Board index FlightGear Development Aircraft

737-300 Classic - New and Improved

Questions and discussion about creating aircraft. Flight dynamics, 3d models, cockpits, systems, animation, textures.

Re: 737-300 Classic - New and Improved

Postby Milo » Tue Jul 15, 2014 1:56 pm

Thank you very much Soitanen. The pushback works at the best. Thanks for making this plane acting so realistic.
Concerning the autopilot bug, it happens randomly when I turn on the autopilot. As mentioned, I use the latest Git version of FGFS. I have tryed to simply configure the autopilot on the ground, turn it on and see if the bug happens. It did some times. I simply turned the heading on, turned the vertical speed to 1800ftpm or something, and turn on the autopilot. It does not bug all the time but too often.
It crashes the whole sim! I have to restart.
Regards
GVX0250 _ FGFS v.Git _ Win10_ Asus N76V, RAM8GB, CPU: i7 2.4
Milo
 
Posts: 48
Joined: Mon Feb 07, 2011 12:11 am
Location: Luxembourg
Callsign: GAX0250
Version: Git
OS: Win8.1

Re: 737-300 Classic - New and Improved

Postby Soitanen » Tue Jul 15, 2014 3:15 pm

Milo, do you try to turn on autopilot, while staying on the ground? If so, that it's wrong usage. But, anyway, I never get this error on my computer.
Correct takeoff procedure (omitting flaps, brakes): turn on both FD. Turn on AT ARM switch. Set V2 on MCP speed window. Press 'd' button on keyboard. It's TOGA button on the throttles in real aircraft. Engines will accelerate to take-off thrust (now only maximum, no derated takeoffs for now), FD will show you desired pitch angle. At 400ft or higher select roll mode (HDG SEL, LNAV, VOR). Turn on autopilot not lower than 350 ft, or it will be disengaged.

New small feature now - high idle thrust. When you are in air or only 4 seconds past after touchdown, engines will work at 32% N1 instead of 22%.
Boeing 737-300. Reworked cockpit, FDM, autopilot and much more. WIP.
Boeing 737-800. WIP. Canvas PFD and ND.
Antonov An-24B. Made from scratch. Very good FDM. 3D model by Adrian. WIP.
Project Russia (some cities, based on OSM with custom objects).
Soitanen
 
Posts: 489
Joined: Sat Jun 16, 2012 7:50 am
Location: Saint-Petersburg, Russia
Version: git
OS: Linux Mint 17

Re: 737-300 Classic - New and Improved

Postby Honzaku » Tue Jul 15, 2014 4:00 pm

Hello Soitanen,

Regarding the document with aerodynamic data for simulator; I am afraid it will be very tought to get this one. I have already tried to find it several times and without any results. Last time I tried to find that document I only found it for B 777-200. But the ftp server where I googled it was accessible only few hours - obviously it was some security bug, which was (sadly for me) promptly fixed. Seems, these data are extremely anxiously protected by aircraft manufactures. Probably selling these data for simulator manufacturers is big portion of their business along with selling the aircrafts itselfs.
Maybe if you have some good friend in company which maintain some of these simulator or maybe if your friend captain has, you could get these unofficially from him.
Not sure if it would be affordable for us to implement these data in flightgear because it is very large document, but it is very interesting to read it anyway. It contains several thousand of pages of graps of coefficients. It is nearly exactely same as JSBSim but unlike the ordinary flightgear aircraft which is driven by (I guess) about 20 look tables, this FDM is driven by several hunderts of these.

BTW: If it werent ilegal it would be great to show this documet publicaly to all of the people who still insist on the statement the JSBSim is just for game but YaSim is for professional simulation.
:wink:
Regards

Honza
Honzaku
 
Posts: 88
Joined: Sat Jan 14, 2012 6:00 pm
Location: Between LKPR and LKVO, Czech Republic

Re: 737-300 Classic - New and Improved

Postby Soitanen » Tue Jul 15, 2014 5:30 pm

By the way, Honza, did you try latest FDM of this aircraft? What can you say about it? (Note about bug - thrust is too big on high altitudes, will solve it)
Boeing 737-300. Reworked cockpit, FDM, autopilot and much more. WIP.
Boeing 737-800. WIP. Canvas PFD and ND.
Antonov An-24B. Made from scratch. Very good FDM. 3D model by Adrian. WIP.
Project Russia (some cities, based on OSM with custom objects).
Soitanen
 
Posts: 489
Joined: Sat Jun 16, 2012 7:50 am
Location: Saint-Petersburg, Russia
Version: git
OS: Linux Mint 17

Re: 737-300 Classic - New and Improved

Postby Honzaku » Sat Aug 09, 2014 7:01 pm

Hello Soitanen,

Finally I have tested your latest version of B733. So here is my opinion on FDM:

-FD and autothrust work fine during the T/O. Very nice feature.

- Recently I am bussy so I havent had time to studying the procedures, therefore I didn't check the model against the "real numbers" in flight. But Boeing is quite plesant to fly in the free air now. Thats my impression.

- Takeoff. This is, in my opinion, curently the biggest issue of this model. I wasn't happy with the way it rotated before you started to make any changes of FDM and unfortunately I am still not satisfied with it.

I tested the default load setting of the ACF as you got it when the simulation is started. takeoff weight is aprox. 116000lbs. Vr should be betveen 133 and 141 kt according to the V speeds tables in the help. Sadly there is no indication of the COG positon along the lenght of MAC so it is impossible to set the trim correctly and check it againts real load diagram. So I follow your recommendation from this discussion to use trim about 4.25 units. Used flaps 5 as usuall for T/O for 733. When "virtual copilot" told me Vr somewhere about 140kts I pulled the mouse and nothing happend I pulled more and still nothing happened. Far beyond the V2 the beast finally rotated but immediately after being airborn it extensively rotated nose up so even massive forward pressure on the yoke (ehm, mouse) didnt prevent me from overshooting the desired climbing attitude. instead of desired 15degs I managed to stop the rotation at about 25degs. When I finally stabilized the ACF in the attitude comanded by FD, I tried to centre the control by 5 button to see for which attitude is the ACF balanced with the trim of 4.25 units. And ACF pitched nose up to 30 or more degs so I had to immediately apply strong forward pressure to prevent the stall.
In next step I tried to determine which trim is needed to keep the ACF around 15deg attitude with flaps 5 and engines in TO regime. It was around 3.2 units. So for this load configuration the T/O trim must be smaller than 3.2 unit to leave some space for pulling of the yoke by the pilot during the rotation and establishing the climbing attitude and keeping it. But this trim moves the rotation even further beyond the V2.
To be sure the problem of the rotation is caused by insufficient authority of the horizontal tail to rotate the plane to correct attitude and not by insufficient lift at the Vr and 10 to 15deg attitude I did the last test. At flaps 5, gear down I keep the attitude between 12 and 15 degs aprox. the speed was kept between 135 and 142 kts. even with less than 70 % of N1 I was able to keep horizontal flight or establish 5fps climb. With about 80%N1 I nicely climbed about 10fps at 15 degs and 142kts. And I definitely wasnt in the ground effect. So in the right attitude the plane is able to airborn at Vr or higher.

Conclusion:

Atleast for this load settings the authority of the horizontal tail is sufficient for controlling the ACF in the air. but is absolutely insufficient to rotate the plane at Vr from the taxiing attitude to climbing attitude.

The pitch moment of the horizontal tail required to rotate the plane to climbing attitude on the ground is far bigger than pitch moment required to stabilize the plane in initial climb in the air. And that is serious problem. The moment required to initiate the rotation must be smaller than the pitch moment required to keep the initial climb attitude.

How to solve it:

Create function which shows the actual position of COG along the MAC in the fashion of Tu-154B. Leading edge body station of MAC (LEMAC) for B733 is 625.6in. datum for body stations is 130in before plane nose (so nose tip has body station 130). But check the body station diagram in SRM (Structural repair manual) you hopefully downloaded from my link to pulsaraviation since it is not available there anymore. Body station notation is bit confusing. 625.6 can means the start of MAC is 625.6 inches from the datum but not necessary. If there are some body station indexed with letters before this station this means there are inserted some section. So for example if there are body stations 351a, 351b and 351c inserted and each of them has e.g. 40in so the actual position of every station behind station 351c is actually 3*40in further from the datum than it is in its name. ie the station 400 is positioned at 400+3*40= 520in from the datum. The length of the MAC of B733 is 134.46in. Having these we can check where is the actuall position of the ACF and whether it is within the limits.

Do the pitch moment analysis at the initial climb and in the rolling just before the rotation. The COG is fore of the main gear. During the rotation the plane has to pivot around the main wheel. Weight of the plane concentrated in the COG generate nose down moment given by weight times to arm between the COG and main gear, thrust generates nose down moment given by thrust times to height of the engine above the gear wheel centre. Total lift act in the COG and generates nose-up moment given by lift times the arm to the main gear. But because the aerodynamic reference point is usually aft of the COG there is nose down moment given by the lift times to arm betveen COG and AERORP as well. At reaching Vr the sum of these pitch moments is nose down and must be nearly balanced by aerodynamic nose up moment of the stabilizator, but must be still bigger. By pulling the elevator you add that bit of the nose-up moment which overcome all of the nose down moments and the plane starts to rotate. But you also establish the new aerodynamic moment equilibrium to keep the plane more or less in the climbing attitude and climbing AOA. This must be fullfiled for all load combination within the approved load and balance envelope. Correct ammount of stabilizer trim should provided the same control feel to the pilot during the rotation for any of these cases. That means the pilot is trained to rotate approximately 3 deg/sec and to do so he has to move the yoke by aproximately same speed and has to overcome aproximately the same resistance of the control for any lodcase from load and balance envelope provided the plane is correctly trimed and the Vr is correctly calculed. In other words the pilot should have same feeling during the rotation and works still the same with control as he is drilled regardless the load of the plane provided it is within the load envelope.

Than try to adjust the stabilizer moment coefficient, elevator moment coefficient, AERORP position to fullfill all of these criteria in the whole flight envelope. And chack the correct main gear position.

Best Regards

Honza
Honzaku
 
Posts: 88
Joined: Sat Jan 14, 2012 6:00 pm
Location: Between LKPR and LKVO, Czech Republic

Re: 737-300 Classic - New and Improved

Postby Honzaku » Sat Aug 09, 2014 9:57 pm

The takeoff rotation dynamics is nicely and easily described for example in this document:

http://faculty.dwc.edu/sadraey/Chapter% ... Design.pdf

on the pages 44-49. Modified equations could give you required momement coeficient of stabilizer and elevator.needed for rotation.
Honzaku
 
Posts: 88
Joined: Sat Jan 14, 2012 6:00 pm
Location: Between LKPR and LKVO, Czech Republic

Re: 737-300 Classic - New and Improved

Postby Soitanen » Sun Aug 10, 2014 6:55 am

Hello, Honza!
Current CG position is already computed, you can see it in FMC at takeoff page. Also there is desired stab position for takeoff due to selected flaps.
Also, I didn't redo v-speeds calculation of first officer, but thay are slightly wrong.
Now I'm concentrated on engine thrust table, I want to match QRH climb and descent numbers. After this work I will try to redo stab and elevator moment into direct force on tail. This, as I think, will give the same behaviour in the air and improved rotation behaviour. Moment from stabilizer is calculated correctly, elevator guessed. By the way, I fly with joystick and for me rotation seems ok.
Boeing 737-300. Reworked cockpit, FDM, autopilot and much more. WIP.
Boeing 737-800. WIP. Canvas PFD and ND.
Antonov An-24B. Made from scratch. Very good FDM. 3D model by Adrian. WIP.
Project Russia (some cities, based on OSM with custom objects).
Soitanen
 
Posts: 489
Joined: Sat Jun 16, 2012 7:50 am
Location: Saint-Petersburg, Russia
Version: git
OS: Linux Mint 17

Re: 737-300 Classic - New and Improved

Postby Honzaku » Mon Aug 11, 2014 1:04 pm

Hello, Michael!

Current CG position is already computed, you can see it in CDU at takeoff page. Also there is desired stab position for takeoff due to selected flaps.


I checked that, but this data are left empty on the takeoff screen. Maybe I have to press some "calculate" or "enter" button on CDU to get it. Honestly, since I prefer classic analog aircrafts I don't have experience with these computers and I am used to hand calculation and graphs. :) So I could do something wrong.
BTW: Latest version I have (from git) has graphical bug - CDU keyboard and frame are white, seems the texture is not loaded.

Also, I didn't redo v-speeds calculation of first officer, but thay are slightly wrong.


What do you mean by redoing the V-speeds? Current (old) speeds are wrong because they don't reflect the values for real plane or you have to calculate new V-speeds which fit better to your FDM? Hope it is the first case.
Just one thought about the speeds. I didn't checked how it is implemented in JSBSim, but airspeed indicator shows indicated airspeed which differs from the true airspeed. the difference due to altitude is included in JSBSim I think, but I doubt there is also included aircraft type specific position error which should be coded manually. I am not sure there are calibration graphs in documents we have. But this data is included in POH quite often.

Moment from stabilizer is calculated correctly, elevator guessed.


I think so, that's why I meant the problem with rotation could be caused rather by wrong position of AERORP or main gear than wrong stab/elev moments. Because it seems to me the moment needed to rotate the plane on the ground is inadequately high compared to moments needed to control the plane in free air. But maybe the default load settings are outside the approved load envelope. I didn't check that.

Now I'm concentrated on engine thrust table, I want to match QRH climb and descent numbers.


I am crossing fingers to you. But I am afraid this could be very difficult to do with so few aerodynamic data we have. Performance is about lift drag and thrust. The more accurate these data is the more accurate simulation is with respect to the aircraft performance. As you can see in Tu-154 JSBSim or Boeing 777 aerodynamics documents, this data is dependent also on Mach number. Looking at the B737.xml the lift is based on one lift curve which is only modified by adding some constants due to flaps, spoilers, etc. Drag is based again on one curve modified by adding constants again, there is at least simple mach correction also provided by adding constants due to mach nr. On the other hand most of the flight is performed on rather small AOA for which these approximations could be acceptable.
BTW: You model lift with one curve and add some constant portion of lift for each flap stage. Basically you move the lift curve along the y-axis (in 2D plot) That is how flaps works. The problem is, for higher degrees of flaps, the leading edge slats are also automatically extended. Such lift curve is moved in y-direction towards to higher lift values due to flap effect but also has significantly extended "linear" part due to slat effect, so the maximal lift occurs at higher AOA's.

By the way, I fly with joystick and for me rotation seems ok.

I know, mouse is bad for control. That is the reason why I don't want to do any reviews of planes, since I think the simulation should be controlled with proper hardware - yoke/stick, pedals and throttle quadrant. On the other hand in this rotation problem I don't think there is so big difference in behavior between the mouse and stick. The problem is that to rotate the plane you have to strongly deflect the stick/ mouse back from the neutral position but immediately after rotation you must strongly deflect it forward from its neutral position to keep desired attitude. With mouse it is just moving the mouse, with stick you can feel the resistance of the spring in those movements. But the effect on the plane should be similar. Weird.

Just one unimportant bug to note: if I want to restart simulation by using menu, the simulation falls and you have to start it again.

Regards

Honza
Honzaku
 
Posts: 88
Joined: Sat Jan 14, 2012 6:00 pm
Location: Between LKPR and LKVO, Czech Republic

Re: 737-300 Classic - New and Improved

Postby Soitanen » Tue Aug 19, 2014 5:53 pm

Hello, Honza!

Look into property tree. There you will find /instrumentation/fmc/cg and /instrumentation/fmc/stab-trim-units. This is current CG position (MAC related) and required stabilizer position for takoff with flaps defined in /instrumentation/fmc/to-flap.

About CDU bug - strange, latest git (simgear, flightgear, fgdata, b733) works fine for me, so I can understand what's wrong with your version.

About v-speeds. Yes, current realization is wrong, not match real QRH. So, in futurre I will re-do this table (it's not so easy, and it mus use abilities of derated takeoff).

About ARP position. Now it's placed at 25% MAC, because I have moved moments from stabilizer and elevator into direct force, located at stabilizer. ARP must move when flaps are extended and with Mach (from 0.615 in this aircraft), but I don't have the numbers now.

Flaps/slats - I know what you mean. And even I have proper graph on the paper, but not digitized it yet (lack of time at summer). Now, lift gain is calculated properly, so if you fly with the normal envelope, you will not feel the difference.

Gear position, empty aircraft CG - all is triple checked, there is no any errors.

About high Mach dynamics - I have numbers for drag increase due to Mach from book, and some reverse engineered data from QRH, so drag will be modelled properly. Lift curve must be slighly higher (on low alpha), but must be not so high, as for low Mach (this determines minimum speed at high altitudes). So now I will have little error in pitch on flight level flight. And again, if fly in normal envelope, there is no difference in minimum speed(but this will be reverse engineered from different data too).

So, try latest uploaded version, how do you feel rotation now? By the way, do you use TOGA button on takeoff, or you moves throttles to 100%?
Boeing 737-300. Reworked cockpit, FDM, autopilot and much more. WIP.
Boeing 737-800. WIP. Canvas PFD and ND.
Antonov An-24B. Made from scratch. Very good FDM. 3D model by Adrian. WIP.
Project Russia (some cities, based on OSM with custom objects).
Soitanen
 
Posts: 489
Joined: Sat Jun 16, 2012 7:50 am
Location: Saint-Petersburg, Russia
Version: git
OS: Linux Mint 17

Re: 737-300 Classic - New and Improved

Postby Honzaku » Fri Aug 22, 2014 9:11 pm

Hello, Soitanen!

Look into property tree. There you will find /instrumentation/fmc/cg and /instrumentation/fmc/stab-trim-units. This is current CG position (MAC related) and required stabilizer position for takoff with flaps defined in /instrumentation/fmc/to-flap.


Ah, I see. There it is. I looked for the data in CDU menu in virtual cockpit, where it wasn't. Since the two numbers (CG pos and corresponding trim units) are very important I would recommend to print theme somehow to the screen in the way of your AT helper or in the CDU menu (best option).

About CDU bug - strange, latest git (simgear, flightgear, fgdata, b733) works fine for me, so I can understand what's wrong with your version.


It is probably operation system specific bug or it is the error of my installation on my computer (Win XP 32bit FG 3.0) since I installed FG 3.0 on the laptop with WIN 8 64bit and the CDU textures are OK there.

About v-speeds. Yes, current realization is wrong, not match real QRH. So, in futurre I will re-do this table (it's not so easy, and it mus use abilities of derated takeoff).


I understand.

About ARP position. Now it's placed at 25% MAC, because I have moved moments from stabilizer and elevator into direct force, located at stabilizer. ARP must move when flaps are extended and with Mach (from 0.615 in this aircraft), but I don't have the numbers now.

Gear position, empty aircraft CG - all is triple checked, there is no any errors.


OK, these were just my ideas what could be wrong according to my previous experience.

So, try latest uploaded version, how do you feel rotation now? By the way, do you use TOGA button on takeoff, or you moves throttles to 100%?


I use TOGA button and it works nice. I have much better feeling from the rotation now, however it seems to be still bit "over-rotated" compared to my "etalon" aircrafts i.e. TU-154, but it is much, much better than it was. Now, I don't dare to tell any conclusion whether it is good or not good without having and using correct v-speeds and corresponding engine settings....and at-least joystick with the possibility of trimming.

Carry on with your great work on this plane.

Best regards

Honza
Honzaku
 
Posts: 88
Joined: Sat Jan 14, 2012 6:00 pm
Location: Between LKPR and LKVO, Czech Republic

Re: 737-300 Classic - New and Improved

Postby Soitanen » Thu Sep 11, 2014 11:15 am

Now v-speeds calculated properly and derated take-offs are possible!

Bad news - derate settings are not accessible via FMC, you need to use property-tree.

Go to /instrumentation/fmc/derated-to/
There is 2 bool values:
- method-derate-20k
- method-assumed
You can enable one or another or all methods.
If using assumed temperature method, selected temperature must be setted in property assumed-temp-degc.

After all preparations, TOGA button will move throttles to reduced thrust.

V-speeds calculated automatically with respect to temperature, elevation, flaps setting for take off and aircraft weight.
Boeing 737-300. Reworked cockpit, FDM, autopilot and much more. WIP.
Boeing 737-800. WIP. Canvas PFD and ND.
Antonov An-24B. Made from scratch. Very good FDM. 3D model by Adrian. WIP.
Project Russia (some cities, based on OSM with custom objects).
Soitanen
 
Posts: 489
Joined: Sat Jun 16, 2012 7:50 am
Location: Saint-Petersburg, Russia
Version: git
OS: Linux Mint 17

Re: 737-300 Classic - New and Improved

Postby Milo » Mon Sep 29, 2014 5:57 pm

Soitanen,

Could you ask Tikibar if you could implant his 747-8s CDU in this 737-200?
It would be great to get the paging in LEGS and be able to put the flightleves directly in the CDU ...

Regards
GVX0250 _ FGFS v.Git _ Win10_ Asus N76V, RAM8GB, CPU: i7 2.4
Milo
 
Posts: 48
Joined: Mon Feb 07, 2011 12:11 am
Location: Luxembourg
Callsign: GAX0250
Version: Git
OS: Win8.1

Re: 737-300 Classic - New and Improved

Postby Soitanen » Mon Sep 29, 2014 6:21 pm

Milo, I think it would be better to ask Gijs to change generic CDU code, which is used at least in 744, 733. In this case we will improve group of aircrafts.
Boeing 737-300. Reworked cockpit, FDM, autopilot and much more. WIP.
Boeing 737-800. WIP. Canvas PFD and ND.
Antonov An-24B. Made from scratch. Very good FDM. 3D model by Adrian. WIP.
Project Russia (some cities, based on OSM with custom objects).
Soitanen
 
Posts: 489
Joined: Sat Jun 16, 2012 7:50 am
Location: Saint-Petersburg, Russia
Version: git
OS: Linux Mint 17

Re: 737-300 Classic - New and Improved

Postby Gijs » Mon Sep 29, 2014 6:23 pm

Right, please feel free to send me any updates to the generic CDU and I'd happily push them to Git. It makes no sense to have dozens of CDUs with more or less the same code ;-)
Airports: EHAM, EHLE, KSFO
Aircraft: 747-400
User avatar
Gijs
Moderator
 
Posts: 9544
Joined: Tue Jul 03, 2007 3:55 pm
Location: Delft, the Netherlands
Callsign: PH-GYS
Version: Git
OS: Windows 10

Re: 737-300 Classic - New and Improved

Postby Milo » Mon Sep 29, 2014 8:00 pm

Soitanen,
To answer the question you asked on the 752 topic, there are differences between the generic and the improved 752 CDU, simply by push Ctrl-C and you'd see the the left screen buttons are all active so i can give the Flight level for each waypoint, and to me, the most relevant point is the ability to change the pages in the LEGS screens.
Yes, I do love your autopilot logic, that is why i love the 733 and I hope that the 3dModel will also be improved soon.
Everyday I plan to start doing some job in Blender and fill the little gap between the horiz.stabilizer and the fuselage but I never come to it ... :?

Regards
GVX0250 _ FGFS v.Git _ Win10_ Asus N76V, RAM8GB, CPU: i7 2.4
Milo
 
Posts: 48
Joined: Mon Feb 07, 2011 12:11 am
Location: Luxembourg
Callsign: GAX0250
Version: Git
OS: Win8.1

PreviousNext

Return to Aircraft

Who is online

Users browsing this forum: No registered users and 17 guests