reported Artificial Horizon not self correcting. Artificial Horizon NOT messed up on rolling (unlike the real thing)

[-0303-]

The Artificial Horizon is very often badly out of whack (henceforth "wrong"), as many have noted. A little surprised that despite really trying (loop, roll ...) I have yet to find a quick way to get it "wrong". I'd love to make a short track but I couldn't.

 

Interestingly, one real Mk IX (see film), messes up it's horizon anytime it rolls and then takes 9 minutes to self correct. Two things. DCS Spitfire horizon does not get "wrong" on rolling and it never self corrects (at least not in 80 minutes).

 

Anyway, after 50 minutes, I manage to make it wrong by doing "donuts" and  then finally I could do the test I set out to do:

1) Parked on runway, RPM 2200 (to provide suction) for 40 minutes+, horizon does not not self correct.

2) Fly plain and level for 40 minutes+, horizon is not self correcting.

Note1, the 50 minutes Batumi - Soochi were flown straight and level to see if it would get wrong spontaneously. It didn't)

Note2, I found the behavior after it's "wrong" weird. If I turn, in the air or on the rwy, I can get the horizon "right" again)

 

Real life video references.

1) Hadley, Spitfire MK IX.

https://youtu.be/Njjl_smtMoQ

Within the first 40 sec, one can visually see the horizon creep in self correcting while on the ground.

 

2) Spitfire Mk IX crossing the channel.

https://youtu.be/O-_AwgYxUW0

At 17:01 it rolls. Horizon gets instantly wrong. Watching from 18:06 (film cut) horizon is 90 degrees wrong. After 7 min (25:00) it's corrected itself in roll axis and after 2 more minutes it corrected itself in pitch axis too (total 9 min to completely self correct). The real Mk IX, in the film, rolls twice (at least), at 17:01 and at 27:15 (31:40, 32:40, 37:50, 38:47, 39:08, 39:40). Both (all) times the horizon gets instantly screwed up. Seems any roll always messes up the horizon in the real thing. In DCS it does not.

 

Anyone so inclined can fast forward my track to the end (~10:24 [F2] time) to quickly verify non straight horizon is not just me.By [F2] time, I mean the lower right corner clock.

~

 

Track timestamps ([F2] view) to fast forward to:

8:00:00 Takeoff Batumi.

8:49:30 Full stop Soochi, horizon still good. Repair (nothing) Refuel.

8:52:00 Start donuts. Horizon gets "wrong".

8:59:15  - 9:42:15 Parked runway, horizon "wrong", RPM 2200 for 40+ min. Observe horizon not correcting.

9:43:40 - 10:25:12 Takeoff, course 270 (gyro) fly plain & level 40+ min. Horizon not correcting.

 

Installation completely kosher, no mods no nothing.

DCS 2.7.6.12852 Open Beta (recorded track)

DCS 2.7.6.13133 Open Beta (replay)

Intel Core i7 3630QM @ 2.40GHz (Max Turbo 3.40 GHz)

16.0GB Dual-Channel DDR3 @ 798MHz

2047MB NVIDIA GeForce GT 635M

447GB KINGSTON SA400S37480G (SATA-2 (SSD))

 

(~ 9:43:40) Soochi takeoff. Horizon "wrong"

 

(10:23:40) After 40 min+ flight plain and level . Horizon still "wrong".

spitfire_horizon_v2.trk

Edited September 28, 2021 by -0303-

[Mr_sukebe]

I'm sure that it used to self correct, prior to the hotfix on 22nd Sept.  

[-0303-]

I've had the Spitfire since ~2019 and fly it more than anything else. Though I never "properly" tested it, I can't remember it ever working correctly.

Edited September 23, 2021 by -0303-

[Nealius]

It worked for me the last ~40 hours I put on the Spitfire since just before 2.7 released. I flew it right before the Mossie release and the horizon would self-correct in short order, and rarely wen't wrong at all. I haven't flown the Spit since the Mossie came out, so it must be an issue that came with the last two patches.

[NineLine]

Thanks guys, I am on it.

[NineLine]

So I played the included track, and for me, it was fine. So not sure what to say there.

 

I cant imagine what would cause it to be different between two of us. Can you do a repair and try again, just for the heck of it?

[-0303-]

^

By this I assume your artificial horizon was "horizontal" & centered at end of track? That's decisively different from what I'm seeing (diagonal). So, something is unambiguously different, something that can be identified.

 

What could be different?
Historically, one bug did not reproduce for ED when [Combined Arms], which I don't have, was included. The modules I have installed are (note, no Marianas):

[I-16]
[Normandy 1944 Map]
[WWII Assets Pack]
[Bf 109 K-4 Kurfurst]
[UH-1H Huey]
[F-86F Sabre]
[Flaming Cliffs 3]
[Spitfire LF Mk. IX]
[P-51D Mustang]

 

Repaired, reran track again

Ran repair off the Win 10 start menu (as opposed command line) and checked these:
Check all files (slow) ✓
Search for extra files after repair ✓

Renamed SAVED GAMES to create a "fresh" one.

 

Reran the track (real time). Same result.

Zipped logs just after program exit.

 

I'm still interested if it happens to more people than me? Could more people please run the track?

Real time full run 2h20min with 5 min to spare before mission end. Divide by whatever time acceleration used. Can set Windows alarm if run in background.

• Note 1. I found I must hit [F1] immediately after start of replay, to not lose function of some vision keys (to zoom in on horizon while replaying). Not terribly important, would start other thread if I wanted to discuss that.
• Note 2. Yes, the track ends out west to sea. The weirdness of horizon includes it (may) straighten out if turning south back towards Batumi again (it did earlier track version).

 

Logs.zip

Edited September 25, 2021 by -0303-

[TJpek]

Can confirm the bug report: Flew for a total of 2h yesterday, the artificial horizon didn't mess up on various rolls while flying around for 1h, but after some dogfights it ended being biased to the left, we flew another 30 minutes after that and it never self corrected.
For as long as I can remember it has been this way: the artificial horizon ends up messing itself up and there's no manual way to reset it in the spitfire

[Nealius]

It's intermittent for me. On first flight the ADI porked itself on takeoff and never corrected. Second flight the ADI was reliable throughout the flight.

 

Maybe it's one of those bugs that happens the first flight after clearing fxo/metashaders2 folders?

Edited September 30, 2021 by Nealius

[NineLine]

On 9/25/2021 at 4:14 AM, -0303- said:

(it did earlier track version).

In this one, after you did all the crazy moves on the ground I did see the gauge out of whack and never recentering.

It does say this on the side of them:

 

So I am not sure it isn't possible to mess up the real thing if you try hard enough.

[-0303-]

^

Quote

crazy moves

 

No crazy moves required (*1). Just spawn ... and wait. I discovered where it happens and can easily and quickly reproduce it. Made a new short track ... and a movie.

 

I found the horizon needles drifts visibly randomly whenever the engine shuts down or (less so) with the engine just idling. I want to interpret this as modeling insufficient suction causing drift by vibration (*2). That would be great, but the problem is that the horizon never self correct once the engine have enough RPM.

 

It "always" drifts on engine shutdown. In submitted track it also drifts instantly on spawn (idling runway). Just realized, it was my 50x speed up that suddenly made "drift on idle" visible, whereas I otherwise usually don't see it. Again, I believe the drift-on-insufficient-suction is just fine, the bug is the horizon NOT self correcting on higher RPM (2650 cruise track).

 

The track
Spawned, noticed immediate drift, revved up a short while just to see if it would start self correction. No. Cycled engine off / on six times. Spawning would have been enough but I wanted the needle at 60 degrees so it would be unambiguous. Recorded at 50x. Took off, flew straight and level (1x real time) aiming for gyro course 320 (~Batumi runway extension out to sea). Still at 60 degrees 10 minutes on. Quit.

10 minutes, again, because video of a real Spitfire Mk IX takes 9 minutes to fully self correct. Neither I or the real Mk IX fly perfectly level, obviously an insane demand to correct the horizon.

 

The film
I filmed 50x speed on runway and then filmed 10x speed flying. 8min:07sec. You can obviously skip and Youtube can be sped up 2x.

 

Also, again, I note that the real Spitfire has it's horizon go coco for cocopuffs anytime it rolls, it would be nice to see that in DCS. The DCS horizon is aligned after 10 rolls if it was aligned before the roll (just tested). If it was unaligned it goes back to the same unalignment (kind of, something more is happening).

 

Attached new logs and new track.

Again, more people could try this shorter track? Use speed up, track stops at 8:49:37, so look after 49 (sped up) minutes if horizon leveled or as below (~video thumbnail).

 

 

*1) Note for NineLine. This did happen:

        Historically, one bug did not reproduce for ED when [Combined Arms], which I don't have, was included

I reported bug, ED  couldn't reproduce until having the same set of modules installed as I had.

 

*2) The needles drift randomly and independently of each other. Random in speed, direction and length (they even switch direction while  drifting). Sometimes one needle drifts much more than the other. All this is good to my mind, they should be random. Just noting, don't expect needles to drift just like in this video / track in other tracks, they won't. But they do drift just the same replaying track so ... canned random?

 

Logs.zip spitfire_horizon_v3.trk

Edited October 1, 2021 by -0303-

[-0303-]

On 9/30/2021 at 3:15 PM, TJpek said:

Can confirm the bug report: Flew for a total of 2h yesterday, the artificial horizon didn't mess up on various rolls while flying around for 1h, but after some dogfights it ended being biased to the left, we flew another 30 minutes after that and it never self corrected.
For as long as I can remember it has been this way: the artificial horizon ends up messing itself up and there's no manual way to reset it in the spitfire

Thanks. Every word is perfectly consistent with my ~2.5 year experience.

I have [WWII Assets Pack], I don't have [Combined Arms]. See my list above. I wonder if modules can make a difference? It did with another bug.

 

The rolling. Tímestamps of every roll in the channel video.

17:01, 27:15, 31:40, 32:40, 37:50, 38:47, 39:08, 39:40. Yeah, I'm sad.

 

Edited October 1, 2021 by -0303-

[NineLine]

Can you tell me which modules you guys have and I can try and check that angle? Thanks.

[NineLine]

6 hours ago, -0303- said:

^

 

No crazy moves required (*1). Just spawn ... and wait. I discovered where it happens and can easily and quickly reproduce it. Made a new short track ... and a movie.

 

I found the horizon needles drifts visibly randomly whenever the engine shuts down or (less so) with the engine just idling. I want to interpret this as modeling insufficient suction causing drift by vibration (*2). That would be great, but the problem is that the horizon never self correct once the engine have enough RPM.

 

It "always" drifts on engine shutdown. In submitted track it also drifts instantly on spawn (idling runway). Just realized, it was my 50x speed up that suddenly made "drift on idle" visible, whereas I otherwise usually don't see it. Again, I believe the drift-on-insufficient-suction is just fine, the bug is the horizon NOT self correcting on higher RPM (2650 cruise track).

 

The track
Spawned, noticed immediate drift, revved up a short while just to see if it would start self correction. No. Cycled engine off / on six times. Spawning would have been enough but I wanted the needle at 60 degrees so it would be unambiguous. Recorded at 50x. Took off, flew straight and level (1x real time) aiming for gyro course 320 (~Batumi runway extension out to sea). Still at 60 degrees 10 minutes on. Quit.

10 minutes, again, because video of a real Spitfire Mk IX takes 9 minutes to fully self correct. Neither I or the real Mk IX fly perfectly level, obviously an insane demand to correct the horizon.

 

The film
I filmed 50x speed on runway and then filmed 10x speed flying. 8min:07sec. You can obviously skip and Youtube can be sped up 2x.

 

Also, again, I note that the real Spitfire has it's horizon go coco for cocopuffs anytime it rolls, it would be nice to see that in DCS. The DCS horizon is aligned after 10 rolls if it was aligned before the roll (just tested). If it was unaligned it goes back to the same unalignment (kind of, something more is happening).

 

Attached new logs and new track.

Again, more people could try this shorter track? Use speed up, track stops at 8:49:37, so look after 49 (sped up) minutes if horizon leveled or as below (~video thumbnail).

 

 

*1) Note for NineLine. This did happen:

        Historically, one bug did not reproduce for ED when [Combined Arms], which I don't have, was included

I reported bug, ED  couldn't reproduce until having the same set of modules installed as I had.

 

*2) The needles drift randomly and independently of each other. Random in speed, direction and length (they even switch direction while  drifting). Sometimes one needle drifts much more than the other. All this is good to my mind, they should be random. Just noting, don't expect needles to drift just like in this video / track in other tracks, they won't. But they do drift just the same replaying track so ... canned random?

 

Logs.zip 9.96 kB · 0 downloads spitfire_horizon_v3.trk 183.48 kB · 1 download

 

This track was messed up for me this time, I can report with this track.