Hi Curt, Ok, removed my joystick, and entered a '5', but still crashed while just in 'circle' mode - no route entered ;=((
As usual Atlas provides a good 'view' as to what happened - added - ATLAS="--atlas=socket,out,<IP>,5500,udp" to output to Atlas running in a 2nd machine... See - http://geoffair.org/tmp/uas-01.jpg for a graph of the flight... The two blips in the graphs show the first stall, but it recovers and begins to climb back, and the 2nd the second stall, this time too low to recover, so into the drink ;=(( CRASH! This is a view of the 'crazy' flight track http://geoffair.org/tmp/uas-02.jpg Obviously the pig-tail loops are the 'stalls'... remember with NO joystick attached and starting with centered controls (NumPad 5)... And if you want to load this track into Atlas, or further study speeds, etc, then this is the Atlas track data :- http://geoffair.org/tmp/uas-01.txt Then on the NEXT flight I tried :- IO="--generic=file,out,10,uas-02.csv,playback" Then I added a header line, to help analyze it in say an OpenOffice spreadsheet import - see - http://geoffair.org/tmp/uas-02.csv On this 2nd flight, this crash took longer, since it (randomly) turned left first, where as mentioned it holds more stable, but then eventually went into a right turn, stalled, recovered, stalled again, and CRASHED... And as you know well, downloading this file, and using say - $ ./fgfs --fg-root=/point/to/fgfs/data --timeofday=noon \ --aircraft=f-14b-uas --carrier=Vinson \ --generic=file,in,10,uas-02.csv,playback --fdm=external you too can enjoy this fateful flight ;=)) In 'chase' view, you can clearly see the right roll increase, the nose coming up, and the stall, recovery, then repeated, and BANG, into the water... I know it is difficult to work on, debug, fix something that obviously does not happen in your case... Maybe if you do not enter any route, or something... And this is all with SG/FG git of 2011-09-14... Any other ideas? Regards, Geoff. On Thu, 2011-09-22 at 14:00 -0500, Curtis Olson wrote: > On Thu, Sep 22, 2011 at 1:25 PM, Geoff McLane wrote: > Hi Curt, > > A pleasure, and FUN ;=)) > > Yes, I know a low frame rate can play havoc when > you are trying to fine control an aircraft from > its attitude feedback, and I should have mentioned my > rate, but is always in the high 50-70 fps range in this > Ubuntu machine... so should NOT be a factor... > > > Ok, 50-70 should be perfect. > > I just did another few runs, and this time it crashed > just while circling... it was in a right bank, which > got too much and the nose came up, and it stalled... > I am mostly in the 'chase' view... > > > This is really strange. I have seen nothing like this except when I > inadvertantly applied external control inputs through a strange > combination of linux virtual desktops and flightgear capturing the > hotkey to come back to the FlightGear virtual desktop. > > > So two thoughts here. If you have a joystick connected, could you try > unplugging it to see if that helps? Could you also press "5" on the > numeric keypad to make sure all the flight control inputs are > centered. Because of the way the F-14b FCS is wired together in > combination with the yasim flight surfaces, you can still input > elevator and aileron and trim and cause conflicts that you might not > see in other simpler aircraft that use aileron and elevator directly. > > > The first time this happened at 2000 feet, it caught > itself - leveled a bit and bumped the throttles, and > began climbing back... > > But a little later, 20-30 secs, it happened again, and > this time was still too low to recover, and SPLASH... > > I had not previously let it fly in the 'circle' mode > for too long, but now note if I leave it in circling > mode, it will eventually end up in the water... seldom > lasts more than 5 or 10 minutes... > > You seem to be deliberately holding its speed down > around 150 - I see air-brakes come up when greater > than this, and throttle back - and although flaps (I think > full flap?) are still applied, 150 must be quite 'low' > for this sleek bird... > > > Normal landing approach in the real aircraft I believe is about 120 > kts? I fly 135 kt approaches in the simulator. It should be able to > hold 150 kts with the flaps down pretty easily. The point of slowing > way down when circling is to keep the circle radius small enough so > you can see what you are looking at. If you fly the circle at 600 > kts, your radius will be 20 miles (just guessing) :-) and you won't be > able to see anything. > > And I am not sure how many degrees each marking on > the hud bottom bank indicator represents, and while it > starts the banking in between the 1 and 2 of the 'big' marks, > at the stall point it is beyond the 2nd big mark, approaching, > even reaching the 3rd big mark, which is on the > horizontal - ie 90 degrees! > > At the moment of stall it loses 1200-1400 feet in 1-3 > seconds... while it can happen in a right or left turn, > it does seem to happen quicker in a right turn... > > I now understand the 'reset' is a full sim reset, > but that is not too helpful if you have set up say a > particular weather, wind or something that you want to > repeat... must get around to feeding that in, in > the command, so a reset puts it back (I hope)... > > > Well complain to the developers if a "reset" resets too > agressively. :-) > > If you could describe a bit more where some of this > is decided/calculated I too could try tweaking some > values... > > I would probably bump the speed a little, and really > watch the bank angle... those stubby little wings do > not give much lift anyway, but the slender body gives > close to none ;=)) > > As mentioned, I too have more than a passing interest > in automated flight control... > > Regards, > Geoff. > > PS: OT: I too searched a little for the expression to be > a 'bunny', but could not really find anything ;=(( > > I am sure it comes from my Australian origin, and > has the meaning to take the rap, take the medicine, > sort of to be the guinea pig... > > > Ok, I can understand that. Here we test cosmetics on little bunnies > (so I'm told) but I'm sure we have to shave all their hair off first. > > Why poor little bunny rabbits feature I just do > not know ;=)) > > Maybe from when Australia had a big war on > rabbits _MANY_ years ago, and put out millions > of traps for the bunnies, as well as other methods, > like poisons - myxomatosis... > > So to be a bunny was to be trapped ;=(()) > > Maybe other Australian's have a better memory than me, > and can explain it better, but meantime I will blame > my parents, or the Australian educational system, for > giving me such a stupid expression ;=() > > On reading up on 'to be a Guinea Pig', another > very confusing expression - not really a 'pig' and not > from Guinea! - I found a reference that in Johnston's > Natural History, they go by the name Spanish Coney. And > 'coney' was the old name for a rabbit, a bunny... Huh! > > > In Peru Guinea Pig is a delicacy ... cui ... never had it myself that > I'm aware of. > > Maybe the early immigrants to Australia decided to mix > it up even more! ;=)) Or got it confused on the long > boat ride half way around the world... > > > Well just to summarize, if your frame rates are solid in the 30-60+ > range, then the next thing I'm wondering about is a joystick or other > means of extraneous control inputs that could be confusing the F-14b > AFCS. > > > Thanks, > > > Curt. > -- > Curtis Olson: > http://www.atiak.com - http://aem.umn.edu/~uav/ > http://www.flightgear.org - http://gallinazo.flightgear.org > ------------------------------------------------------------------------------ All of the data generated in your IT infrastructure is seriously valuable. Why? It contains a definitive record of application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-d2dcopy2 _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
