Attitude indicator will not right itself

[pierrewind]

On some modules, once the artificial horizon developps an error it is unable to correct it. 
This has been observed on:
-The F5E
-P51
-FW190
-(problem likely on other airplanes as well)

Picture taken on T/O, 30 minutes later the problem was still there (or maybe increased). There wasn't any significant maneuvering.
Doing 90degree turns showed the gyroscope reference as being wrong (opposite bank at 180 degres, pitch up/down at the 90 degrees). Reference wasn't regained in flight. Once on the ground it seemed to be able to correct this error within 10 minutes with an expected 2 degrees per minute correction (spooled up).



The fact that the gyroscope is unable to regain the reference in flight is a big problem. It for example will make uncaging the attitude indicator in flight quite impractical or render the attitude indicator information unusable after it developps errors. 

Ps: Similar to:

 

Edited April 21, 2021 by pierrewind

[pierrewind]

Here is the track applicable to the P51

Gyro Bug 2.trk

[FoxxyTrotty]

Can I just confirm for the sake of avoiding confusion, that you are not referring to gyroscopic drift here? DCS does simulate the gyro drift, and you must cage/uncage during level flight periodically to level it out. It does seem a little excessive to me in some modules (i know what you mean about easy flight/gentle turns still throws it off) but this is drift, and not some sort of bug that prevents it from being caged. I'll give those aircraft you mention a go and see if I am prevented from caging/uncaging to level out the drift.

[FoxxyTrotty]

ive just flew the P51 and mig15 and was able to cage/uncage like before to level out the drift

[Frederf]

Typically attitude gyros of this period are self-correcting and will continuously calibrate to local vertical over time. Dramatic attitudes like aerobatics can cause AIs to tumble or otherwise not maintain calibration but with mild maneuvering this isn't an issue. Gyros of this type will have a cage which will prevent attitude changes beyond certain limits. Usually the AI will recover itself but if it doesn't (or doesn't quickly enough) the manual cage is there. Low vacuum produces poor performance like on the ground.

 

It's an AN-5736-1 (Sperry 656768 being an example)and more information about its design and performance specification would be helpful to find what its performance is like.

Edited April 20, 2021 by Frederf

[pierrewind]

Thank you for the answer.

Attitude indicators have been designed with system that keep it level. It usually is accomplished through small exhaust ports within the gyroscope. Then with precession it will bring the gyroscope level. This does introduce small errors with turning and accelerations. Other more advanced gyroscopes may use other systems. 

It also allows the gyroscope to maintain a constant reference in regard to the earth "plate" . At the moment, it seems the attitude indicators are not able to regain this reference. 

Not being able to regain/maintain the level reference if a delibarate choice is a wrong choice. (note: on the ground, it is able to regain reference)
 

The earth rotation has influence but this instrument is designed to account for it  

In flight, you should never have to cage/uncage unless you run into some specific situations such as hitting the attitude indicator limits. 

Edited April 20, 2021 by pierrewind

[-0303-]

Two good videos.

Training video describing Artificial Horizons from the oldest pneumatic to newer electrically driven. Their basic functionalities and errors. The pendulous vanes of pneumatic horizons described (applicable to every WW2 Warbird).

Pedagogic explanation of what precession is.

EDIT

University of Nottingham guy describes precession hands on. http://www.sixtysymbols.com/videos/axial_precession.htm

On second thought, this video instead, watch 8:20 - 16:55. It's a pedagogically much better explanation of precession. Yes, he talks about gyro-compass, but precession is precession, everything applies to gyro-horizon as well.

 

Edited January 15 by -0303-
remove blank lines added by ???

[-0303-]

I made tracks of P-51D, Bf 109 K-4 and F-86F. Simplest shortest track:

• Air start
• cage horizon
• roll 90 degrees
• uncage horizon
• roll level and fly for 10 minutes watching for self erection.

None did. In 10 minutes a minimum of 2 degrees per minute should make for 20 degrees or something noticeable. I believe I understand that a fundamental functionality of any artificial horizon, equally important as the rigid-in-space property, is that they self erect towards the gravity vector (~2-8 degrees per minute). Note, only talking of ye olde mechanic gyro horizons.

I did find specs for the F-86F "J-8" (electric) artificial horizon. Gleaning the essential: self erection rate is 2-4 degrees per minute in bank, 2-5 degrees in pitch (If "0.2" isn't a typo I only watched roll self erection anyway).

This page states Attitude Indicator Type J-8, MIL-I-5133A is for F-86-F. Found specs for MIL-I-5133C here (pdf):

Bf 109_horizon_v1_10min.trk F-86F_horizon_v1_10min.trk P-51_horizon_v1_10min.trk

Edited October 17, 2021 by -0303-

[Machalot]

You may need to limit your test to a 5-10 deg offset, not 90 deg, based on what it says in the F-86 spec you quoted. At 90 deg the self erecting feature might be outside its design limits. 

[-0303-]

On 10/16/2021 at 8:35 PM, Machalot said:

You may need to limit your test to a 5-10 deg offset, not 90 deg, based on what it says in the F-86 spec you quoted. At 90 deg the self erecting feature might be outside its design limits. 

I saw that. I thought it wouldn't make sense for it to not self erect from any angle. I wanted unambiguous results.

Still, I made tracks starting from 10 degrees and 30 degrees. Tricky to set just 5 degrees. Given the "5 degrees in 6 min" they should self erect in 12 minutes and 36 minutes respective (ed/add: should happen faster see next post)  which are the times I tested (+ ~1 extra minute to make sure they got 12 and 36). They got worse. 10 degree right bias ended at 20 degree right bias, 30 degree right bias ended at 45 degree right bias.

Also made 3 hour tests with P-51D and F-86F. Parked with enough rpm for suction and electricity (generator lamp off F-86F). Both crept much to slow (P-51 ~0.5 degree per minute) and worse, kept on trucking after passing "level". P-51 started at 60 degree right bias, I quit at 30 left bias. F-86F started at 90 degree right bias, I quit again at 30 left bias. Sparing people those tracks.

In [F2] view, note clock rightmost and bank angle (check for zero and look at horizon). Pause may be useful. 38:18 min and 14:19 min track times respective. Obviously use speed up, no point watching real time.

F-86F_horizon_v3_start30end45deg_in36min.trk F-86F_horizon_v4_start10end20deg_in12min.trk

Edited October 21, 2021 by -0303-

[-0303-]

Just realized, last sentence of F-86F gyro spec: "...horizon bar [pitch] and bank index [roll] ... return from 5 degrees to zero ... 6 minutes..." makes for a pitiful (5/6) 0.83 degrees per minute. But the pendulous vanes action, by their nature, obviously acts more powerfully the further from alignment the horizon is. Watch pierrewind 2 min video and think about it. If correcting air ports are only partially open at 5 degree misalignment, they're fully open at some point further away (10, 30 , 45 misalignment?) exerting max realignment push (making for the full advertised 5 degree per minute?).

Then my last two tracks shouldn't require 12 min and 36 min but something much shorter. Also now I think maybe that's why the initial F-86F gyro spec paragraph talks about a specific interval  "...10 to 5 degrees shall be between 2 to 4 degrees per minute...". Realignment speeds differ with the degree of misalignment.

Edited October 17, 2021 by -0303-