FDE Help....

[John_Cillis 2,406]

I am trying to create an updated FDE for the recently released "Jupiter 2" (from "Lost in Space")... My accomplishments so far:~Can take off nearly vertically with full flaps~Can land vertically~Stays smooth and stable from 10KTS to roughly 2700KTAS (FS2002/FS2004's maximum, it seems)~Can climb to FL 990 in less than 60 secondsIn order to accomplish all of this, I really had to tone down the elevator effectiveness, as it was nearly impossible to control the pitch at high speed. Problem is, at lower speeds (<100KTS with flaps extended), there is almost no pitch and roll response even though the aircraft is well above stall.I've poured over the FDE (both .air and aircraft.cfg) trying to look for a parameter that lets me adjust the effectiveness "curve". I'd like to get control effectiveness more responsive at low speeds without making it overresponsive at high speeds. When I am done with this, I will send the FDE to the Jupiter 2's author for inclusion in his next update.... I have to tell you that I have the high speed piece so nicely done, that flying through the 3-D clouds at 9000 feet at 2600KTAS has to be seen to be believed.Any thoughts?Regards,John

[Guest Douglas K]

Hi John, Forget the aircraft.cfg file, that's for casual FD maniacs only.;) What you want is in the airfile. Look for tables 517 (elevator), 518 (ailerons), and 519 (rudder). They are concerned with Cm_de, Cl_da, and Cn_dr vs q, or to put it more simply the control effectiveness of the elevator, ailerons and rudder in response to dynamic pressure (airspeed). The tables read from left (low speed) to right (high speed), the X values are related to the airspeed or Mach number, and the Y values to the control effectiveness. They usually begin with X values of '0' and proceed to '1500', and Y values of 1.0 to 0.5 in most of the stock MS aircraft. Try increasing the Y value of the second XY pair from the left (it will probably have an X value of 180-300 and a Y value of 0.5-0.8 depending on the airfile type) from 1.0 to 2.0 and then test for results. To change the Y values:With AirED, use the numberpad and arrow keys. NUM lock on. Left and Right arrow keys move the highlight left and right. Numpad 4 and 6 also move the highlight left and right. Up and Down arrow keys increment and decrement the data by one. Numpad 7 and 1 increment and decrement the data by 10. Numpad 8 and 2 increment and decrement the data by 100. Numpad 9 and 3 increment and decrement the data by 1000. The X values can be moved left or right if necessary, just press X on the keyboard, type in what you want (you might want to try 70, 80 or 90 for control response under 100 KIAS) and hit enter to move the line to the desired position. If you want, the Y values can also be edited this way instead of using the numberpad.It can take a little trial and error (sometimes a lot), and flight-testing to achieve exactly what you want, but this will solve your problem. I use 518 quite often to lower aileron control effectiveness at low airspeeds and model the greatly reduced aileron control response you often find with some real aircraft when flying slow. Best RegardsDouglasEdit: If you find you have an airfile that doesn't contain these tables, then use AirED's copy to clipboard feature to grab them from another airfile and then use the add from clipboard feature to paste them into yours.

[John_Cillis 2,406]

Thanks--I'll give that a try. I suspected these tables may come into play--your explanation seems to document them pretty well. -John

[John_Cillis 2,406]

Just a follow-up.... Your suggestion did the trick, although I had to add these records from another .air file. I also found a record for trim as well since that also became too sensitive at high speeds.... It flies like a champ now.... Thanks again...-John

[Guest Douglas K]

That's great John, glad to hear you got it working. I'm always happy to help if I can. Best RegardsDouglas

[Guest Ron Freimuth]

"What you want is in the airfile. Look for tables 517 (elevator), 518 (ailerons), and 519 (rudder). They are concerned with Cm_de, Cl_da, and Cn_dr vs q, or to put it more simply the control effectiveness of the elevator, ailerons and rudder in response to dynamic pressure (airspeed). "The tables read from left (low speed) to right (high speed), the X values are related to the airspeed or Mach number, and the Y values to the control effectiveness. They usually begin with X values of '0' and proceed to '1500', and Y values of 1.0 to 0.5 in most of the stock MS aircraft. .." I'm happy to see people are using the technical 'abbreviations' I added to Aired.ini. Besides, they are short to write. ;) The official terms are in Greek letters with subscripts. But that's mainly done in textbooks; technical class notes, etc. generally use something close to what I decided on. ----- Note 'q', dynamic pressure, is measured in lb/ft^2. Long ago we thought those x axis were 'IAS'. I remember some values for q vs IAS. A C172 flies at about 110 kts IAS, the corresponding q is 40 lb/ft^2. q varies with EAS^2, but at lower speeds IAS^2 is pretty close. Say one wants to increase rudder and elevator effect at low speeds (since there is no prop blast effect on the tail in FS9 this may be desirable). Say to double the control moment at 30 kts IAS. Note 110/30 is about 3.7. This means q is 1/3.7^2 the 40 lb/ft^2 at 110 kts. About 110/15: 7.5. So, one would set the rudder and/or elevator moment vs q tables to 2 at 7.5. You can see how low that is. I did the above in my head, since one still would want to experiment, it's close enough for a start. Regardless, it's good to set those tables to 1.0 at q =0. Otherwise, the rudder and/or elevator will deflect double the set limits when checked before TO. This is because those tables change how far physical surfaces deflect. One might consider they model the limited strength of the pilot in moving the controls at higher airspeed values. While hydraulic operated control surfaces, such as the rudder, are often designed to run out of force to keep for moving the surface too much at higher speeds. In fact, 180 kts may start to limit how far the rudder of a jet transport will deflect. Considering q varies with IAS^2, 180 kts would be about 105 lb/ft^2. Note slower AC will see no effect from any 'q' related table that is flat to q = 80 or so. Since that corresponds to roughly 150 kts IAS. AFSD displays q very accurately. Ron

[Guest Douglas K]

I fear the damage to my vocabulary is permanent. I find myself using those definitions when thinking and making notes about flight dynamics for MSFS aircraft. And it is shorter and more convenient to jot down a note during flight-testing to increase Cd_df or CY_r than it is to write drag coefficient flaps, side force yaw rate etc.I've had mixed results with increasing rudder and elevator vs. q to compensate for the loss of prop wash effect in FS9. It always ends up as a compromise, and I've spent hours tailoring things to suit the takeoff phase of flight, only to find that the approach to landing or most especially the stall and spin characteristics were unacceptable. As far as I'm concerned, in the final analysis nothing is an acceptable substitute for propwash effect except...propwash effect.Here's to hoping that no further damage will be done to the flight model in the next version of FS. :-beerchug The tables make better sense to me now, many thanks for the explanation. I knew they didn't correspond precisely with airspeed, and much of the time consuming work in the past has been simply to create the desired effect at the correct speed. I'm going to put this new info to work; it should let me start out much closer to my goals than before.I had a hilarious result during my earliest experimenting with these tables. After setting 517 to 3.0 at 0, I was checking out a joystick problem from spot plane view and saw my elevators flipping well past the vertical in both directions. It took me more than a moment to figure out why.I decided against limiting the effectiveness at high speeds to model non-boosted controls, it just didn't feel right to me. You can still move the stick as before, only the response is diminished. It feels more like a stretched cable or a cable that has slipped a pulley than the solid resistance of high air loads.But I agree that modeling the rudder system on a jet transport seems a perfect use for 519 Cn_dr(q). With data from the real aircraft and a little work you could create a very realistic model of an aircraft that uses a blowdown system to reduce control surface deflection at high airspeeds.That seems a better use for control authority vs q than decreasing 518 to model for example the really poor roll rate of an A6M Zero at speeds in excess of 250 mph. Even a Sumo wrestler would be hard pressed to fully deflect the stick at those speeds, and in MSFS it seems easier and more realistic to simply limit how much you deflect the joystick above 250, rather than spending a lot of time working with Cl_da(q) to limit the aircraft's control response from the same controller inputs. This is just my personal preference; others may find it perfectly acceptable. The low end of the speed range is a really good use for table 518. As I mentioned before, you can model the sloppy feel of light aircraft ailerons in slow flight or just before touchdown. Thanks again for the info Ron, I'm off to play with it.Best RegardsDouglas

[Guest Ron Freimuth]

>I fear the damage to my vocabulary is permanent. I find>myself using those definitions when thinking and making notes>about flight dynamics for MSFS aircraft. And it is shorter and>more convenient to jot down a note during flight-testing to>increase Cd_df or CY_r than it is to write drag coefficient>flaps, side force yaw rate etc. You may be getting more familiar with them than I. However, a few, such as Cm_q, and Cl_p now just roll off my tongue.>I've had mixed results with increasing rudder and elevator vs.>q to compensate for the loss of prop wash effect in FS9. It>always ends up as a compromise, and I've spent hours tailoring>things to suit the takeoff phase of flight, only to find that>the approach to landing or most especially the stall and spin>characteristics were unacceptable. As far as I'm concerned, in>the final analysis nothing is an acceptable substitute for>propwash effect except...propwash effect. Prop effect on horiz stab was lost in FS2K2, but the effect on the rudder still worked. It looks like a realistic P-Factor is now in FS9; however the losses don't make up for it. ;( TAS, q, and Mach Number are what count aerodynamically. It turns out that FS only gets at IAS in the CAS-IAS calibration settings. It uses them to generate the IAS variable. Which is ONLY a variable for the ASI and pilot. The flight model 'works off' q, Mach and TAS, IAS is only for the pilot. We eventually figured out that the flight and powerplant models are in Engineering units, and consistent with real aerodynamics. Other than for bugs. Though, the XML 'types' provide several different scaling factors, some of which are rather odd. Further, it appears the old, legacy scaled integers may often not have a scaling factor that works, and one has to scale the integer manually. FS 98 and earlier used some metric values, as scaled integers. >The tables make better sense to me now, many thanks for the>explanation. I knew they didn't correspond precisely with>airspeed, and much of the time consuming work in the past has>been simply to create the desired effect at the correct speed. And, 'q' varies a lot more than 'IAS'. But, is mainly what the AC responds to.>I had a hilarious result during my earliest experimenting with>these tables. After setting 517 to 3.0 at 0, I was checking>out a joystick problem from spot plane view and saw my>elevators flipping well past the vertical in both directions.>It took me more than a moment to figure out why. Right, that's why I set those tables to 1.0 at q=0. I figure I'm not as likely to see excessive deflections when I'm busy in the cockpit.>I decided against limiting the effectiveness at high speeds to>model non-boosted controls, it just didn't feel right to me.>You can still move the stick as before, only the response is>diminished. It feels more like a stretched cable or a cable>that has slipped a pulley than the solid resistance of high>air loads. It's pretty important in jets, otherwise control are too sensitive. I generally set those tables to start dropping around q=150 or so. But, only down to about 0.8 to 0.7 at q=300 for the elevator and ailerons. While I now make the rudder drop very rapidly above q ~ 100. Note if the 'moment factor' is proportional to 1/q above 100 it would be set to 0.5 at 200 and 0.3 at 300. However, I think there was reason to make it drop even more rapidly at first. OTOH, one doesn't want the sensitivity to change too rapidly or the pilot will notice an abnormal effect when changing q (speed) in such a range. I know the Lear45 was way too senitive in cruise in FS 5, I had to use the autopilot to hold cruise altitude. There were (and still are) adjustments to adjust KB flight control vs speed, but few use the KB nowadays. I figure a lot of additions were added to the flight model after FS5; FS98 had most of what is now available. But, that wasn't of much value, since none of use know what they did back in the FS98 days. >But I agree that modeling the rudder system on a jet transport>seems a perfect use for 519 Cn_dr(q). With data from the real>aircraft and a little work you could create a very realistic>model of an aircraft that uses a blowdown system to reduce>control surface deflection at high airspeeds. I only started doing that a few months ago, after someone told me that the 707 rudder deflection is limited at higher speeds automatically by the size of the hydraulic cylinder. Later, I ran into 'control and gain profiles' for one of the jet fighters. The 'fly by wire' controller limits how much surfaces can deflect. The Yaw Damper still doesn't work, but much of the problem with setting realistically low Yaw Damping, Cn_r was that the rudder could yaw a jet way too much at higher speeds. With the deflection now limited to +/- 6 degrees or so at higher speeds (q), the defective Yaw Damper isn't really needed. >That seems a better use for control authority vs q than>decreasing 518 to model for example the really poor roll rate>of an A6M Zero at speeds in excess of 250 mph. Even a Sumo>wrestler would be hard pressed to fully deflect the stick at>those speeds, and in MSFS it seems easier and more realistic>to simply limit how much you deflect the joystick above 250,>rather than spending a lot of time working with Cl_da(q) to>limit the aircraft's control response from the same controller>inputs. This is just my personal preference; others may find>it perfectly acceptable. >>The low end of the speed range is a really good use for table>518. As I mentioned before, you can model the sloppy feel of>light aircraft ailerons in slow flight or just before>touchdown. I've increased the aileron control at approach speeds, since some jet transports try to maintain a constant roll rate regardless of speed. However, some beta testers didn't like my change. ;)>Thanks again for the info Ron, I'm off to play with it.>Douglas In part, it's all the details of the FS 'flight model' that keep people from getting the results they expect. Including details of how the CoL, CG, LG, etc. locations in aircraft.cfg. If the AIR file is reasonable, much of getting good ground and air handling is set in aircraft.cfg. And, I still have plenty of problems getting what I want. In part because I make a big effort to use information from the real AC, including LG and fuel tank positions. BTW, There is a new Aircraft.dll for Orbiter. While Orbiter hardly competes with FS, any addon's, including details of the new DLL will be completly public, and as time goes on, be developed for what the designer needs. Models can be done in G-Max, etc. The 'scenery' can be modified with an MS texture app. 'Mesh' has 'level of detail' up to '9'. So, I guess it is similar to FS 'textures'. All the variables in Orbiter are 'metric'. I'm still wresting with km/hr, m/s, Pa, etc. However, I expect panel gauges with units pilots understand will eventually be done. Ron