BUG: ACS (AP) oscillation at high Mach numbers.

[DarkFire]

TEST CONDITIONS:

Standard DCS day.

No wind.

No clouds.

No precipitation.

 

FLIGHT CONDITIONS:

Flying at 12,000m

Accelerating using maximum afterburner.

Altitude held using ACS "H" mode.

 

OBSERVED PROBLEM:

With "H" mode engaged, when speed reaches between Mach 2.25 and 2.30 an abrupt pitch oscillation sets in, initially as high as ±25 m/s vertical velocity. As speed increases to M2.55 the oscillation gradually damps down to ~±6 m/s vertical velocity.

 

Switching off the altitude hold mode stops the oscillation but on re-engaging the autopilot the oscillation reappears.

 

Attaches is a track and my DCS log file.

 

A small request: please, please, I beg, for the love of ($belief system) can the trim reset be made to work on the roll channel? :cry_2:

Edited November 27, 2016 by DarkFire

[SinusoidDelta]

The manual states the maximum speed is M2.35. I'm still surprised how greatly that number can be exceeded. At those speeds I don't think there should be any guarantee for care free handling.

 

Also from the manual:

With a further increase of the indicated airspeed, the lateral handling gradually deteriorates, which becomes noticeable at airspeeds greater than 1200 km/h.

[DarkFire]

The manual states the maximum speed is M2.35. I'm still surprised how greatly that number can be exceeded. At those speeds I don't think there should be any guarantee for care free handling.

 

Also from the manual:

 

True, but as far as I know the "Speed above Mach 2, not more than 5 minutes" comment from the manual is due to friction heating of the canopy and elements of the airframe. Aerodynamically it's quite capable of flying at faster speeds.

 

I could understand if, maybe due to limits on computing power available to the ACS, the system begins to lose stability above certain speed as the system struggles to cope with increasingly extreme data inputs. I could also imagine that the ACS system increasingly fails to compute proper outputs given 'edge of envelope' inputs.

 

All that being said, the sudden onset of the pitch oscillation at quite a precise speed and it's harmonic oscillation characteristics, despite the fact that the ACS is supposed to be a deterministic system, is more suggestive of an anomaly of the in-game ACS programming as opposed to an accurate simulation of the real system.

 

To be a bit more concise: I can imagine that the real Su-27 ACS system suffers from decreasing stability as flight approaches the edges of the envelope due to the limitations of the system, but what we have hear I think is probably an anomaly of how the game programming works rather than a simulation of real limitations.

[Ironhand]

Ran your track and took control several times to test a hypothesis. If you disengage the AP prior to that problem range and retrim that "anomaly" disappears. I took over at just under Mach 2.0 several times and down to as low as Mach 1.5. (I stopped there due to lack of time.) At no point did the anomaly reoccur. So how much the AP has to deal with a mis-trimmed aircraft seems to be the factor here. It can only compensate a finite amount before it's overwhelmed.

 

EDIT: Don't know if this is actually a bug or not. But that seems to be what's occurring. Don't know what the system's real world limitations are.

Edited November 30, 2016 by Ironhand

[DarkFire]

Ran your track and took control several times to test a hypothesis. If you disengage the AP prior to that problem range and retrim that "anomaly" disappears. I took over at just under Mach 2.0 several times and down to as low as Mach 1.5. (I stopped there due to lack of time.) At no point did the anomaly reoccur. So how much the AP has to deal with a mis-trimmed aircraft seems to be the factor here. It can only compensate a finite amount before it's overwhelmed.

 

EDIT: Don't know if this is actually a bug or not. But that seems to be what's occurring. Don't know what the system's real world limitations are.

 

Indeed it does. Re-tried it a couple of times and found the same thing. So what we have is a 'trim out of range' anomaly rather than an absolute trim issue. This is pure speculation on my part, but I wonder if this sort of anomaly and a few other minor quirks were produced as a result of adjustments made to the modelling of the ACS when it was improved in the patch a few weeks ago?

 

Have to admit, given a choice between the ACS working perfectly at speeds down to ~320 Km/h, with a better AOA / G limiter and all the rest of the improvements we now have; and a funky autopilot at Mach 2.5, I'd take the other improvements any day. Realistically, in pretty much any combat mission in the Su-27, how often do we really need to cruise around at M2.5 at 40K feet? :)

[Ironhand]

Indeed it does. Re-tried it a couple of times and found the same thing. So what we have is a 'trim out of range' anomaly rather than an absolute trim issue. This is pure speculation on my part, but I wonder if this sort of anomaly and a few other minor quirks were produced as a result of adjustments made to the modelling of the ACS when it was improved in the patch a few weeks ago?

 

Have to admit, given a choice between the ACS working perfectly at speeds down to ~320 Km/h, with a better AOA / G limiter and all the rest of the improvements we now have; and a funky autopilot at Mach 2.5, I'd take the other improvements any day. Realistically, in pretty much any combat mission in the Su-27, how often do we really need to cruise around at M2.5 at 40K feet? :)

... And the "fix" for the problem is simple enough...retrim once you approach those higher airspeeds. It's one of those things I would never have noticed on my own simply because I'd be trimmed for close to my airspeed, anyway, before I'd engage the AP if I were to use it. Different user approach, I guess. :)

[Pocket Sized]

They should probably do what was done on the F-15: introduce an artifical speed limit at max allowable Mach. The 15 has been demonstrated ~0.1M over the DCS speed limit IRL and has similar issues with heating.

[DarkFire]

... And the "fix" for the problem is simple enough...retrim once you approach those higher airspeeds. It's one of those things I would never have noticed on my own simply because I'd be trimmed for close to my airspeed, anyway, before I'd engage the AP if I were to use it. Different user approach, I guess. :)

 

Yep, the only time I ever sit there at 12,000m accelerating to M2.5 is for a max altitude run. Otherwise it's fairly pointless.

 

They should probably do what was done on the F-15: introduce an artifical speed limit at max allowable Mach. The 15 has been demonstrated ~0.1M over the DCS speed limit IRL and has similar issues with heating.

 

Not a bad idea. If the speed were to be limited to M2.35 (or is it 2.32?) I don't think anyone would really notice. Plus as you said anything except an SR-71 or a MiG-25/31 travelling at that sort of speed is going to start melting anyway.

[SinusoidDelta]

They should probably do what was done on the F-15: introduce an artifical speed limit at max allowable Mach. The 15 has been demonstrated ~0.1M over the DCS speed limit IRL and has similar issues with heating.

 

I agree. Currently the Su-27 far exceeds its quoted maximum speed of 2.3M. At It's probably not high on the priority list considering all that ED is working on at the moment. It also doesn't seem to provide any real advantage AFAIK.

[GGTharos]

The F-15 has demonstrated speeds up to M2.7 IRL, officially, and higher un-officially. It didn't melt down.

 

Similarly the Su-27 is capable of higher speeds than posted, and it probably won't melt down either.

 

There should be no artificial speed limit. Your fuel will naturally limit the amount of time you spend at those speeds to something below the time required to destroy your aircraft.

 

Any damage-inducing limitations should be based on IAS/CAS, not on mach number.

Edited December 1, 2016 by GGTharos

[Pocket Sized]

There should be no artificial speed limit.

 

Tell that to BST :P

[GGTharos]

There is an artificial speed limit of M2.6 for both F-15 and Su-27 last I checked, but it was a while ago.

 

Also, made a mistake - the damage should be based on EAS, not CAS, IMHO

[DarkFire]

There is an artificial speed limit of M2.6 for both F-15 and Su-27 last I checked, but it was a while ago.

 

Also, made a mistake - the damage should be based on EAS, not CAS, IMHO

 

Yes, I gather that the F-15C (used to?) reach 2.606M and the Su-27 would amusingly go all the way up to 2.62M, though it now tops out at about 2.54M.

[Pocket Sized]

I think the speed limit is based more on ground/true airspeed.

 

I've reached it at a variety of altitudes and it always seems to be bang on 1500 knots true, despite mach number changing with altitude. I'll have to do more testing though.

[GGTharos]

Why? GS/TAS has no bearing on what the aircraft experiences in terms of force. It's EAS describes the airspeed upon which the stresses are based. The engine itself might have a different dependency but I have my doubts.

 

I think the speed limit is based more on ground/true airspeed.

 

I've reached it at a variety of altitudes and it always seems to be bang on 1500 knots true, despite mach number changing with altitude. I'll have to do more testing though.

[bolek]

So what we have is a 'trim out of range' anomaly rather than an absolute trim issue. This is pure speculation on my part, but I wonder if this sort of anomaly and a few other minor quirks were produced as a result of adjustments made to the modelling of the ACS when it was improved in the patch a few weeks ago?

 

I don't think so, this kind of behavior has always been there since PFM. Autopilot was pretty broken before this ACS patch, but even then it behaved much better if the aircraft was trimmed for the speed (and had to be re-trimmed if major changes occurred while autopilot was engaged).

[Pocket Sized]

Why? GS/TAS has no bearing on what the aircraft experiences in terms of force. It's EAS describes the airspeed upon which the stresses are based. The engine itself might have a different dependency but I have my doubts.

 

The speed limit I'm describing is artificially added by the FM, so I see no reason it couldn't be based on TAS. Plus, 1500 is a pretty round number but it very well could be based on Mach.

[DarkFire]

I don't think so, this kind of behavior has always been there since PFM. Autopilot was pretty broken before this ACS patch, but even then it behaved much better if the aircraft was trimmed for the speed (and had to be re-trimmed if major changes occurred while autopilot was engaged).

 

Well, yes and no. Before the PFM update it was perfectly possible to engage the autopilot at ~1040 Km/h TAS and it would be rock steady all the way from there up to 2770 Km/h TAS which is the maximum possible speed I've ever seen. It never generated such a huge pitch oscillation around 2.3M.

 

What you may be thinking of was the issue whereby if the AP was engaged beneath 560 Km/h IAS it would generate a yaw oscillation.

[GGTharos]

I understand where you're coming from, and I would agree that in absence of anything else, a TAS limit of 1500 is ok. I would however prefer it to be based on EAS, as this is what damage is actually based on.

 

If you check fighter manuals, their speed limitations mention airspeeds in KTS, not mach numbers. Consider that the F-15 for example has a prescribed never exceed speed of 800KCAS (IIRC) - 800KCAS at 1000' is a vastly different TAS from 40000', as you know.

 

In this case, KCAS is what you get from your instruments, but IMHO it should be based on EAS which would explain those official high speeds better (EAS is more generous with altitude increase).

 

 

The speed limit I'm describing is artificially added by the FM, so I see no reason it couldn't be based on TAS. Plus, 1500 is a pretty round number but it very well could be based on Mach.

[DarkFire]

Ran your track and took control several times to test a hypothesis. If you disengage the AP prior to that problem range and retrim that "anomaly" disappears. I took over at just under Mach 2.0 several times and down to as low as Mach 1.5. (I stopped there due to lack of time.) At no point did the anomaly reoccur. So how much the AP has to deal with a mis-trimmed aircraft seems to be the factor here. It can only compensate a finite amount before it's overwhelmed.

 

EDIT: Don't know if this is actually a bug or not. But that seems to be what's occurring. Don't know what the system's real world limitations are.

 

Yep, I've tested this again and you're exactly right: anything over about a 1.2M difference between the speed at which the AP is engaged and the final speed produces the pitch oscillations, so I'd say that the maximum trim difference that the AP can cope with corresponds to about a 1.2M speed difference.

[Ironhand]

Yep, I've tested this again and you're exactly right: anything over about a 1.2M difference between the speed at which the AP is engaged and the final speed produces the pitch oscillations, so I'd say that the maximum trim difference that the AP can cope with corresponds to about a 1.2M speed difference.

This might only factor in at high speeds. Starting slow and increasing to over a 1.2M speed difference didn't seem to make any difference to the AP function. At least that was the case the few times I tried.

[LJQCN101]

Well, in the RL manual, Operational Limits part, it reads: ACS operation is prohibited at M > 2.0. Reason: Mild oscillation occur.

Edited December 7, 2016 by LJQCN101

[DarkFire]

Well, in the RL manual, Operational Limits part, it reads: ACS operation is prohibited at M > 2.0. Reason: Mild oscillation occur.

 

Ah, well there we go, thanks for posting that. Wish there was an English translation of the actual Su-27 flight manual available...