LNAV not working in ILS approaches

[0zz]

Tried my first ILS approach today. ILS03 App to WAAA. Course and NAV radio set correctly

Up until transition I was in GPS mode. It automatically switched to LNAV got the ident of the localizer on screen and all fine. But AP switched to heading mode and steered off course.

Tried it several times. I mean if you are exactly on course and set the heading bug to runway direction, with magnetic deviation, it flys as if it is doing approach. However, this is not LNAV at all.

I don't think KFC225 should act like that. Anyone else experiencing same thing?

[sdflyer]

Where you in APR mode?

[0zz]

Not while in GPS mode. I generally hit approach button after GNS switches to VLOC, before the GS capture marker. 

[sdflyer]

5 hours ago, 0zz said:

Not while in GPS mode. I generally hit approach button after GNS switches to VLOC, before the GS capture marker. 

If my memory serves me well APR must be engage for GPS LPV guidance.

[0zz]

But this is an ILS approach. The earlier GPS navigated part shouldn't have an effect on the final approach. I might as well being vectored to final with HDG hold. It felt like VNAV is properly executed with localiser guidance, but LNAV is not.

Hmm. There is something not right with how AP behave. I mean I didn't used bendix/king in real life but quite a lot of addons use the same device. For example, the one on Carenado B58 keeps vertical and lateral attitude when AP button is activated. In BN-2 it directly goes to VS mode with 0 setting, resulting a sharp pitch down during climb. Im not sure if this is how it behaves in real life or not, but it is disturbing.

Weird. Maybe someone with a real life experience on BN-2's can pinpoint the issue.

[Ch.Cole]

I don't quite understand. I just tried an ILS approach by flying HDG until the needle moves, then I tried NAV, which captured the localizer, then APR, which guided me down on the GS. Then I did another approach where i directly went from HDG to APR. Did you change the CDI on the GPS from GPS to VLOC?

[tADEUS73]

He talks about the autopilot-behavior where the GPS guidance automatically changes to NAV/ILS when the localizer is caught. It's a standard feature of most planes that use the G530 and it was also working for me lately in the BN-2 when I was flying it. It sometimes malfunctions if you play around with the approach waypoints and not just fly it as given. That saying, it also malfunctions due to many other reasons in all the planes that use this system. Sometimes the turn is to sharp and the plane cannot make it and resigns (happened to me in other planes and the BN-2) or the plane will give up trying to go down the G/S if you are not low enough - also seems to be happening in many planes. It's rare but it happens.

That saying I don't remember if I just allow the autopilot to do its thing when it switches automatically or I press APR in addition to that

[0zz]

Exactly thanks tADEUS73. After some more testing, I finally noticed the tiny INOP sticker on FD button. Man, I am blind!

I think the behaviour I was missing was FD mode, as it is not coded (yet?) AP module tries to make up for it by force switching to HDG or VS mode instead. 

That cause, abrupt behaviours like I explained. It steers out of course during GPS to Localiser transition if heading bug is not set as same as course bug. Or it just zero out vertical speed when AP is activated during climb.

 I am not sure if there is a plan to implement FD mode on AP. If not, at least the code has to be amended for a smoother transition.

Simply, when AP button is activated, it should hold the attitude until vertical and lateral modes selected. VS shouldn't be automatically reset to zero.

Or heading should automatically be matched to course when GNS auto-switches CDI.

Edited May 7, 2020 by 0zz

[tADEUS73]

Well I never had to set the heading bug to the runway direction in any of my G530 planes for it to work. Just fly the approach on GPS, wait for it to change to NAV or change it yourself, when it's centered on the localizer press APR to enable the vertical guidance. Works (almost) every time. At least if you aren't on some crazy position in relation to the glide slope already, otherwise the plane will most of the time give up. I also didn't notice a violent change from GPS to NAV.

[Graeme_77]

The GPS should not be able to “auto-slew” a mechanical HSI, the pilot must always set the course bar on the HSI to the GPS DTK value. Likewise flying an ILS it must have the course set on the HSI. The GPS drives the HSI, and the autopilot “reads” from the HSI, not the GPS. The BN2 as implemented currently does not require this (I’ve reported as a bug) but in general it’s a good idea to operate the HSI correctly. 

 

(That’s the reason the GNS gives you the next course and a countdown timer approaching the waypoint)

Edited May 7, 2020 by Graeme_77

[Graeme_77]

Having done hours (literally - it's 03:55AM!) more research on this, the KFC225 does support a form of GPS Steering.  With the GNS530W in GPS mode it signals to the KFC225 to use the GPS ARINC 429 information to steer with rather than the analog HSI deviation, although it must use analog guidance for GPS approaches - see the link below. The KFC225 effectively can work in an analog or digital mode, and that mode is triggered by a pinout on the GNS530W.

Now, for reasons apparently from the 1990s, when the GNS530W is changed between GPS and VLOC mode, the KFC225 will, IRL, drop back to ROL mode. 

So when the GNS530W is used with a KFC225 the auto-switching between GPS/VLOC must be disabled by the installer so it's always a pilot action to change over.

You can read about here: https://www.twincessna.org/pdf/KFC-225-Internal-GPSS.pdf

In the sim model, the GPS/VLOC swap is causing the KFC225 is going to HDG rather than ROL, but only in certain circumstances, as has been observed in this thread. That's not ideal - however setting the HDG bug to the inbound track during or just before intercept is a really good idea anyway, so by doing this you can mitigate the problem.

I've got a bit more research to do and will update my bug report accordingly, but that will need to wait for a week or so.

Edited May 9, 2020 by Graeme_77

[0zz]

Thanks @Graeme_77 That makes much more sense now. If you have access to Carenado B58, check the KFC on that mod. I think the logic on that one is more accurate.It is a good reference for a realistic rendition.

I noticed, the erratic behaviour when AP is activated during the climb is due to the command I am using. 2x 'servos_up_one_level'. First time it activates AP, and second time it toggles VS. But since the AP button auto-activates VS in BN-2, second command cause the auto-enabled VS mode to deactivate. Then AP auto-switches to ALT mode, that causes the pitch up attitude to zero out. 

 

[Graeme_77]

Thanks Ozz,

I had the Carenado B58 in XP10 with the REP - I'm afraid to say I hated it - the logic was nowhere near realistic. The Islander is at least trying to do something like what the real thing does, the Carenado was behaving like a modern fully digital autopilot. 

The AP engagement logic isn't right at the moment as it remembers the previous VS rather than syncing. If you bind "Autopilot VVI Sync" and use than just before engaging the AP with "Servos toggle" the behaviour is acceptable. This has already been fed back to Torquesim - and all it really needs is a pitch mode engagement, then everything else would be more or less as per the real thing. In the absence of pitch mode "Autopilot VVI Sync" then "Servos toggle" immediately does the job.

Further work on the ROL/HDG mode on approach shows the AP will drop from NAV mode to HDG when there is a VOR/ILS tuned, and from NAV to ROL when there is no VOR or ILS tuned. This is not correct, it should simply go to ROL when swapping between GPS and VLOC CDI.

However, I feel that as the LR 530W does not show the required prompts that Garmin put there specifically for the KFC225 that the most useful behaviour may be to work on the basis the "swap and ROL" shouldn't be there, and allow the AP to maintain NAV mode provided there is a valid VOR/ILS source. This, after all, is what Garmin were trying to achieve anyway, within the limitations of the KFC225.

Alternatively, if Torquesim would implement "external GPSS" where the AP would use the HDG mode (with a switch to swap between HDG bug and GPS) and simulate a fully analog control system then that may be more practical and more realistic given the constraints with the LR 530W.

[Graeme_77]

6 hours ago, 0zz said:

Thanks @Graeme_77 That makes much more sense now. If you have access to Carenado B58, check the KFC on that mod. I think the logic on that one is more accurate.It is a good reference for a realistic rendition.

I noticed, the erratic behaviour when AP is activated during the climb is due to the command I am using. 2x 'servos_up_one_level'. First time it activates AP, and second time it toggles VS. But since the AP button auto-activates VS in BN-2, second command cause the auto-enabled VS mode to deactivate. Then AP auto-switches to ALT mode, that causes the pitch up attitude to zero out. 

 

Ozz, the usual cause of a "snatch" during AP engagement is the X-Plane flight director handing. Do you use flywithlua? If so, perhaps this little script will help you. It creates a keybinding called "Smooth AP engagement on Torquesim Islander". It works best with a long keypress, around 0.5 - 1.0 seconds. On the keydown it activatesthe (hidden) flight directors and resyncs the VS target, then on key release activates the autopilot - you should get a smooth engagement with ROL/VS every time. If it snatches, make sure your are in trim before attempting to engage the AP, and hold this keybinding for just a little bit longer.

function activate_BN2P_autopilot1()
	command_once("sim/autopilot/servos_fdir_off")
	command_once("sim/autopilot/fdir_on")
end

function activate_BN2P_autopilot2()
	command_once("sim/autopilot/vertical_speed_sync")
end

function activate_BN2P_autopilot3()
	command_once("sim/autopilot/servos_on")
end

create_command("FlyWithLua/autopilot/BN2P AP On",
               "Smooth AP engagement on Torquesim Islander",
               "activate_BN2P_autopilot1()",
               "activate_BN2P_autopilot2()",
               "activate_BN2P_autopilot3()"
)

 

Edited May 9, 2020 by Graeme_77

[0zz]

Interesting. I thought REP B58 was quite realistic, as it does not auto-slew HSI. How did Simcoders missed that?

Roger. Will change the bind script to add Autopilot VVI Sync before servo activation. Thanks for the tip.

Systems simulation in BN-2 need more work, more depth in general IMO. Hope AP rework will be one of the early updates.

EDIT: Just saw the script. I use Spad Next to handle bindings. I think I can easily create a scripted event using the simulation events you listed there. Thanks

Edited May 9, 2020 by 0zz

[StefanH75]

7 hours ago, 0zz said:

Interesting. I thought REP B58 was quite realistic, as it does not auto-slew HSI. How did Simcoders missed that?

 

AFAIK Simcoders does not alter the AP logic. They are focussed on engine and flight model.