Is it a P3D thing

[captdini 3]

HI folks,

This is something that has been gnawing at me for sometime, and the reason I bring this up in the General Forum is that I have noticed this in all PMDG a/c (NGX, 777, 747v3). It also occurs in the FSL A320X, and I am sure in other payware aircraft's too. I am referring to the aircraft unable to hold altitude properly in a level turn with the Autopilot engaged. By this, let me explain. Assume I am flying at 1500', speed at maneuvering speed for current flap setting, and A/P at SPD, HDG, ALT HOLD. Now if I turn the MCP Heading selector 90 degrees left, the A/P commands a turn to the selected heading. Now as the bank increases(max 25), after a few moments, the aircraft starts descending and loses anywhere between 30-60 feet. After sometime the A/P recovers the lost altitude slowly and levels off at the MCP altitude. Now, when rolling out of the turn on the selected heading, the aircraft begins to climb, once again by 20-40 feet and the autopilot will correct it back to the MCP altitude. 

I spoke to a friend of mine who is a TRE on NG's and asked him if this behaviour is also the same in the aircraft. He said, he has not ever noticed any such large deviation in the aircraft in the same scenario. The altitude deviation is extremely marginal, if any he said. On one of his last simulator sessions, I asked him to video tape and send me this behaviour from the simulator, which I am attaching. I have also attached a very similar scenario taken from P3D to compare. I hope you can make out what I am saying from observing these 2 videos. You will observe that the Level D sim, the Auto Flight system is much 'quicker' in countering the V/S deviation and the altitude deviation may be hardly 10 feet, if any.

The reason I ask this is because I know how much work PMDG and all the other top notch developer's have put in tweaking the flight model "outside" the stock P3D/FSX engine. Therefore, I was wondering if it is a P3D limitation that the A/P is slow in "reacting and countering" what are the natural laws of physics.

Thanks for your patience. I would love to upload the videos as soon as I know how, so that you can see what I mean.

Regards,

Dinshaw Parakh

https://drive.google.com/file/d/0BxOboTRxOXHtRkZCODJnalZFRHM/view?usp=sharing  - P3D video

https://drive.google.com/file/d/0BxOboTRxOXHtbTdGWnZJSERxbXc/view?usp=sharing   - Level D 737 video

 

 

 

 

[Chock 17,769]

PMDG state that their 747-400 is simulated within 5% of the real aircraft's engine and flight performance, and indeed it is. That might seem like a big margin for error, but in a 'glass half full' kind of way, you should view it not as 5% inaccurate, but rather as 95% absolutely accurate.

And that's not because PMDG couldn't be arsed going that extra five percent, it is because the parameters of how accurate the base simulation is, which in spite of taking some simulation aspects outside of the base simulation, as things like the PMDG 747 and the FSL A320 do, are among the reasons why it could, or could not, be classed as a Level A, B, C or D simulator. Microsoft Flight Simulator, and the derivatives of it, such as FSX-SE and P3D, have never been constrained by any requirement to be accurate enough to be certified for real pilot training. That is not to say you couldn't learn from them, just not in an official capacity. They are essentially for entertainment purposes. That FSX and P3D are like that, has both positive and negative aspects for your home simulation.

To examine what that 5% discrepancy means, let's have a look at some of the requirements for a simulator to be classed as meeting the requirements for pilot training to a professional level...

Level D simulators are required to have a motion system which reacts within 150 milliseconds of what the real aeroplane would do, a Level C simulator can have this happen within 300 milliseconds of what the real aeroplane would do, and the respective visual systems also have to be within those parameters too. Level A and B simulators have no such requirement for that motion to be so represented. Level D simulators have to have a collimated field of view of at least 75 degrees horizontally and 30 degrees vertically, with Level C simulators requiring that to be 45 degrees and 30 degrees. But you might be surprised to learn that a Level A, B or C simulator is not required to model any of the following: ground effect, Mach effect at high altitude, effects of airframe icing, reverse thrust effect on control surfaces, aeroelastic representations, representations of non linear functions due to sideslip etc, etc. Your PMDG 747/737/777 etc, can and does simulate all of those things.

Now, just think about that for a second: Your 20 quid copy of FSX-SE, or your fifty quid copy of P3D, with a seventy quid add-on aeroplane, has the capability to do all that stuff, and display a full 360x360 degree visual simulation of it whilst doing so (assuming you have the monitors or projectors to make that physically possible). You can have fly by views, external camera views, AI aircraft, dynamic weather downloads, AI aircraft flying around, ATC, etc, and all whilst simulating that stuff which even a Level C simulator is not required to be able to replicate. Not one of those things just listed in this paragraph, which your home flight sim does, is a requirement for a multi-million dollar Level A, B, C or D simulator, and the aforementioned list in the previous paragraph is only a requirement for a Level D simulator. So it's a 'swings and roundabouts' kind of deal in that your home sim can do absolutely loads of things which a Level D simulator cannot even come close to doing. The reason a Level D simulator costs millions, is because apart from the actual physical hardware of the cockpit and displays on a motion platform, it has to be able to mimic the real aircraft to a lot closer parameters than the 5 percent margin of error which things like PMDG's 747 work to, largely by virtue of the base simulation program it runs in. But since the PMDG 747 is seventy quid, and a Level D simulator will probably cost several million quid, I think we can forgive it having to use a base platform which constrains it to within a certain level of accuracy, and one is really asking a bit too much of it to be doing the kind of thing - on a base platform which is over ten years old and limited in VAS - that literally costs millions to achieve on a Level D simulator with custom built computing hardware and indeed the same avionics software which the real aeroplane has.

Yes there is some inaccuracy in how FS works, even with the very best add-on aeroplanes, but let's not forget that we are talking about PMDG managing to get their aeroplanes within 5 percent of the real thing's performance on a base platform that you can buy off Steam for 20 quid, and run on a computer that probably most people these days have in their home, whether they are into flight sims or not.

Amazing is an understatement when you think about it like that. Most people who have flown a PMDG version of an airliner a few times could probably get the real thing on the deck safely if they had to. Really, it's nothing short of astonishing.

[downscc 1,327]

Agree with Alan and I'd like to add that the turn is one area that uses more of the FSX/P3D flight simulator than other areas, which are entirely processed outside of the platform.

It should handle a 90 deg turn but when the turn physics became an issue during beta testing the developer did the best he could with the FSX/P3D engine; otherwise, it could have been three more months of development to take the turn outside the box, and they would have felt a business based pressure to increase price.  I will accept what we have, and most do.

Try hand flying any airplane in IMC conditions and make a 90 deg turn without altitude variance..., you pass an FAA checkride if you can hold it within +/- 50 ft., which no pilot with less than 30 hours of instrument time is likely to achieve.

[captdini 3]

Dear Alan and Dan,

Thank you for your detailed replies and I appreciate that. As you both have stated, it is to do with an aging Simulator platform, and it is exactly what I felt too, hence the Topic title. I am definitely not comparing costs and level of immersion, realism, etc between a Level A-D simulator and a desktop platform, it would most definitely be foolhardy and immature to do that. Apples and Oranges.

Like you both said, I am more than happy and content with how far PMDG and the other's have gone in giving us Sim and aviation enthusiasts these extremely realistic planes to play around with, given the limitations of P3D and FSX :) And I am more than happy with 95 % realism as stated. 

It was more a comment on the Devil is in the details kind of thing. The above scenario is in no way a deal breaker, and it would be foolish for me to think so. If they manage to find a work around for this, then fabulous. If not, it will not diminish my enjoyment one bit, either way.

But once again, thanks for your replies and this discussion, which was very enlightening. 

Regards to both of you,

Dinshaw Parakh.

[Paul_Smith 92]

14 hours ago, downscc said:

...it should handle a 90 deg turn but when the turn physics became an issue during beta testing the developer did the best he could with the FSX/P3D engine; otherwise, it could have been three more months of development to take the turn outside the box, and they would have felt a business based pressure to increase price.  I will accept what we have, and most do. ...

I am not sure I can really accept that. This isn't a case of discovering an issue during pre-release testing. This is the PMDG's third go at a 747, second go at a 737 and they have numerous other aircraft under their belt, all of which have PMDG's own autopilot implementations. If they were aware of excess slippage during AP controlled turns, I am pretty confident that they would have fixed it a long time ago. I would ask the OP why they are doing significant AP controlled turns at 1500'? 

[VHOJT 165]

10 hours ago, Paul_Smith said:

 I would ask the OP why they are doing significant AP controlled turns at 1500'? 

Because that's a perfectly reasonable use of an autopilot.  No need to ask that at all.

Cheers,
Rudy

[captdini 3]

17 hours ago, Paul_Smith said:

I am not sure I can really accept that. This isn't a case of discovering an issue during pre-release testing. This is the PMDG's third go at a 747, second go at a 737 and they have numerous other aircraft under their belt, all of which have PMDG's own autopilot implementations. If they were aware of excess slippage during AP controlled turns, I am pretty confident that they would have fixed it a long time ago. I would ask the OP why they are doing significant AP controlled turns at 1500'? 

Dear Paul,

To answer your question, if you see in my Original post, I mentioned that I asked him to specifically get a few turns going in level flight and video tape the A/P behaviour with reference to the altitude loss/gain for me to make a comparison. It was taken after the session was over. Therefore there is nothing strange why they were doing 'significant' turns at 1500'. It may well have been 2500', 5000' or any altitude, which is not relevant here. If you notice the ND/PFD in the video, it is a classic visual circuit pattern they are in. Flap 5, Downwind leg. As you may well know, the 737 visual circuit is normally flown at 1500 feet, you will find this in the QRH. I can safely guess, this was done in the interest of time. It would obviously make no sense wasting time to climb straight ahead to say 5000' and do the same 90 degree left and right banks, which you could do at circuit altitude. 

It was not part of any Simulator profile, if that is what you are thinking. It was purely to demonstrate my request. I hope that answers your question. And as Rudy says, it is perfectly reasonable use of an Autopilot.

Regards,

Dinshaw Parakh

[AirbusCG 52]

I have seen some cases in the TFDi 717 forums where they traced an altitude gain/loss during turns back to Active Sky 2016's winds.
When they flew in clear skies it made perfect turns, with AS16 enabled it started to climb/descent.

The behaviour that the 717 showed looks pretty much the same as the behaviour I see from 'my' PMDG 747.

Maybe OP should try it without a weather engine and see what happens.

[captdini 3]

1 hour ago, Emi said:

I have seen some cases in the TFDi 717 forums where they traced an altitude gain/loss during turns back to Active Sky 2016's winds.
When they flew in clear skies it made perfect turns, with AS16 enabled it started to climb/descent.

The behaviour that the 717 showed looks pretty much the same as the behaviour I see from 'my' PMDG 747.

Maybe OP should try it without a weather engine and see what happens.

Hi Emanuel,

I didn't have any weather engine running, just stock P3D weather. Weather was fair, nothing added. I only added winds of 124/12 just to mimic what I saw in the video sent by my friend. This behaviour happens even with NIL wind conditions IIRC. For whatever it is worth, are you saying in your 747, with clear weather, the turns do not lose/gain altitude? I will give it a shot anyway and report back.

Regards,

Dinshaw Parakh.

P.S. Have any of the PMDG development teams noticed this behaviour in the development cycle or is it just happening to me? Just curious, nothing else

[xkoote 57]

OP,

there are more of these things happening related to the autopilot. I have written it off to just PMDG making a decision to compromise based on any number of factors. In my opinion there is a certain degree of the stock autopilot in PMDG aircraft. The 777 system is A FAR CRY from anything that the FSX/P3D autopilot was ever envisioned to handle. I have learned to accept it and now I find ways around it. Another example is when (at least for me) I do a Go Around, the system has trouble maintaining the G/A speed. After a while speed drops to the amber band and full TOGA is applied. No big whoop. I guess we all wish for it to be perfect. With all other systems simulated to damn near perfection, it is easy to demand the same from other systems.

 

Cheers,

[mtrainer 136]

 Over time, the simulation shows that the physics and system interactions simply gets better and better.  The history of flight simulation is an incredible one.  Looking forward to the future.

Mark Trainer

 

[AirbusCG 52]

On 22.3.2017 at 11:04 AM, captdini said:

Hi Emanuel,

I didn't have any weather engine running, just stock P3D weather. Weather was fair, nothing added. I only added winds of 124/12 just to mimic what I saw in the video sent by my friend. This behaviour happens even with NIL wind conditions IIRC. For whatever it is worth, are you saying in your 747, with clear weather, the turns do not lose/gain altitude? I will give it a shot anyway and report back.

Not saying that at all Dinshaw, I never tried it without. Only in the TFDi 717 I did and over there it helped and worked fine without Active Sky.
That's why I had the idea that this may also happen here in frist place.
I always noted the behaviour, but always put it down as Active Sky induced thing, just like it was in the 717. Perhaps it is not though?

[captdini 3]

40 minutes ago, Emi said:

Not saying that at all Dinshaw, I never tried it without. Only in the TFDi 717 I did and over there it helped and worked fine without Active Sky.
That's why I had the idea that this may also happen here in frist place.
I always noted the behaviour, but always put it down as Active Sky induced thing, just like it was in the 717. Perhaps it is not though?

Hi Emanuel,

Very interesting point you make there. Maybe, TFDi have managed to make a partial workaround and implement it. It is like FSL, who have nailed the taxi/friction problem. The 320 taxis and maintains speed without the need to add thrust every now and then to maintain a taxi speed. At the same time, the FSL A320 exhibits the same loss/gain in height during turns. Having said that, the PMDG birds are no less spectacular in what they do. Especially, the new Queen with the curved CRT screens is something so unique and different, it has to be marveled at. I guess, it is up to each developer to choose which boundaries they want to push. 

With reference to your earlier post, I did do the same with NIL weather(Clear skies) option and had the same results as earlier.

Regards,

Dinshaw Parakh

[AirbusCG 52]

Some developers focus on this and others on that. I think PMDG found a very good balance in the Queen! It's most certainly a stunning airplane!
And yeah, the curved CRTs are a really nice feature! Just like the glass layer above them which make the whole screen look very very good!

For the climb/descent bug it should help to raise a ticket directly with PMDG to be sure they are aware of it and will look into it.