DCS: AH-64D Flight Model discussion

[GrEaSeLiTeNiN]

In spite of all the 'challenges' flying this aircraft, George puts it in a hover really well (when he's not going into VRS ). Hmm...  

[Sandman1330]

46 minutes ago, GrEaSeLiTeNiN said:

In spite of all the 'challenges' flying this aircraft, George puts it in a hover really well (when he's not going into VRS ). Hmm...  

Hovering it is actually easy, no problem there. I did notice the roll axis seems more sensitive after the last update, but adjusting saturation easily solved that.

It’s the unexpected and unpredictable behaviour in dynamic manoeuvring that’s weird. 9 times out of 10 during a specific manoeuvre it will be fine, but that 1 time in 10 it will bite you with a snap roll. I am still trying to determine any kind of pattern or consistency in it.

[Floyd1212]

I posted a short video in this other thread (before it got locked) demonstrating a sure-fire way to induce a snap roll.

https://forum.dcs.world/topic/324117-uncommanded-catastrophic-rolls/?do=findComment&comment=5202767

With the tail rotor applying forces above the CG, giving it a lot of left pedal will create a rolling moment to the right.  Combine this with a left turn, and a quick drop of the collective and you can induce a roll to the right.

Be gentle with the inputs, and you can avoid situations like this.  It's like the Porsche 911 of the 80's.  It would only take you one time to learn not to lift while the tail is getting loose...

[Sandman1330]

29 minutes ago, Floyd1212 said:

I posted a short video in this other thread (before it got locked) demonstrating a sure-fire way to induce a snap roll.

https://forum.dcs.world/topic/324117-uncommanded-catastrophic-rolls/?do=findComment&comment=5202767

With the tail rotor applying forces above the CG, giving it a lot of left pedal will create a rolling moment to the right.  Combine this with a left turn, and a quick drop of the collective and you can induce a roll to the right.

Be gentle with the inputs, and you can avoid situations like this.  It's like the Porsche 911 of the 80's.  It would only take you one time to learn not to lift while the tail is getting loose...

Interesting, though I have a hard time believing the rolling moment from the tail rotor would be forceful enough to generate such a significant and sudden roll. I also fly a helicopter with a high mounted tail rotor, and the rolling moment isn’t even noticeable, let alone enough to cause such an undesirable flight condition.

I wonder if the wonkiness in the ball and the SCAS response to it that I shared above plays into it. Trying to center the ball in response to the incorrect ball indications may lead to more left pedal / tail rotor thrust than desired and therefore more roll moment…

[GrEaSeLiTeNiN]

3 hours ago, Floyd1212 said:

I posted a short video in this other thread (before it got locked) demonstrating a sure-fire way to induce a snap roll.

https://forum.dcs.world/topic/324117-uncommanded-catastrophic-rolls/?do=findComment&comment=5202767

With the tail rotor applying forces above the CG, giving it a lot of left pedal will create a rolling moment to the right.  Combine this with a left turn, and a quick drop of the collective and you can induce a roll to the right.

Be gentle with the inputs, and you can avoid situations like this.  It's like the Porsche 911 of the 80's.  It would only take you one time to learn not to lift while the tail is getting loose...

Thanks for the vid. I tried it out and understand what's going on. It actually made me pay attention to the collective. I find the torque effect when raising and lowering the collective rather pronounced. I mean I could basically yaw L/R with the collective  Maybe it's because I use a Warthog throttle. That said, I'd imagine the real aircraft exhibits the same logic but to perhaps a lesser extent, as I'd assume safety considerations in the design process would have mitigated such an effect.

[Sandman1330]

7 hours ago, GrEaSeLiTeNiN said:

Thanks for the vid. I tried it out and understand what's going on. It actually made me pay attention to the collective. I find the torque effect when raising and lowering the collective rather pronounced. I mean I could basically yaw L/R with the collective  Maybe it's because I use a Warthog throttle. That said, I'd imagine the real aircraft exhibits the same logic but to perhaps a lesser extent, as I'd assume safety considerations in the design process would have mitigated such an effect.

No that’s legit, the aircraft should have significant yaw with collective movements. In fact it’s a bit under pronounced in the DCS Apache, because of the SCAS yaw channel helping you out (assuming it’s working correctly).

One of the emergencies we practice regularly is landing with a stuck pedal. If I stick the pedal on a student at 90% torque, where we require 70% to hover IGE, that 20% torque difference means the aircraft will have significant rotation on landing.

Edited April 23, 2023 by Sandman1330

[Poptart]

On 4/22/2023 at 8:00 PM, Sandman1330 said:

Interesting, though I have a hard time believing the rolling moment from the tail rotor would be forceful enough to generate such a significant and sudden roll. I also fly a helicopter with a high mounted tail rotor, and the rolling moment isn’t even noticeable, let alone enough to cause such an undesirable flight condition.

I wonder if the wonkiness in the ball and the SCAS response to it that I shared above plays into it. Trying to center the ball in response to the incorrect ball indications may lead to more left pedal / tail rotor thrust than desired and therefore more roll moment…

Yep. I notice that when I try to keep the aircraft in trim, that huge roll force is much more pronounced and it takes an unusual amount of left stick to counter. It's not as bad in nose to tail trim. I've started flying that way since the update.

[Apache 64]

I almost hesitate to write about it, but in my opinion the Apache flies more predictably again since today's fix (DCS 2.8.4.39731 Open Beta).

Although of course nothing has changed again, I think both the center of gravity has shifted for the better again, and the hold modes work more reliably again, as the roll tendency is less pronounced again.

Just my observations/impressions.
Makes spontaneously more fun again.

PS: If I'm the only one who feels this way, just check it off as confusion

 

[BIGNEWY]

6 minutes ago, Apache 64 said:

I almost hesitate to write about it, but in my opinion the Apache flies more predictably again since today's fix (DCS 2.8.4.39731 Open Beta).

Although of course nothing has changed again, I think both the center of gravity has shifted for the better again, and the hold modes work more reliably again, as the roll tendency is less pronounced again.

Just my observations/impressions.
Makes spontaneously more fun again.

PS: If I'm the only one who feels this way, just check it off as confusion

 

No changes were made in the last patch, if you are seeing a difference it could be you. 

[Silver_Dragon]

I recomend the Casmo video about the AH-64D...

 

[Apache 64]

vor 50 Minuten schrieb BIGNEWY:

No changes were made in the last patch, if you are seeing a difference it could be you. 

Muscle memory is a bitch

[placsea]

1 hour ago, Apache 64 said:

Muscle memory is a bitch

I think your hitting on something that people maybe need to be more aware of. If you don't fly a helicopter in DCS for a while, apache or whatever you are going to have to reacclimate. It's not a personal failing to fiddle with the control axis to get them usable and comfortable. The more time you spend flying and practicing you can dial things into your liking. We don't fly helicopters in DCS because they are easy. 

[Apache 64]

vor 9 Stunden schrieb placsea:

I think your hitting on something that people maybe need to be more aware of. If you don't fly a helicopter in DCS for a while, apache or whatever you are going to have to reacclimate. It's not a personal failing to fiddle with the control axis to get them usable and comfortable. The more time you spend flying and practicing you can dial things into your liking. We don't fly helicopters in DCS because they are easy. 

I totally agree with you there. But I actually fly the Apache several hours a day. Hence the idea that muscle memory plays a big role here. With each hour of practice, the feeling for flying after a patch gets better and you adjust accordingly. Then, when a new patch comes along and it suddenly goes really well in the same period, you might think something has changed because of an update.

So all is well. Practice, practice, practice, as has been said so many times.

[Raven434th]

This in OUR experience is NOT the case,...we can fly everyday for a week and struggle to stay alive without the machine killing us every other flight...then a patch comes out and its fine...then we maybe fly another week and a patch comes out and its a nightmare...then its patched and its fine again...then another patch and its squirrelly  as F...we DON"T fly it for a week or two due to frustration....it gets patched.... and we fly it like we're seasoned apache pilots it and totally OWN the beast. And the cycle continues....it's been a rollercoaster ride. As of yesterday...it's been user friendly and good to fly. Handles like I'm in control not it...........so far..........

Edited May 5, 2023 by Raven434th

[Sandman1330]

On 4/22/2023 at 1:53 PM, Sandman1330 said:

I did some quick testing, and found something interesting.

I set the apache up at about 85% torque and 100kts. This resulted in a slight climb, but I wanted to ensure airspeed and torque remain constant in the turn so as not to change coordinated flight parameters. With the ball centered, I then started a turn as tight as I could without losing airspeed or having to increase torque. So, with the collective and pedals frozen, I turned the aircraft and maintained (within 5kts / 5%) my parameters.

In both turns, the ball immediately falls to the inside of the turn, as if it's following gravity, not the G loading of the aircraft as it should. This is where it got interesting:

In a left turn, the SCAS yaw channel slowly motored over to re-center the ball in the turn. On roll-out, the ball deflected heavily to the right, until the SCAS motored back to my trimmed center, where the ball re-centered. So in a left turn, it seems the SCAS wants to correct for the incorrect ball behaviour, resulting in a significantly uncoordinated turn.

In a right turn, none of that behaviour was exhibited - the SCAS stayed centered in the trimmed position, and on roll out the ball centered again on it's own.

So in addition to the ball behaviour, there seems to be some inconsistency in how the yaw channel deals with it.

Add to this, if I try and center the ball myself in the turn (as a good helicopter pilot should always do), I'm actually putting the aircraft in a significantly uncoordinated situation. I wonder if this uncoordinated flight is causing some of the wonkyness / uncommanded roll that we are seeing...

Here's a track file of my testing:

 

 

apache ball.trkUnavailable

@BIGNEWY or @Raptor9, this post above, should I create a dedicated bug report? Or is this acknowledged already? I'm happy to create a dedicated bug thread if this hasn't been noted yet.

My bug report would be less about the behaviour of the SAS, and more about the behaviour of the ball indicator (falling to the inside of the turn with gravity rather than indicating a coordinated turn)

Edited May 5, 2023 by Sandman1330

[Raptor9]

43 minutes ago, Sandman1330 said:

should I create a dedicated bug report? Or is this acknowledged already? I'm happy to create a dedicated bug thread if this hasn't been noted yet.

My bug report would be less about the behaviour of the SAS, and more about the behaviour of the ball indicator (falling to the inside of the turn with gravity rather than indicating a coordinated turn)

Short answer, they are known and are being worked on.

Long answer: it's not so much the trim ball as much as the flight model itself that needs adjustment, which is still ongoing. Specifically, the yaw offset to the left in forward flight and the adverse yaw when rolling the aircraft require adjustment. Those are what are contributing to the behavior you are seeing that is creating the uncoordinated flight when rolling into and out of a turn without collective and pedal movements, and are known. Also keep in mind the electronic trim ball indication in the AH-64D is derived from inertial data, which makes it behave more reliably and promptly, but it won't behave the exact same way as the skid/slip indicator in an analog gauge, so I caution against making those direct comparisons.

The other side of it is Turn Coordination. When the force trim isn't pressed and ground speed is >40 knots, SAS Turn Coordination logic will attempt to keep the trim ball at its current reference (which is updated when the force trim is released). However, its current correction rate is much too slow which is why you see it slowly fade in the input after the fact, unless the roll is accomplished very slowly. The Turn Coordination input rate is also a known factor that needs adjustment, coupled with the factors of the exaggerated yaw offset and adverse yaw effect.

[Sandman1330]

45 minutes ago, Raptor9 said:

Short answer, they are known and are being worked on.

Long answer: it's not so much the trim ball as much as the flight model itself that needs adjustment, which is still ongoing. Specifically, the yaw offset to the left in forward flight and the adverse yaw when rolling the aircraft require adjustment. Those are what are contributing to the behavior you are seeing that is creating the uncoordinated flight when rolling into and out of a turn without collective and pedal movements, and are known. Also keep in mind the electronic trim ball indication in the AH-64D is derived from inertial data, which makes it behave more reliably and promptly, but it won't behave the exact same way as the skid/slip indicator in an analog gauge, so I caution against making those direct comparisons.

The other side of it is Turn Coordination. When the force trim isn't pressed and ground speed is >40 knots, SAS Turn Coordination logic will attempt to keep the trim ball at its current reference (which is updated when the force trim is released). However, its current correction rate is much too slow which is why you see it slowly fade in the input after the fact, unless the roll is accomplished very slowly. The Turn Coordination input rate is also a known factor that needs adjustment, coupled with the factors of the exaggerated yaw offset and adverse yaw effect.

Awesome, thanks for your response. Since the issues are known I'll not clutter up the bug threads and let you guys work.

Cheers!

[pii]

On 5/5/2023 at 9:51 AM, Raven434th said:

it's been user friendly and good to fly. Handles like I'm in control not it...........so far..........

I sense a little sarcasm but in general, I agree with it all except the last part above.

It's still wonky for me.  waiting again  

[Durham]

Recently got a Rhino FFB joystick primarily for force trim on AH-64D.  Got that working and it really is transformative having the cyclic behave as the manufacturer intended.  However, have experienced the uncontrollable roll (usually to the right) which sometimes ends up in inverted flight.  Nonetheless, spent two hours last night trying and failing to replicate, irrespective of how irresponsible I was with the cyclic roll.  Reasonably convinced by the discussion above that it is more to do with the controller(s) fighting the SAS system, so now have that up and am monitoring.  Will try again tonight messing around with load out, density altitude to see if I can predictably induce the uncontrolled behavior and will update here if I succeed.  What does seem obvious is that if you fly carefully, blip the trim repeatedly, she flies in a very believable and predictable manner, but sometimes, if you fly like an idiot, it will bite you and if you are low to the ground you crash and burn.  Yet to have it affect a squadron multi-player mission, only when messing around flight testing in single player have I lost control.  Ho hum.

[pii]

On 5/8/2023 at 1:35 PM, pii said:

I sense a little sarcasm but in general, I agree with it all except the last part above.

It's still wonky for me.  waiting again  

UPDATE:

Well, I have to take what I said above back.

Apparently, I missed an update and once I updated the AH-64 flew a lot better in most flight-related areas.

If DCS didn't do anything to the FM I'd say something else is affecting the FM. It flies completely difference now. 

[iFoxRomeo]

DCS 2.8.5.40170 Open Beta

 

...

• FM. Considerable change of flight dynamics model (WIP):
  • The range of roll and pitch control has been increased, and at the same time, the stability of the helicopter has been increased in all flight modes, including:
    • Helicopter controllability at low-speed maneuvering has been improved;
    • Increased helicopter roll stability at submaximal speeds;
    • Improved behavior in transitional modes (roll-in/out and pitch-in/out)
  • The trim position of the helicopter and controls in hover and level flight modes has slightly changed.
• FMC/SCAS
  • SCAS tuned for modified flight dynamics model;
  • SCAS altitude channel damping and auto-shutdown behaviour was remodelled to react at specific limit parameters of the engine and vertical speed

...

[hreich]

Anybody know what this option in special options is:  -- man-trk ramp up speed?

[hreich]

I just tried to engage hover hold --below 5 kts, and my helo managed to hover...But on IHADS display it stated that speed is around 10 knts?! Is that normal behaviour?

 

 

[Raptor9]

9 minutes ago, hreich said:

I just tried to engage hover hold --below 5 kts, and my helo managed to hover...But on IHADS display it stated that speed is around 10 knts?! Is that normal behaviour?

Position Hold (and Velocity Hold) are based on ground speed, not airspeed.

[Apache 64]

DCS 2.8.5.40170 Open Beta:

Thank you ED for the latest changes on the FM and the tuning on the systems arround.

This is great pleasure to flight the Apache now. Even much more than before.

And i love it in every aspect.

Now we can have slow turns without or with minimum pedals, assisted from SCAS.

Hold modes work also better than before.

The whole flight less wonky, less rolling more precise... a big step in the right direction.

Many, many thanks and keep up the great work.

...back to cockpit now....more flying.