Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
You started up the engines, firewalled the throttle, let the RPMs stablize, released the brakes, and the aircraft pitched *up*??? That's clearly unphysical. Why ? The nose pitches down with power and brake application. So, releasing the brakes makes the nose pitch up. Not immediately, the aircraft pitches up after rolling for a few seconds. I think the main problem really is the rapid increase in airspeed, which is unnatural, and doesn't occur if both engines are used. Actually, it diverges for some reason, I just made another test and ended up with an aircraft at 35 deg, some roll to the right, and 1346kt! This is a few seconds after turning one engine on, running it at full speed, and releasing the brakes. No flaps, all controls centred. Andras === Major Andras e-mail: [EMAIL PROTECTED] www:http://andras.webhop.org/ === ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
Andy Ross writes: 2. Maintaining a straight heading is hard during the early part of the takeoff roll, but the text describes S-curves rather than violent spinning as the problem for inexperienced pilots. Is that with or without braking being applied? I can confirm that I execute lots of S curves during takeoff in the DC-3 when using the brakes method. It only spins violently when you try to correct yaw divergence with a flapping rudder. As far as I can tell, that's without braking. Braking during a takeoff roll would be so unusual for typical pilots that stories about flying the DC-3 aimed at a modern audience would be sure to mention it. One of the narratives specifically mentioned just tapping the rudder pedals rather than making large rudder inputs during the early part of the takeoff roll (i.e. at slower speeds) to avoid the s-curves. Hey, now that's really good information. This would *definitely* help with directional stability. You can lock the tailwheel by simply removing the castering=1 bit from the gear definition. This could be pretty easily made settable at runtime via a property. I tried that, and it's an improvement, but the tailwheel seems to slide sideways too easily. You can see it most clearly from external view, where applying only a light differential brake causes the tail to rotate sharply. Could there not be enough weight on the wheel? Note that I set castering=0 rather than removing the attribute completely. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
Andy Ross writes: OK, if anyone wants to try it before I get home, the following 5-line patch adds support for a settable castering attribute for gear objects. Apply it to the YASim directory, and then replace the tail wheel definition in dc3.xml with this: !-- Tail wheel; has castering selectable by a wheel lock -- gear x=-17.3 y=0 z=-1.5 compression=0.2 control-input axis=/controls/tailwheel-castering control=CASTERING/ /gear Then bind a key to toggle it, and we're set. Hopefully I haven't broken anything. I'll test it this evening. That sounds like an excellent start. One peculiar property of the DC-3 tailwheel I read about is that it can be locked only when aligned with the longitudinal axis of the plane. If you activate the lock when the tailwheel is turned, it will not engage until the wheel passes through the longitudinal axis, at which point it will snap on. That way, there's no risk of locking the wheel at an oblique angle, but you can engage the lock early while turning onto the runway. If that's too fancy, we could just have the lock instantly snap the wheel to position for now. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
David Megginson wrote: Note that I set castering=0 rather than removing the attribute completely. I saw it in a slow, taxiing turn at around 10kt or less, but I had done the modification myself before you posted yours. I'll try it with exactly your suggestion. Ah; this is my fault. You got faked out by the dumb YASim parser. It looks (well, looked, it uses the more robust control-input mechanism now) only for the *presence* of the castering attribute, not its value. So specifying castering=0 told YASim that the wheel *is* castering. Like I said, it's fixed now; don't yell at me. :) Attached is the DC-3 file I was using last night, which maps the castering bit to /controls/tailwheel-castering. I'm going to check the patch in right now, so give it whirl and see if it works for you. For now, perhaps locking the tailwheel could automatically snap it to 0 deg steering angle. We could even handle that in the input bindings, if there were a pseudo-steering property for the tailwheel. Sorry, I wasn't clear. That's exactly what happens right now. The way it works is that castering causes the gear to ignore the whole issue of steering direction and simply ignore all force along the ground plane. This is nice and simple, and generally has the right effect (even for the DC-3 tail wheel, excluding the falling into place feature). But it makes the falling-into-place feature harder to implement. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com Men go crazy in conflagrations. They only get better one by one. - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
I wrote: Attached is the DC-3 file I was using last night, which maps the castering bit to /controls/tailwheel-castering. I lied again. Now it's attached. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com Men go crazy in conflagrations. They only get better one by one. - Sting (misquoted) airplane mass=16865 approach speed=58 aoa=13 control-setting axis=/controls/throttle[0] value=0.4/ control-setting axis=/controls/throttle[1] value=0.4/ control-setting axis=/controls/mixture[0] value=1.0/ control-setting axis=/controls/mixture[1] value=1.0/ control-setting axis=/controls/propeller-pitch[0] value=0.5/ control-setting axis=/controls/propeller-pitch[1] value=0.5/ control-setting axis=/controls/flaps value=1.0/ control-setting axis=/controls/gear-down value=1/ /approach cruise speed=180 alt=23200 control-setting axis=/controls/throttle[0] value=1.0/ control-setting axis=/controls/throttle[1] value=1.0/ control-setting axis=/controls/mixture[0] value=1.0/ control-setting axis=/controls/mixture[1] value=1.0/ control-setting axis=/controls/propeller-pitch[0] value=1.0/ control-setting axis=/controls/propeller-pitch[1] value=1.0/ control-setting axis=/controls/boost[0] value=1.0/ control-setting axis=/controls/boost[1] value=1.0/ control-setting axis=/controls/flaps value=0.0/ control-setting axis=/controls/gear-down value=0/ /cruise cockpit x=-1.63 y=0.41 z=0.76/ fuselage ax=0 ay=0 az=0 bx=-19.75 by=0 bz=0 width=2 taper=0.5 midpoint=0.25/ wing x=-7.02 y=1.09 z=0.95 length=11.58 camber=0.1 chord=4.77 taper=.286 sweep=19 dihedral=4 stall aoa=14 width=3 peak=1.5/ flap0 start=0.04 end=0.38 lift=1.5 drag=2.0/ flap1 start=0.38 end=0.97 lift=1.2 drag=1.5/ control-input axis=/controls/flaps control=FLAP0/ control-input axis=/controls/aileron control=FLAP1 split=true/ control-input axis=/controls/aileron-trim control=FLAP1 split=true/ control-output control=FLAP0 prop=/surface-positions/flap-pos-norm/ control-output control=FLAP1 side=left prop=/surface-positions/left-aileron-pos-norm/ control-output control=FLAP1 side=right prop=/surface-positions/right-aileron-pos-norm/ control-speed control=FLAP0 transition-time=5/ /wing hstab x=-18.12 y=0.41 z=0.27 length=3 chord=3 taper=.545 sweep=15 stall aoa=14 width=3 peak=1.5/ flap0 start=0 end=1 lift=1.5 drag=2.0/ control-input axis=/controls/elevator control=FLAP0/ control-input axis=/controls/elevator-trim control=FLAP0/ control-output control=FLAP0 prop=/surface-positions/elevator-pos-norm/ /hstab vstab x=-18.12 y=0 z=0.95 length=2.72 chord=3.95 taper=.462 sweep=2 stall aoa=14 width=3 peak=1.5/ flap0 start=0 end=1 lift=1.5 drag=2.0/ control-input axis=/controls/rudder control=FLAP0 invert=true/ control-input axis=/controls/rudder-trim control=FLAP0 invert=true/ control-output control=FLAP0 prop=/surface-positions/rudder-pos-norm/ /vstab propeller x=-4.09 y=2.52 z=-0.95 mass=2000 moment=35 radius=1.4 eng-power=1200 eng-rpm=2700 turbo-mul=1.5 cruise-power=660 cruise-speed=195 cruise-rpm=2100 cruise-alt=2 min-rpm=1600 max-rpm=2700 actionpt x=-2.45 y=2.52 z=-0.95/ control-input axis=/controls/throttle[0] control=THROTTLE/ control-input axis=/controls/starter[0] control=STARTER/ control-input axis=/controls/magnetos[0] control=MAGNETOS/ control-input axis=/controls/mixture[0] control=MIXTURE/ control-input axis=/controls/propeller-pitch[0] control=ADVANCE/ control-input axis=/controls/boost[0] control=BOOST/ /propeller propeller x=-4.09 y=-2.52 z=-0.95 mass=2000 moment=-35 radius=1.4 eng-power=1200 eng-rpm=2700 turbo-mul=1.5 cruise-power=660 cruise-speed=195 cruise-rpm=2100 cruise-alt=2 min-rpm=1600 max-rpm=2700 actionpt x=-2.45 y=-2.52 z=-0.95/ control-input axis=/controls/throttle[1] control=THROTTLE/ control-input axis=/controls/starter[1] control=STARTER/ control-input axis=/controls/magnetos[1] control=MAGNETOS/ control-input axis=/controls/mixture[1] control=MIXTURE/ control-input axis=/controls/propeller-pitch[1] control=ADVANCE/ control-input axis=/controls/boost[1] control=BOOST/ /propeller !-- Tail wheel; has castering selectable by a wheel lock -- gear x=-17.3 y=0 z=-1.5 compression=0.2 control-input axis=/controls/tailwheel-castering control=CASTERING/ /gear !-- Main wheels; note mapping of rudder input to effect differential braking -- gear x=-6.08 y=2.52 z=-3.49 compression=1 retract-time=7 control-input axis=/controls/parking-brake control=BRAKE/ control-input axis=/controls/brakes[0] control=BRAKE/ control-input axis=/controls/rudder control=BRAKE square=1 src0=-1 src1=1 dst0=0.5 dst1=-0.5/ control-input
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
At 02:47 PM 5/22/2002 -0700, you wrote: David Megginson wrote: 1. According to the author, at least, differential braking is bad form while taxiing the DC-3; you should use differential power instead except for very tight turns. I'll buy that. But working dual throttles during the takeoff and landing rolls can't possibly be a good idea, right? In that regime, you're still stuck with rudder and braking only. During the landing roll (with no significant prop wash), you're stuck with braking only. 2. Maintaining a straight heading is hard during the early part of the takeoff roll, but the text describes S-curves rather than violent spinning as the problem for inexperienced pilots. Is that with or without braking being applied? I can confirm that I execute lots of S curves during takeoff in the DC-3 when using the brakes method. It only spins violently when you try to correct yaw divergence with a flapping rudder. Just to clarify what I said earlier: the reason that it looks like a rudder problem is that turning the plane a little bit with the rudder is possible. But once it is pointed little bit away from the velocity vector, it begins turning *farther* away very rapidly. If you don't correct this immediately, the aircraft will rapidly be so far out of whack that the rudder is incapable of correcting the yaw. Thus, what started out as a tiny rudder input diverges into a ground loop. But it's caused by a *lack* of rudder authority to correct the problem, not by too much authority causing it. Does that make more sense? [snip] Differential braking should be kept to a minimum in any airplane, for two reasons: (1) An airplane is a really lousy automobile. It has about as little undercarriage as it can get away with (one has only to look at pictures of an airplane and a truck scaled to the same size to realize this), and every brake application is hard on its pitiful little brakes. (2) Differential braking tends to scrub rubber off some very expensive tires. So differential power becomes the steering method of choice in airplanes that have it available. Light taildraggers generally have steerable tailwheels, and being single-engined, they always have some prop blast over the tail; consequently they're not very hard to steer in the takeoff roll. Larger taildraggers don't have steerable tailwheels because the steering forces would require powered controls which were not in use when they were designed. In the Gooney Bird one must line up on the runway, lock the tailwheel, and hold the wheel firmly back until there is full tail surface control. Prior to that point, you aren't really steering a heading: you're just holding yaw rate to a minimum. The airplane will turn somewhat in a crosswind; this can be dealt with to some extent by judiciously positioning and aiming the airplane before starting the roll. The divergence you mention is present in a real taildragger; it's just a basic instability in the yaw axis resulting from most of the weight being supported in front of the cg. When the fuselage is misaligned with the direction of motion, the side force on the wheels is destabilizing. I don't know exactly how the tailwheel lock is implemented in the DC-3; in the AT-6, the last couple of inches of aft stick travel center and lock the wheel. It's an ideal arrangement, because if you don't have the stick back the tailwheel won't do you any good anyway. rj ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
[Flightgear-devel] DC-3 takeoff roll: partial solution
Here's a lot of information on taxiing the DC-3: http://www.douglasdc3.com/dc3taxi/dc3taxi.htm The most important point is that the DC-3 tailwheel must be locked for takeoff and landing (i.e. it doesn't caster freely). Also of note: 1. According to the author, at least, differential braking is bad form while taxiing the DC-3; you should use differential power instead except for very tight turns. 2. Maintaining a straight heading is hard during the early part of the takeoff roll, but the text describes S-curves rather than violent spinning as the problem for inexperienced pilots. Locking the tailwheel should help a lot, but we'll also have to make sure that the tailwheel has the right amount of authority. By the time the tailwheel starts to lift, I'd expect that the rudder should be becoming more effective. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
David Megginson wrote: 1. According to the author, at least, differential braking is bad form while taxiing the DC-3; you should use differential power instead except for very tight turns. I'll buy that. But working dual throttles during the takeoff and landing rolls can't possibly be a good idea, right? In that regime, you're still stuck with rudder and braking only. During the landing roll (with no significant prop wash), you're stuck with braking only. 2. Maintaining a straight heading is hard during the early part of the takeoff roll, but the text describes S-curves rather than violent spinning as the problem for inexperienced pilots. Is that with or without braking being applied? I can confirm that I execute lots of S curves during takeoff in the DC-3 when using the brakes method. It only spins violently when you try to correct yaw divergence with a flapping rudder. Just to clarify what I said earlier: the reason that it looks like a rudder problem is that turning the plane a little bit with the rudder is possible. But once it is pointed little bit away from the velocity vector, it begins turning *farther* away very rapidly. If you don't correct this immediately, the aircraft will rapidly be so far out of whack that the rudder is incapable of correcting the yaw. Thus, what started out as a tiny rudder input diverges into a ground loop. But it's caused by a *lack* of rudder authority to correct the problem, not by too much authority causing it. Does that make more sense? Also, recognize that implementing prop wash would have the effect of increasing rudder authority during the takeoff (but not landing) roll, which will also help. Locking the tailwheel should help a lot, but we'll also have to make sure that the tailwheel has the right amount of authority. Hey, now that's really good information. This would *definitely* help with directional stability. You can lock the tailwheel by simply removing the castering=1 bit from the gear definition. This could be pretty easily made settable at runtime via a property. You don't have to worry about the skidding authority of the wheel -- skidding friction (to first order, anyway, for tires that aren't melting) is the same for wheels of all shapes and sizes. I really should read through this site more carefully. It's got lots of good stuff. The fantastic quote in question is: CAUTION: THE TAIL WHEEL LOCK MUST BE LOCKED DURING TAKE OFF AND LANDING. Sounds like good advice to me. I'm not at home right now; can someone remove the castering setting from the dc3.xml file and try it? If this is the solution, then I'll add a property-based control for castering tonight. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com Men go crazy in conflagrations. They only get better one by one. - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
Andy, CAUTION: THE TAIL WHEEL LOCK MUST BE LOCKED DURING TAKE OFF AND LANDING. Sounds like good advice to me. I'm not at home right now; can someone remove the castering setting from the dc3.xml file and try it? If this is the solution, then I'll add a property-based control for castering tonight. Works, but isn't perfect. At least I can take off now, but if I try to use only one engine, the aircraft soon pitches up and crashes with the front wheels still on the ground, the tail stuck on the tarmac. Two engines work fine, though. Andras === Major Andras e-mail: [EMAIL PROTECTED] www:http://andras.webhop.org/ === ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
Andy, I just made two recordings of flights with the DC3, but can't play them back because fgfs segfaults. I can put them on the web if that helps (maybe even to debug the segfault...). Andras === Major Andras e-mail: [EMAIL PROTECTED] www:http://andras.webhop.org/ === ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
but if I try to use only one engine, the aircraft soon pitches up and crashes with the front wheels still on the ground, the tail stuck on the tarmac. Two engines work fine, though. But here you've lost me. Normally, the aircraft state with all three wheels on the ground is not called a crash. :) Sorry, sorry, that should have read tail stuck IN the ground. Attached screenshot taken within 3sec after releasing brakes, after this, the plane pitches up even more, and fgfs hangs, moaning about terrain intersections. Maybe it's the two fronts wheels taking off rather than the tailwheel being buried, but in any case airspeed builds up much too quickly. Sorry for the confusion. Andras === Major Andras e-mail: [EMAIL PROTECTED] www:http://andras.webhop.org/ === attachment: dc3-1engine.jpg
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
Major A wrote: Sorry, sorry, that should have read tail stuck IN the ground. Attached screenshot taken within 3sec after releasing brakes, after this, the plane pitches up even more, and fgfs hangs, moaning about terrain intersections. Maybe it's the two fronts wheels taking off rather than the tailwheel being buried, but in any case airspeed builds up much too quickly. One note is that the fact that the wheels are drawn inside the ground is probably a red herring. David's 3D model and the YASim geometry description of the DC-3 aren't in perfect agreement, so you can sometimes see this artifact. We need to clean this up at some point, but that's clearly not the bug you're seeing. Just to be clear: You started up the engines, firewalled the throttle, let the RPMs stablize, released the brakes, and the aircraft pitched *up*??? That's clearly unphysical. Clearly the only right thing here would be for the aircraft to accelerate slowly (the DC-3 gets about 0.3G of acceleration at the start of the run), and maybe ground loop. Under no circumstance should it be pitching until there is enough airflow over the horizontal stabilizer to lift the tail. This happens somewhere around 40kts or so, I think. Unfortunately, I've never seen anything like this behavior before. I'll try it this evening, but I'm pretty sure I would have noticed it before. Can anyone else reproduce this? Is there anything weird about your platform (other than the ugly window manager, that is)? The hanging of flight gear is just YASim detecting the crash and ceasing simulation. You should be able to recover by selecting File-Reset from the menu. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com Men go crazy in conflagrations. They only get better one by one. - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
You started up the engines, firewalled the throttle, let the RPMs stablize, released the brakes, and the aircraft pitched *up*??? That's clearly unphysical. Why ? The nose pitches down with power and brake application. So, releasing the brakes makes the nose pitch up. ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] DC-3 takeoff roll: partial solution
but if I try to use only one engine, the aircraft soon pitches up and crashes with the front wheels still on the ground, the tail stuck on the tarmac. Two engines work fine, though. But here you've lost me. Normally, the aircraft state with all three wheels on the ground is not called a crash. :) Sorry, sorry, that should have read tail stuck IN the ground. Attached screenshot taken within 3sec after releasing brakes. Sorry for the confusion. Andras === Major Andras e-mail: [EMAIL PROTECTED] www:http://andras.webhop.org/ === attachment: dc3-1engine.jpg
