Can a realistic turn & slip indicator be programmed?

[Guest Tom Goodrick]

I would just add that what the ball sees is total local acceleration. Consider it like coffee in a round-bottom cup held with the mouth parallel to the floor of the aircraft. When sitting still on the ground, the total acceleration is 1 g pointing straight down at the ground. But when flying, the total acceleration is usually near 1 g but varies a little. When doing a proper turn, the total acceleration should point down perpendicular to the floor of the aircraft and the ball should show that by staying in the center. In training we used to have to do "720's" or a steady turn twice around. During all that time the ball should stay centered. The coffe in a cup would stay centered in the cup even though the cup tilted with the aircraft. You would feel a force pushing you straight into the seat with no need to tilt your head. In fact, if you tilt your head in that maneuver, you get more messed up because the forces tell your body that the head should be straight but the eyeball says something is crooked.I looked in the parmeters that can be read by gauges hoping to finnd accel X, accel Y and accel Z. From that I could make a 'perfect' ball gauge. But all I found was TURN_COORDINATOR_BALL and who knows what drives that stupid thing.When I do a high-bank turn in a Learjet, there is no adverse yaw and the doggone ball should stay centered all the time. But you can see the yaw rate stay low even with a steady high bank until you kicj the rudder into the turn. Something in FS is terribly wrong. I don't think it is the gauge. I think it is the way the rudder works mathematically. It is screwed up.

[Guest Ron Freimuth]

>I would just add that what the ball sees is total local>acceleration. ... Which is different from 'slip'>I looked in the parmeters that can be read by gauges hoping to>finnd accel X, accel Y and accel Z. From that I could make a>'perfect' ball gauge. But all I found was>TURN_COORDINATOR_BALL and who knows what drives that stupid>thing. But, there are Velocity X, Velocity Y, and Velocity Z parameters. I know they can be differentiated to give reasonable accelerations: ax, ay, az. Ron

[Guest Tom Goodrick]

There is a better method. Just read and display the sideslip angle, beta. This is available in the parameter INCIDENCE BETA and it reads in degrees. I fashioned a quick and dirty little digital xml gauge that displays the angle to the nearest 10th of a degree. Then I used it to make adjustments in the Cessna Skylane RG I imported to FS9 from FS02.The rudder had the power to maintain a slip of 14.5 degrees left and about 13 degrees right (Directions refer to the yaw, not the slip.) There was no adverse yaw. I then went through the stability derivatives for roll and yaw. I came up with a combination that is more of a charicature of the real plane than a true representation. But it shows adverse yaw and requires that you step a little on the rudder pedal to counter it in order to make a smooth turn. It takes some practice to make nice turns but you can roll into a turn, hold the turn with zero sideslip and then roll out using a combination of aileron and rudder in each part of the maneuver. A very small amount of rudder is needed to hold the turn in a 30 degrre bank. I am currently playing with it. When I finish I will post it somewhere for downloading. It proves that we were all wrong in blaming FS for screwing up this rudder and ball business. We are also to blame - those of us who should know better. I never looked at the stability derivatives because I knew they could be a real can of worms. I hoped the planes would perform well enough that such fine tuning would not be needed. But that was a bad attitude. Now I am suspicious of everything in this sim - even the resolution of the rotation equations for a rigid body which is largely germain to this rudder-ball problem. Do they take into account properly the relative values for each of the moments of inertia in determining the roll-yaw coupling? The problem is that, if they do, we can screw it up by messing with some of these derivatives. Or, put another way, messing with these derivatives makes it possible to invent whole new laws of physics with no relation to reality what so ever. Of course, many people have been doing that for years by not using accurate moments of inertia.For that reason, we must tread lightly. I will be writing more about this with some examples.

[Guest Ron Freimuth]

>There is a better method. Just read and display the sideslip>angle, beta. This is available in the parameter INCIDENCE BETA>and it reads in degrees. I fashioned a quick and dirty little>digital xml gauge that displays the angle to the nearest 10th>of a degree. Then I used it to make adjustments in the Cessna>Skylane RG I imported to FS9 from FS02. I display pitch, roll, yaw, alpha, and their deriviatives in my XML Jet Test gauge. I'm confused about Yaw; is it also Beta? that what I assumed. However, I'm not sure Beta = zero means the G vector is down (ball centered). It seems one can be in a slip, then bank the AC to center the ball. I have noted 'autorudder' only locks the ball and disabled the rudder. Beta is the same with it on or off. Another MS FAKE! >The rudder had the power to maintain a slip of 14.5 degrees>left and about 13 degrees right (Directions refer to the yaw,>not the slip.) There was no adverse yaw. I then went through>the stability derivatives for roll and yaw. I came up with a>combination that is more of a charicature of the real plane>than a true representation. But it shows adverse yaw and>requires that you step a little on the rudder pedal to counter>it in order to make a smooth turn. I use published SD's when possible. They appear to work well for SEL's. However, I generally adjust the Control Derivatives downward to keep control sensitivity from being too high. Real AC give FFB and that is a good cue for control moments. I set Cm_de so I can get about +3.5 G and -2.5 G. Cl_da to get a reasonable roll rate (at an appropriate TAS), and have taken to setting Cn_dr so I can get about +/- 15 deg beta. The C172SP 'fix' I did is popular, it has SD's hand adjusted to be near the real ones calculanted by Smetana. I wanted to do a quick fix and start from scratch, rather than copy what I already had in a C172N. The C172N is mostly set to Smetana's SD's. In fact, I have his DOS programs which calculate them -- however haven't had the ambition to put in all the details to reproduce his values. Besides reducing the Control Deriviatives we have found the Vertical Wing Offset affects Wing Leveling. Seems Cn_beta should include any wing offset effects but I found I had to reduce Cn_beta from the published value when the wing was set above the CG to avoid too much wing leveling. People have still complained that my smaller SEL's don't hold a bank as a real AC will. Howver, if I reduce Dihederal Effect they isn't stable enough to hold a good heading. Funny, since larger AC don't have this problem, even though the SD's may be similar.>It takes some practice to>make nice turns but you can roll into a turn, hold the turn>with zero sideslip and then roll out using a combination of>aileron and rudder in each part of the maneuver. A very small>amount of rudder is needed to hold the turn in a 30 degrre>bank. I am currently playing with it. When I finish I will>post it somewhere for downloading. It's been over 30 years since I flew a PA-28; howver roll-yaw dynamics in my SEL's seem nominal. Drag in a slip appears to be too low, but there is some. AFSD will report 'drag due to sideforces' when Herve' and I get this figured out better. >It proves that we were all wrong in blaming FS for screwing up>this rudder and ball business. We are also to blame - those of>us who should know better. I never looked at the stability>derivatives because I knew they could be a real can of worms. Most FS AC problems are due to poor parameters, not the basic flight model code. Beyond that, until one really gets into the details he may well be confused. I've concetrated on one thing at a time in my FD's and only more recently have gotten more into the Yaw-Roll Couplings. However, most of the effects are not too hard to understand if one thinks about them. I found 'Cn_p Yaw Moment - Roll Rate' was exactly what it says. When one is rolling into or out of a bank the ball moves to the side. It takes rudder to neutralize this. 'Cl_r Roll Moment - Yaw Rate' is impossible to eliminate (but ailerons compensate for it). I only recently realized this is due to the fact one wing is moving faster than the other when the AC is Yawing. Looking from the rear, a yaw to the left would create a CCW roll. When the roll rate is zero the ball should tend to return to center. There is a smaller 'Cl_dr Roll Moment - Rudder' which may require a limited amount of aileron in a turn. While 'Cn_da Yaw Moment - Aileron' (Adverse Yaw) is typically small and may even be zero. 'Cn_beta Yaw Moment - Sideslip' (Weathervane Stability) is what keeps an AC moving more or less 'straight'. Similar to how a weathervane aligns with the wind. Finally, 'Cl_beta Roll Moment - Sideslip' is the Dihederal Effect. One has to be careful with the signs of these parameters. FS convention is often reversed compared to some other conventions. Generally, its best to take a known good AIR file and be sure he sets the same signs it has. I think the MS AC at least get the signs right. ;) >I hoped the planes would perform well enough that such fine>tuning would not be needed. But that was a bad attitude. Now I>am suspicious of everything in this sim - even the resolution>of the rotation equations for a rigid body which is largely>germain to this rudder-ball problem. Do they take into account>properly the relative values for each of the moments of>inertia in determining the roll-yaw coupling? Everything can be tested. Perhaps some people here will be encouraged. The best tests are quantitiive. One might set related parameters to 0 to simply things (avoid other effects). I have confidence in the basic Roll dynamics. Cl_p, Cl_da, and Roll MoI appear to combine to generated appropriate roll dynamics. I have some formulas in Aired that give Helix Angle, Roll Rate, and Roll Time Constant. I tested a P-80A with heavy wing tanks at FL 350 and the Roll TC appeared to be consistent with the formula. Near one second. Something one doesn't even see with low Roll MoI and low altitudes. A quantitive test of Cl_da could consist of setting unbalanced wing tanks, calculating the extra_weight * distance and checking to see if aileron deflection was correct in "Cl_da * deflection (rad) * S*q*b = Unbalanced Roll Moment". One might check with the excess fuel set for the other wing also, since engine torque should have some effect. Just the same, roll rates are correct for specifice Roll Damping and TAS, so Cl_da (and its 2048 scale factor) are not in question. perhaps Cn_da, Yaw Moment - Ailerons (Adverse Yaw) would be worth checking. One need only check to see how much rudder it takes to neutralize a specific Cn_da (make it higher than normal). Both are normalized to S*q*b so that cancels out. If Cn_dr = |Cn_da| then a given aileron deflection should requre the same magnitude of rudder deflection to cancel the yaw. I don't think there are other effects to confuse such a test. Assuming sideslip is returned to zero. I expect Herve' will be adding Moments and Side Forces to AFSD. He will start on displaying TBL 473, Pitch Moment - alpha. More will be added as time and feasability allow. The idea being that if one can see how large certain components are then he has a better idea what is going on and if it reasonable. In the end, all forces and moments must sum to zero. That requires accelerations to be accounted for, but AFSD already does that to some degree as far as CD and CL go. ron

[Guest Ron Freimuth]

Douglas, I was going to reply to your message, but think I covered a lot of points in the reply I just wrote to Tom.Ron

[Guest]

> However, I'm not sure Beta = zero means the G vector is down>(ball centered). It seems one can be in a slip, then bank the>AC to center the ball. Beta=0 does not necessarily mean the ball will be centered. Because the ball measures specific acceleration, an aircraft could be in a bank with sideslip and accelerating laterally and still have a centered ball (at least momentarily). What is needed to drive the ball is aircraft acceleration in body axes plus gravitational acceleration transformed to body axes.> I have noted 'autorudder' only locks the ball and disabled>the rudder. Beta is the same with it on or off. Another MS>FAKE!Yikes! That's sad. Think of all the folks out there doing stall/spins turning final who think they're flying coordinated.

[Guest Karl R Pettersen]

HiI've tried to "fix" the default DC3 now using the techniques mentioned in this thread, but no success. Does anyone have an .air-file that takes care of these problems? Especially the wrong roll action. I've tried negative value on this, but nothing happens, it still rolls towards the wrong direction.I'm using Trev Morson's aircraft.cfg which fixes other (delay, power, etc) parts of the dynamics, but don't think this is relevant to rudder or engine out scenarios.

[Guest Ron Freimuth]

>> However, I'm not sure Beta = zero means the G vector is down>>(ball centered). It seems one can be in a slip, then bank>the AC to center the ball. >>Beta=0 does not necessarily mean the ball will be centered.>Because the ball measures specific acceleration, an aircraft>could be in a bank with sideslip and accelerating laterally>and still have a centered ball (at least momentarily). What is>needed to drive the ball is aircraft acceleration in body axes>plus gravitational acceleration transformed to body axes. The 'ball' variable is at least partially correct, in fact it may be fixed in FS9. Otherwise, I can add a bit of 'sideslip' to the 'ball' variable to fake a better response. >> I have noted 'autorudder' only locks the ball and disabled>>Yikes! That's sad. Think of all the folks out there doing>stall/spins turning final who think they're flying>coordinated. However, the 'yaw_damper_gain' in aircraft.cfg does appear to let one set a more realistic yaw damper. A YD will help autocoordinate if it works correctly. I only could see two different conditions with that yaw_damper_gain in FS9. 0-0.20 and 0.20 to 1.0. I'm not sure what side the 0.20 belongs to. Regarless setting 0.20 appears to allow several degrees of sideslip angle with the YD on. Further, the YD does appear to move the rudder slightly to counter Yaw. As it should. Ron

[Guest Douglas K]

>Does anyone have an .air-file that takes care of these problems?:)If you want these files, then post a reply here or send me a Avsim P.M., or E-Mail me and I will zip them up and send them along to you.