haven't been able to fly as well with dead zone as without it, regardless of getting used to it. The proper curve will remedy the problem somewhat. You can learn to do things the hard way but I prefer the easy way. Making curve linear isn't about precision but predictability. With linear curve, I can feel with my muscle sense how much more input I'm giving. It's just easier and faster to learn to get along with it.

Conuk · May 8, 2010

So what kinda curve works for you?

care to share the numbers or a screenshot ?.

Am thinking I need to change something. After watching frazor and 0 curves etc spin the shark on a dime... My modded cougar can make the shark stand to attention or nose dive to ground with a gentle push or pull!

Edited May 8, 2010 by Conuk

dburne · May 8, 2010

Exactly.

You paid for the game- why let others tell you how you'll play it?

Agreed - and I am sure this type of thing turns several new folks off to this forum! BS is a very hard core simulation that is difficult enough to learn the ins and outs of, so I would think it has a fairly limited customer base - I have certainly seen some new folks get turned off to it when coming here for help.

Bucic · May 8, 2010

It's a matter of how crappy your controlls are. If they are perfect, then you use real setting - a linear characteristic. If not...

dburne · May 8, 2010

Well I have what I would consider as perfect a stick as I can get. TM Cougar modded with U2 NXT gimbal mod and hall sensors. Smooth as silk. Throttle and rudder pedals modded with hall sensors as well.

I have tested fairly extensively, and I get much smoother control input with a little curve ( less responsive around center ) and a little deadzone.

Bucic · May 8, 2010

Well I have what I would consider as perfect a stick as I can get. TM Cougar modded with U2 NXT gimbal mod and hall sensors. Smooth as silk. Throttle and rudder pedals modded with hall sensors as well.

I have tested fairly extensively, and I get much smoother control input with a little curve ( less responsive around center ) and a little deadzone.

The message that I try to punch through with is that a dead zone per se may be necessary but you can get it using User Curve and, opposite to the standard dead zone, get a perfectly smooth exit from the dead zone.

Conuk · May 8, 2010

Well I have what I would consider as perfect a stick as I can get. TM Cougar modded with U2 NXT gimbal mod and hall sensors. Smooth as silk. Throttle and rudder pedals modded with hall sensors as well.

I have tested fairly extensively, and I get much smoother control input with a little curve ( less responsive around center ) and a little deadzone.

I have the same cougar - will try this.

Frederf · May 8, 2010

The message that I try to punch through with is that a dead zone per se may be necessary but you can get it using User Curve and, opposite to the standard dead zone, get a perfectly smooth exit from the dead zone.

I sense an undue paranoia about the "kink" in the control graph between a standard deadzone and the linear livezone. It is unfair to qualify this kink as not "smooth" just because there is a hard corner in the graph's visuals. Be not deceived by the simple visual appearance. Apart from slop in the center of the joystick itself, linear control mapping is perfectly smooth.

The reason a hard corner is fine as the junction of the deadzone and the livezone coincides with cyclic deflections near the neutral position. Such deflections produce very small forces on the helicopter and are already "smoothed" by this fact of physics.

In fact an abrupt onset in linear control matches what the spring tension is doing in a typical slop+linear spring tension joystick. Matching the spring resistance to the control output is the most direct feeling control one can have.

Adding curvature beyond simple deadzone+linear is fine if one wants to flavor the experience that way but it is hardly necessary nor does it make the concept of a conventional deadzone "idiotic" at all.

Bucic · May 9, 2010

The reason a hard corner is fine as the junction of the deadzone and the livezone coincides with cyclic deflections near the neutral position.

Cachy but false. It could be an argument pro "dead zone to live zone transition is smooth" but we still have an inherent dead zone fault - a no-response zone obviously. The whole "smooth transition" discussion is far less significant.

As for the transition, let the numbers talk (the values are for the same curves from my previous figure, just an example):

For one the User Curve leaves some minimum response near the outer part of the dead zone instead of eliminating it.

Are those big differences? Of course not. But they are still evident in practice.

Also you can ask guys moding their X52 (gimbal far from perfect) hall sensor to get rid of its inherent dead zone why input is desirable even at the very center of joystick travel. Black Shark or aerial refueling in FC 2 can serve as a verification.

Another aspect - the only reason why simulators are that forgiving towards dead zone users is that commercial joystick handles are short. I'll leave you with that riddle. Hint: travel time.

To summarize - I'm glad that "dead zone is an unwanted thing in controller curve" has already been established. With that in mind I let other people decide if they would like to minimize it (Custom Curve) or not (regular dead zone). I already said virtually everything that was to be said in that subject. The real world characteristic and possible pilot induced oscillations (mentioned by Bushmanni) are important aspects of it. It's also related to the riddle I tossed in.

Edited May 9, 2010 by Bucic

Frederf · May 9, 2010

The reason a hard corner is fine as the junction of the deadzone and the livezone coincides with cyclic deflections near the neutral position.

Absolutely true. I'm talking about the Ka-50 cyclic, not the USB joystick. It is undeniable that the beginning of rise in the curve corresponds to the neutral zone of the cyclic. Before any output was registered the cyclic was at neutral! The control input having a "hard corner" is squeezed by being multiplied by another function, namely the function relating cyclic displacement with overall force applied to the airframe. No matter how sharp the change is between 0.0000 N to 0.0000N to 0.0000N to 0.0001N of force... it's still a weak onset of overall force. The ideal, perfect joystick would use exactly 1:1 mapping to the in-game cyclic.

You confuse linear with directly proportional. The red line in your graph is perfectly (piecemeal) linear, it's just not directly proportional.

The problem with the green line is that if I have a joystick with slop +/-10%, this "smooth" curve will make it so that my joystick is actually giving control input within the slop zone (baaaad!). This means I'm actually making a small control input left/right depending on which side my joystick lays in the slop zone which is beyond my control as a user. The joystick should give the exact same control output (post-curve) whether one relaxes the stick from the left or right deflection.

For one the User Curve leaves some minimum response near the outer part of the dead zone instead of eliminating it.

Undesirable!

The red line is what I want, a discontinuity in the behavior of the output because there is a corresponding discontinuity of the behavior of my input device. I want the dead zone to be absolutely dead and I want the live zone to be alive. The "serviceability" of my controller is 0% (untrustworthy) and jumps to 100% (trustworthy) across the slop boundary. Trying to smooth between "dead" and "alive" just means that authority is given to the part of the control range that shouldn't have authority and is withheld from the part that deserves 100% confidence.

If you imagine hooking a spring scale to the top of the joystick and pulling it through its range by this scale you will see that in a graph of tension (Y) vs. stick displacement (X) it will most resemble the red line. You want to have the closest correlation between physical tension and output.

All you're doing with the green line is reducing the slope of input-output mapping across the dead/spring boundary which can be accomplished in other ways without suffering the curve's creep into the deadzone. It's dead for a reason! The joystick is not trustworthy in that range of motion. Placing any output on motion in that zone is allowing output beyond the user's control!

Edited May 9, 2010 by Frederf

Bucic · May 10, 2010

Since I was not going to respond, I'll make it short.

The most important part - the tiny inputs within joystick gimbal crap zone (lots of play) are too valuable to be zeroed and are totally within user control. Although limited control. Even in Saitek Cyborg Evo (not the X model!) or AV8R. This is why further discussion between me and you is pointless. No offense.

The graph I presented is only a slice of the whole characteristic (part of one of the four quarters). I didn't confuse anything. I just consider line from I and III quarter as a complete characteristic. In that case a characteristic with a dead zone is not linear.

Bushmanni · May 10, 2010

Ok, I think I now understand Frederfs point as I can somewhat agree with him. If you have a stick with loose play at the middle position in centering merchanism but no loose play in sensors, the dead zone approach would give more accurate feel. The kind of curve Bucic presented would be good with spiky potentiometers. But I think in case presented by Frederf and not being able or willing to buy a new stick, I would still slightly smooth out the curve when coming out of the dead zone but I definitely would have a dead zone in my curve.

Bucic · May 12, 2010

The centering mechanism play zone is perfectly controllable in Saitek sticks. You can normally perform formation flying or refueling with a solid authority over the stick actual inputs to a game. The only problematic area is when you have to make adjustments just on the ege of the centering mechanism play zone - a bit in the zone and a bit outside the zone.

Talking is fine but I decided to remove the original centering mechanism and replace it with a primitive one outside the casing. Weird that I came up with the idea after ~20 hours of using X52 and not within the first hour :doh:

MaverickF22 · July 5, 2013

Now as you guys have understood that as the more airspeed (first of all), more collective level, the more right rudder (the effect of rotor intersection increases if you fly forward) and the more forward cyclic stick (again, when flying forward) will lead you to an epic blades collision on the right side (your 3 o clock)! If you fly backwards, the above elements (rudder and cyclic positions) will reverse (except for the collective which has to stay) which means that you will have to have more left rudder and stick pulled as your airspeed increases backwards (as long as you can increase it without turning forward:smartass:) in order to obtain a blades collision on the left side (it's all logical).

So in order to avoid touching your blades, simply avoid the above characteristics..., and most of it, the thing that almost everyone might still encounter (even the more experienced ones) while trying to fly as fast as they can in level flight (over 310-320km/h), eager to reach their destination quicker..., will have a feeling of certainty to puss the stick (cyclic) forward for going over a slope or a house, whatever..., then they will force the blades to collide, so avoid, or simply use as little forward stick as possible as your forward speed increases (pulling the stick will increase the distance between the upper and lower rotor discs, but you should now be careful on your G load which can brake your blades again) so it's a bit safer than pushing.

Have a KAMOV day!:thumbup:

MaverickF22 · July 5, 2013

There is one problem though which i can't comprehend!

Why are the rotor discs coming closer and intersect (on the right side as you fly forward) at higher altitudes (lower air density) than at lower altitudes (higher air density) for the same amount of true airspeed which is indicated on the HUD (the KA-50's HUD will normally show you the true/ground speed) or let's say indicated airspeed also, which if you try to maintain while gaining altitude, the blades will come closer to collide on the right side even quicker.

Normally, if you fly higher, where the air density, and in turn, the dynamic pressure which the blades and the whole airframe would feel, and also the indicated airspeed, should be lower, thus reducing the aerodynamic forces for the same amount of true airspeed (indicated on the HUD).

The weird thing is..., that even if the indicated airspeed drops as you maintain a constant true airspeed let's say, and gain altitude, the blades are coming closer and closer to each other on the right side and finally collide if you keep going higher in these conditions, and this seems completely opposite to logic..., because it simulates that the aerodynamic forces on the blades, depending on their azimuthal position (the highest is at 3 or 9 o clock, depending on the direction of rotation when flying forward or backwards) and indicated airspeed are more a function of true airspeed, which has a lower and lower top limit at higher altitudes, which is wrong...!

The aerodynamic forces on the blades should only be a function of each blade's angle of attack and dynamic pressure (which leads to the indicated airspeed), so for short..., the lower the dynamic pressure, then the lower this effect should be and vice-versa for higher pressures. So, in reality, the higher you are flying then the higher the possible true airspeed should be, for the same critical or maximum indicated airspeed where the blades might collide, indicated airspeed which should be the same at every height.

Now tell me what am i missing here! I'd like an answer from an ED member who knows aerodynamics and flight dynamics.

Thank you!

Have a good day,

Mav.

sobek · July 5, 2013

Just a guess but the angle of attack of the blades is higher at high altitudes due to the lower static pressure, which prolly means it reacts stronger to the additional dynamic pressure caused by the IAS, causing it to flap more.

BBQ · July 6, 2013

After many years of frustrating (though fun) flights in the Ka-50, and many different "styles" of flying and using the trim, my advice is to first and foremost understand how the trim system works, and then master basic flight maneuvers like turning, hovering, cruise, etc. You said you've pretty much covered those already -- but I would guess that you really haven't (of course I believe you when you say you have -- but two years from now you'll look back and laugh most likely, on your statement).

It's not the stick, nor a deadzone, nor a curve.

Read all you can in the "trim" and "autopilot" threads. Aside from luck -- I don't see how it would be possible for a new pilot to consistently perform advanced maneuvers like the split - s immelman, etc.

Bushmanni · July 6, 2013

The right side of lower rotor disk raises due to induced flow through the rotor disk, not due to the blade traveling the same direction as the helicopter is flying. When air gets thinner the blades need to work harder (higher AoA) to produce the same amount of lift and therefore the induced flow is increased. How the induced flow decreases blade tip separation in forward flight is a very complex phenomenon that needs to be explained with complex math or wind tunnel demonstration as it's all about air particles interacting with each other. There's a thread about this behind the link, sadly without any comments from ED.

http://forums.eagle.ru/showthread.php?t=99594

JG14_Smil · July 6, 2013

Don't fly so fast.

Actually, it IS the curves and controller setup that make the Ka-50 fly smoothly.

BBQ · July 6, 2013

Don't fly so fast.

Actually, it IS the curves and controller setup that make the Ka-50 fly smoothly.

Then how is it that I can fly like a hot knife through butter with zero adjustments to curve, deadzones, etc.?

:music_whistling:

Bucic · July 7, 2013

]The right side of lower rotor disk raises due to induced flow through the rotor disk' date=' not due to the blade traveling the same direction as the helicopter is flying.[/u'] When air gets thinner the blades need to work harder (higher AoA) to produce the same amount of lift and therefore the induced flow is increased. How the induced flow decreases blade tip separation in forward flight is a very complex phenomenon that needs to be explained with complex math or wind tunnel demonstration as it's all about air particles interacting with each other. There's a thread about this behind the link, sadly without any comments from ED.

http://forums.eagle.ru/showthread.php?t=99594

Really? This would be a new one for me.

1. There is no reason to disregard the upper vs lower rotor opposite dissymetry of lift (manual, 4-2).

2. I think you've mistakenly took the airflow to be like this (manual, diagram 5-2) in high-speed forward flight. It's like (manual, diagram 4-16), and I guess it's even flatter at 250+ kph. I'm not saying the induced flow is negligible here. Just saying that the dissymetry of lift is the dominant phenomena responsible for the blade collision.

I've checked the manual just now and there seem to be no section dealing with blades intersection. The chapter 5 AERODYNAMIC FEATURES OF COAXIAL

CONFIGURATION HELICOPTERS only mentions possibility of blade to tail boom collision possibility for normal helicopters (p. 5-2).

As for the blades colliding more eagerly at high altitudes. I can't think of any specific reason why this would happen, right now. Wait... Does it really happen? At higher altitudes blades collide at lower IAS?!

The aerodynamic forces on the blades should only be a function of each blade's angle of attack and dynamic pressure (which leads to the indicated airspeed), so for short..., the lower the dynamic pressure, then the lower this effect should be and vice-versa

I think you're too fixated on forces acting on the blades. Let me put the relations we discuss here this way:

the dangerous upper to lower blade tip distance <- opposition of upper rotor vs lower rotor dissymetry of lift <- dissymetry of lift phenomena

Sign In

rotor collision... what went wrong

Recommended Posts

Bushmanni

EtherealN

Bushmanni

Conuk

dburne

Bucic

dburne

Bucic

Conuk

Frederf

Bucic

Frederf

Bucic

Bushmanni

Bucic

MaverickF22

MaverickF22

sobek

BBQ

Bushmanni

JG14_Smil

BBQ

Bucic

Irregular programming

BBQ

Recently Browsing 0 members