Pitch axis too sensitive

[tonghuix]

Very little yoke movement (I believe only 5~10 millimeter) at pitch axis would cause head up and down very rapidly and abruptly, and it also cause VSI change huge amount, from ±500 to ±1000, does it real?

I also test Cessna 337, CAP10B and DR400, these aircraft use jsbsim as well, but doesn't experience this problem.

And c182s has the same problem too, but not so obvious and making me dizziness.

To re-produce, you have to fly longer and without AP all the time, such us fly from PHOG to PHNL, and using a accuracy joystick such T16000M.

[okroth]

Where is your CG? What you describe sounds like pitch instability that occurs if the center of gravity is moved backwards. Usually the CG needs to be near of 25% of wing chord. If you have high load in the baggage compartment, it could create such a behaviour.

[tonghuix]

@okroth thank you for providing a explanation!

However, I checked the fuel and payload dialog, my CG is 40.95 inch shown in the dialog. And I checked POH, the CG is in Normal Category.

Here is my weight and balance detail:

Pilot: 180 lb Fuel: 40 Gallons / 240.8 lb Gross Weight: 1887 lb

[okroth]

40.95 is only 0.45 inch behind the 25% chord, and pretty OK. Did you open the property tree viewer and watch the elevator inputs? In the JBSIM flight model, the elevator is "filtered" using a kinematic component, which lets the yoke travel with minimum 0.15s from full pull to full push. That could introduce a pilot induced oscillation, as it takes some time after control input to create an effect. During this time, the lack of effect increases the input. You could check by changing line 692 of c172p.xml from fcs/elevator-cmd-norm-filtered to fcs/elevator-cmd-norm

[tonghuix]

@okroth Thank you again!

After change the c172p.xml file, the pitch not such sensitive, but when console property tree with elevator input from -0.01 to +0.01, the head still up and down still, but not quickly and abruptly.

And I also notice that when stay at ground (speed zero) with no air flow to elevator, the plane also head up and down but only response with full yoke, I found this when running up at checking flight control movement when before takeoff checklist.

[okroth]

Was the engine running? The C172P flight model has prop wash implemented that effects pitch moment due to elevator deflection.

[tonghuix]

@okroth oh yes, engine running at ground. But I wonder the prop wash shouldn't effect the plane such obviouse.

[okroth]

The flight model uses the /propulsion/engine/prop-induced-velocity_fps property as airspeed for the elevator and rudder, not the airplane's speed. This may be over-estimating the actual value at the tail as we have a cabin in between that slows down the airstream.

[tonghuix]

I do believe the pitch still too sensitive, does anyone have experience that fly a real Cessna 172?

[okroth]

In reality the plane is flown with force not travel feedback. When cruising with 100kts, an elevator displacement of 8° up is sufficient to pitch the plane by 6° nose up. This results in a triple as high CL giving 3g vertical acceleration. The 8° are achieved by moving the yoke only by 28mm. The resulting force from the elevator's hinge moment, however, is in the 400N range (about 40kg). Usually one uses a few 10N (a few kg) to manoeuvre the plane, resulting in small movements of the yoke. (about a few mm to cm). The rest is trimmed off.

So if your joystick is set to travel from -1 to +1, the necessary movement is indeed very small, especially as the joystick's travel will not be the +/-90 mm of the yoke.

You may like to either change the settings of the joystick, or add a function that works like the F-16 flight computer. This fly-by-wire system uses a speed squared dependent gain to the joystick input such that at any speed the same angle of joystick input creates the same aerodynamic force and therefore acceleration.

[dany93]

Thank you, @okroth for your detailed and very interesting explanations!

[dany93]

@tonghuix Is your joystick accurate enough? Sometimes, the potentiometers are worn and they become noisy, particularly close to the neutral, (not with Hall effect JS). To get an idea about this, try by controlling with the mouse (which, of course, doesn't have this issue). And, important, use the trim.

[tonghuix]

@okroth thank you very much! @dany93 yes, my stick using Hall effect JS (T16000M), very very accuracy, that's why I found this question, because very small 'vibration' will cause a huge and rapidly movement on the aircraft.

Now what I have done is set a larger range of 'dead zone' near neutral, and train myself to adopt this! At least now I can fly a rectangle course in ±200 ft. well still need more practice to meet ACS in ±100 ft.

I do trim off the plane!

[dany93]

I also have a T16000M, that's an excellent JS for its high accuracy for a reasonable price.

Sadly, I've read that these JS work loose with time. Which is a pity for their accuracy, annoying thing near the neutral position.

[tonghuix]

Hi, I suggest we could make the plane response pitch input not such active and fast especially in low air speed.

I found it is very odd that the pitch become more sensitive at low air speed (lower than 75kt), and less sensitive at high air speed (faster than 85kt).

[okroth]

That IS odd. Usually the required displacements (joystick / yoke inputs) are larger on lower speed. To flare out before touchdown, one has to pull out the yoke nearly fully, while in mid-air, one moves it by millimeters. Is the airplane still too sensitive when you throttle down to idle? If not, the propeller induced air speed effect may be too high.

[dany93]

I found it is very odd that the pitch become more sensitive at low air speed (lower than 75kt), and less sensitive at high air speed (faster than 85kt).

I don't find this. At low airspeed, throttle it up makes it more sensitive than at idle, but I need at worst same elevator amplitudes at high regime low airspeed, and more elevator amplitude at low regime. Never less elevator amplitude at lower speeds.

At idle aiming at stall in horizontal flight (without flaps), I need 80 - 90% elevator up just before stall.

@okroth, could you try it and give your opinion about this elevator sensitivity? Unless you share your knowledge (thank you) but you don't have access to this FG c172p. :=)

[tonghuix]

Yes, I also use nearly full displacements at low speed, but the sensitively not triggered by displacements also by small movement.

[tonghuix]

Here is a SOP for experience this effect: @dany93

0. Take off and climb above 1500ft.
1. Set 1500 RPM and maintain altitude
2. Set landing configuration (flap full)
3. Maintain altitude and slow to approximately 45-50 kt just 5-10 above VS0 stall speed. Avoid stall warning activation
4. Engine power as required to maintain airspeed
5. Accomplish level flight, climbs, turns, and descents
6. Recover

Try to feel the sensitive and control efficiency during the slow flight.

[okroth]

First: I used the mouse as flight control.. no force feedback ;-) Second: flew with 45kt and flaps fully extended. But had more than 1500 rpm. How can you steadily fly with 1500 rpm and 45kts? With that settings I had a sink rate of 300fpm. (Which is realistic) Any pull squeaked the stall horn. Nevertheless, control input reactions were as I would expect them; sluggishly, not critical. Check your joystick, it may have worn out. And do not fly that dangerously. In real life you fly full flaps at final approach only, with appropriate sink rate. Actually full flaps only if you approach a airport with short runway. Normal airports with 20° flaps. Reason: with full flaps, there is no real flare, it just sits down; so the approach must be quite precise. No theory, experience.

[tonghuix]

1500 RPM is only a reference to reduce power.

And the landing configuration (full flap extension for c172) slow flight is a check item in ACS of FAA PPL. I believe it could also usable for FG.

[okroth]

OK,

now that manoeuvre makes sense. With about 1950rpm level flight is possible. I did not notice an instability in pitch, however, the plane seems to flat-spin quite easily.

[algefaen]

(Re slow flight sensitivity)

Try to feel the sensitive and control efficiency during the slow flight.

On a C172 I think you'd mainly notice the sluggishness of ailerons and not so much the elevator, since it requires quite a bit of power (->propwash) to maintain altitude at slow flight speeds. As a skydiver I can testify that the air is moving significantly faster in the propwash compared to the surrounding air. (=

I wish I had a good non-destructive, non-hazardous method for logging control input + airspeeds, engine parameters, aoa, etc. in our steam gauge cessna...

[tonghuix]

I just found a lot of nice videos in a C172R (yes not P, but still could be a reference), in these videos, pilot installed a GoPro camera in front of his chest, and it record his control of yoke and the plane's reaction.

https://www.youtube.com/watch?v=WTvh_J0skWo

https://www.youtube.com/watch?v=AKkmmICGsFM

I found the plane react very sluggish than C172P-detailed, not only pitch but also roll.

[okroth]

nice finding! See that even with sluggish reactions, the movement of the yoke for roll is small (only 10° ... 15° turn)

[tonghuix]

@okroth Yep, so I think it could be a bug in this c172p-detailed or in FDM.

It would add some dynamic delay for yoke input, and at slow flight add more delay.

[dany93]

@tonghuix, I think that we didn't understand it by the same way. I read:

the movement of the yoke for roll is small (only 10° ... 15° turn)

Which makes 11 to 17% of full deviation. In small to moderate turbulences. Less in calm air. Am I misunderstanding the message?

[dany93]

@tonghuix,

I have no problems at controlling the aircraft. However, I've made tests and it seems that the aileron authority might be decreased a bit. I almost sure it will improve for you, but this is to be tested and validated in enough different conditions (15 - 20 kt crosswind landing, severe turbulences,...).

In c172p.xml, line 1476, try this Replace the initial value (0.2290) for 0.15 as shown below:

            <function name="aero/coefficient/ClDa">
                <description>Roll_moment_due_to_aileron</description>
                <product>
                    <property>aero/qbar-psf</property>
                    <property>metrics/Sw-sqft</property>
                    <property>metrics/bw-ft</property>
                    <property>fcs/aileron-pos-rad-avgd</property>
                    <value>0.15</value> <!-- ini 0.2290, 12 april 2017 -->

Do a backup copy of your initial file before, and, if you are using git, restore with the initial one before pulling.

[tonghuix]

@dany93 it is good for roll, so how to modify it for pitch?

[dany93]

For pitch I see no good solution for you in the aircraft files. You can try by decreasing the value (initial -1.8, for -1.5 etc....). Not much, there is only a small margin) line 1606

            <function name="aero/coefficient/Cmde">
                <description>Pitch_moment_due_to_elevator_deflection</description>
                <product>
                    <property>aero/function/qbar-induced-psf</property>
                    <property>metrics/Sw-sqft</property>
                    <property>metrics/cbarw-ft</property>
                    <property>fcs/elevator-pos-rad</property>
                    <value>-1.8</value>

But you will see that the risk is of no longer being able to reach the stall AoA. At low speed for landing, you want to do the flare and slowing down, pull the yoke full, the aircraft will not pitch up enough.

The other way is to apply a <power> to your JS elevator axis. Starting by <power>2. It will apply for all aircraft.

[tonghuix]

More video:

• A whole traffic pattern, with a roof-installed GoPro camera https://www.youtube.com/watch?v=Rs1b7uSdCu8

• A beautiful girl's first solo, multi-camera, https://www.youtube.com/watch?v=3jfRs2xsjtc&t=50s

These two videos show pitch and roll are all sluggish, I still suggest add 'delay' to yoke controls, but not rudder.

[okroth]

I have tried to find some FDM coefficients of the C172 from the Internet; the major reference seems to be some data from Roskams Book "Methods for Estimating Stability and Control Derivatives of Conventional Subsonic Airplanes", as others refer to it, namely the UIUC icing model. Another author David Allerton had in the files accompanying his book "Principles of Flight Simulation" a FDM of the C172 with slightly deviating coefficients.

There seems to be a typo in the coefficient Cnda "yaw from aileron deflection" in the C172P, and C172R. It is -0.0053 instead of -0.053 as in Roskam and Alerton's models. You picked it up and doubled it to -0.016 (to compensate the -2,...,+2 range of the summed ailerons), however I think it should be -0.16. With your current value it will fly nicely, but 10 times nicer as the real thing.

Your FDM is the only that has Clp = -0.968, which is double the -0.47 (or -0.484) used elsewhere. This will slow rolling rate significantly (50%).

All other FDMs use Cmde = -1.28. A Cmde of -0.18 definitively makes the plane too sensitive to elevator.

The side force from rudder CYdr is set to 0.15 in your FDM, the value used by anyone else is 0.187.

You aileron authority Clda is 0.229 (compensated by the -2,...+2 range this is 0.115), which again is significantly smaller than the 0.178 used by the others.

Together we have reduced rudder side effects, aileron response(64%), roll rate(50%), drastically reduced adverse yaw(10%), and increased pitch(140%).

I think, tonghuix was right in saying that pitch is too sensitive. It is just that his joystick may have too much gain on top.

[dany93]

Thank you very much @okroth for this research and indications. I'm looking at that.

This FDM had been initially done by David Megginson, with his coefficients (which come from Jan Roskam). When I took it again, I firstly tried to respect as much as possible his work. However many changes have been done since...

[okroth]

Oh, do I read Cessna 182? I always thought the coefficients were computed for a Cessna 172

[dany93]

Yes, Megginson's table data are written for Cessna 182. David Megginson took almost (if not all) of these coefficients for his Cessna 172 FDM. My tendency is to trust him (anyway, he is much better than I am on this). Probably that he had no other set? I discovered these tables only a certain time after the C172 FDM.

[okroth]

The C182 is somewhat larger and heavier (The C172P has a 2450lbs MTOW, the C182 2650). That should also influence the moments of inertia Ixx, Iyy, Izz. The coefficients may be similar, though. Afaik, the C182 is somewhat smoother in air than the C172, which could be just by size and weight.

[okroth]

Found some comparisons. The C182 has 1" shorter wingspan, 1 sqft less wing area, and is 2" longer. It has a larger engine, constant speed prop, and higher weight. I presume the 2" additional length go into the cowling that incorporates six cylinders compared to the 4 in the C172. So, Megginson was quite right to use the C182 coefficients. But we should consider to adjust Iyy and Izz as these are affected by the heavier engine. Iyy should be fine as this is largely dominated by the wings's weight.

[dany93]

But we should consider to adjust Iyy and Izz as these are affected by the heavier engine

Calculating these moments of inertia is very difficult, I don't know the accuracy of Megginson's values. I tried such calculations but they where only approximate. Rather a checking, validation for myself... taking Megginson's values as reference! I remember having tried changing them and having found no noticeable difference. Although I don't remember when and by how much. If necessary, do you have an idea of what we should enter for Iyy and Izz? (keeping Ixx)

[okroth]

I would try to figure out the weights of the 4cylinder and 6cylinder engines (with prop) and then their contributions to Ixx and Izz as point masses under consideration of their distance to center of gravity, Using the parallel axis theorem (Steiner's theorem).

[okroth]

The section 6 of the POHs gave some information:

C182P: empty weight:1772lbs, moment:63300lbs-inch -> CG is 63300/1772=35.7in aft datum engine O-470-U 442lbs @ 17.5 in fore datum. This is 270slug-ft²

C172P: empty weight:1467lbs, moment:57300lbs-inch -> CG is 39in aft datum engine O-320-D2J 281.5lbs @ 19.6 in fore datum. This is 209slug-ft²

C172R: empty weight:1639lbs, moment:64400lbs-inch -> CG is 39.3in aft datum engine IO-320-L2A 297.8lbs @ 18.6 in fore datum. This is 215slug-ft²

So for C172P, Ixx and Izz get (270-209) = 61slug-ft² smaller For C172R, these inertias are reduced by 55slug-ft²

If the original report of Roskam did contain data for a C172, for which I found indications, we have to change Ixx and Izz for the C182 accordingly.

[dany93]

Many thanks once more @okroth, for these informations and (more) having done and detailed the calculations.

In principle I can do these calculations (I have a table for this) but, between the units, CG shift and everything, there is more than enough to stumble. I checked them just to be sure that I understood well that you mean, but having your values as a reference was very precious.

[dany93]

I have changed a few aerodynamic coefficients.

@okroth : Sorry to sollicitate you as much, but do you have the possibility to test the aircraft in FG? Part of this is subjective, based on memory and, when someone complains, I'm not the better one to decide.

CnDa: -0.16 is impossible. Even -0.053 is way too much. I have set it at -0.03. Do you feel it's correct? I think it must not be higher. Rather slightly lower, maybe. The irony is that I was led to decrease it because some people complained that the adverse yaw was excessive... I hope that you judgement will help to arbitrate. Having decreased a bit ClDa compared to the table value (hence increased the aileron deflection for a given effect, see below), I think that the CnDa value must be lower than the tables too. Or we restore ClDa at the table value.

Clp: It was -0.484 at ordinary roll rates (up to 5 rd/s), then (only at higher roll rates, to avoid FDM divergence) -0.968 to -1. I've set it at 0.484.

Cmde: changed (decreased) for -1.28. However, with this value, the stall AoA could no longer be reached (at least with the pilot alone). I recovered this possibility by moving the AERORP a bit foreward (closer to the CG). It seems enough in most situations to reach stall AoA (although with a weak or no margin), but not in a turn with the engine at idle. Do you think that is enough for the stall behaviour, or do I have still to move the AERORP forward? (I'm tempted to do this). Megginson had put it even closer to the CG (at 2.2 inches). I am at 5 inches: AERORP = 44 for CG 39.06.

CYdr: If I remember well, this value had been decreased (by me, after discussions on the forum) to be fitted with the rudder moment.

ClDa. I'm confused about what you are meaning. 0.229 (current value) is the value in Megginson's table, higher (not smaller) than 0.178. However, after my tests, I think that it can be decreased down to about 0.15 without drawback, which would make it less sensitive to the ailerons. I've set it at 0.178.

Moment of inertia: decreased Iyy and Izz after your indications. I think it is not sensitive, but at least it is more rigorous.

There is probably some more refinements to be done (trim presets, maybe more on stall behavior...) but prior to spending more time on this details, I'd prefer having your opinion about the most important points.

Still thank you, I've added your nickname in the FDM file.

[okroth]

First: Megginson has three setups: climb, cruise, and approach, with partly different coefficients.

CnDa: Megginson has -0.0504, -0.0216, -0.0786 for climb, cruise, approach... not sure why they differ, but the C172 does have adverse yaw...UIUC, Roskam, etc all use -0.053.

Clp: 5rad/s is quite a lot ~ 285°/s. All other models (I found) use -0.47...-0.494.

CmDe: FlightGear says about AERORP "the location where the aerodynamic forces are applied to the model". I would expect this to be the 25% line of the wing. According to the POH, this is 25.9 + 0,25*58.8 inches aft datum, i.e. 40.6 inches. That would explain why you can not stall, as the allowed CG range is about 5" fore and 8" aft this point, and when you set it at 44", and fly alone with full tanks, your CG will be too far ahead.

CYDr: I will check. However, it looks as if it uses lv-ft which is in the metrics section: CYdr lv-ft = Cndr bw-ft And lv-ft is a little greater than lh-ft, as the fin is slanted backwards which shifts the aeromatic center back too.

ClDa: Megginson has -0.229 consistently. Roskam had -0.178. I think I will make an estimation using XFOIL and spreadsheet... That went quite well with the rudder hinge moments.

Regarding stall (and spin entry): you may have to consider yaw rate as this is the way to change from one to other. I.e. near stall AoA, the wing yawing forward has a lower AoA, higher lift, higher roll moment (from lift), and less drag. As long both roll moments (from lift) are balanced, the plane will not spiral off. As soon only one wing stalls, the roll moment of the other will push the plane over.

[dany93]

Thank you, @okroth

At least for the easiest, the AERORP at 40.6 seems to work very well, I recover "my" aircraft, with its stall and spin behavior. Possibly even a bit hard, I have to see it (I think I can refine). Very pleasant to fly.

I was afraid for the CG which can be aft the AERORP, but no problem with CG at 44.5 in.

[okroth]

CG may be behind AERORP, it must however be in front of the airplane's neutral point, which can be computed easily using the areas of stabilizer and wing an the horizontal lever arm. In a C172, it is 20.7 in behind the 25% line of the wing, so you still have plenty of space. Nevertheless you got an experience in load and balance :-)

Regarding Ixx and Izz: The 180HP and 160HP engine are the same, the 180HP just has a higher max .RPM. So at least that can be handled easily.

[dany93]

New version with your indications. With them, I'm learning here, consolidating my knowledge there. Very interesting. Thank you.

IMHO, the aircraft has been greatly improved. I hope being right.

[wlbragg]

I like the change, definitely not so sensitive yet I still have a good range of motion. Stall is good!

[onox]

@dany93 Can you update the first comment (or add a new one) of #958 ?

[dany93]

Thank you @onox, thank you @wlbragg. https://github.com/c172p-team/c172p-detailed/issues/958 updated. If you can have a look to check the formulation, thanks.