6 hours ago, LRBS said:

There are a couple of issues here happening, unfortunately, if you point out some discrepancies about the product in their forum, many users, some beta testers including some in the developers' team will feel the need to dismiss any findings or play the blame game or just don't know how the real airplane should behave. Many times if you push the issue your complaints/findings will get deleted. That's your answer not seeing "comments".

 

As an airline pilot (dual qualified in Boeing and Airbus) we are talking about two different airframes with totally different systems and behaviors. What you think about Fenix of being "unrealistic agile" it is not correct, this is how the airplane reacts. Airbus systems are softer and more defined in certain aspects compared to Boeing. 

 

You are right on the money when you mentioned that speed is a factor, especially in this case "LOC intercept".

 

"Usually" when on final vectors to intercept, we are at about 180 KTS, and sometimes during capture, you will notice a slight deviation and bank correction left/right before settling down (not on Airbus). Unfortunately, PMDG has issues with not properly anticipating entering/exiting the turn translating into overshooting or undershooting and high unrealistic bank rates.

 

Regarding that PMDG implemented a brand new LNAV system is totally incorrect, they still have issues with it and is not fixed after so many years.

PMDG like Aerosoft, doesn't like criticism, period. 

I would like to share this information. Perhaps will bring some joy.

 

I tried these settings and I noticed some improvements, but it is needed to edit the ai.cfg located in their specific aircraft folder.

Change these two lines thanks to Simon A. Bergmann to read:

 

 

headingPID = 1.3, 0.01, 2.1, 0.2, 1
nav_ex1PID = 2.5, 0.05, 20, 0.02, 50

 

The problem with this is that the paying customer should not do their work.

Edited December 30, 20223 yr by LRBS

In the real world simulator in the A320 a good rule of thumb when self positioning onto a LOC (no ATC radar vectors) with a 90deg heading intercept, is:

Start the turn when : Distance offset from LOC (shown at bottom of ND if you have extended the RWY centerline and/or waypoint sequencing is correct) is 10% of your Groundspeed.

eg if your Groundspeed is 180 knots , start the 90deg final turn when 1.8nm away from the LOC. 

Some people (Top Guns) use 80% of this eg. 1.5nm . but i need to check it in the real sim. 

 

16 minutes ago, Winkle Brown said:

In the real world simulator in the A320 a good rule of thumb when self positioning onto a LOC (no ATC radar vectors) with a 90deg heading intercept, is:

Start the turn when : Distance offset from LOC (shown at bottom of ND if you have extended the RWY centerline and/or waypoint sequencing is correct) is 10% of your Groundspeed.

eg if your Groundspeed is 180 knots , start the 90deg final turn when 1.8nm away from the LOC. 

Some people (Top Guns) use 80% of this eg. 1.5nm . but i need to check it in the real sim. 

 

That's exactly right, 25 bank and it will line up perfectly, at 45 intercepts with this GS roughly 0.9 NM lead turn. The same method works in Boeings, actually in any airplane. This method was explained to them but completely rejected. And it shows.

I'm very happy that you mentioned this, people need to see this stuff, and there is so much more. THANK YOU!

On 12/30/2022 at 3:41 PM, LRBS said:

Regarding that PMDG implemented a brand new LNAV system is totally incorrect, they still have issues with it and is not fixed after so many years.

It is correct, there was an LNAV update a while ago and while it's probably still not entirely accurate, it's doing a job miles better than before, both with regards to LNAV as well as LOC captures.

There seems to be confusion here about how automation handles a turn onto final.

Firstly, yes PMDG did recently introduce a new and significantly improved LNAV model.  Gotta keep up.  ;)

Secondly, LNAV has exactly nothing to do with how a 737 (and most other aircraft) handle a localizer intercept.  A localizer is a radio beam.  When you're on a heading and you arm Approach or VOR/LOC, you're telling the aircraft to fly straight until it senses the course to that radio beam come alive.  It cannot lead the turn, because it has no way of knowing when to start turning until the localizer is alive and the computer can measure the rate of needle movement.

This is different than flying an LNAV course, which will result in leading turns for a smooth intercept.  It's possible in this case because the FMC is driving and has a known lateral path to follow; it knows in advance where the turns are going to be.  In Approach or VOR/LOC, the FMC is not controlling lateral path.  Thus, even if you've loaded the approach in the box (which you typically would do), the automation has no way to lead the turn.

The reality is that a real 737 will often make a 2 step LOC capture: an initial full bank towards final as the LOC comes alive, and then a shallow out after a second or two of needle movement rate has been sensed, then a steepening bank again to establish on LOC as it centers up.  Sometimes you'll even see an S turn, depending on wind, groundspeed, and intercept angle.  This is the reason many of us will choose to use LNAV to join a localizer if we're on a long final and down low; it'll make the intercept much smoother because the FMC can predict the turn in advance.  That, or I'll often just hand fly it since I can eyeball the ND and predict my turn that way.  Simmers are in general more married to the automation than real pilots; it's perfectly ok to click the magic off and just fly.  ;)

This behavior is typical of any aircraft that purely uses localizer sensing to join.  Every airliner I've ever flown behaves this way.  Now, there are aircraft that will use FMC path as a sort of pseudo-LNAV to predict turns even when not in LNAV or "managed" mode; maybe the Airbus does that, I don't know.  

But TL;DR - there's nothing wrong with the way the PMDG captures a localizer, from what I've seen it's realistic.  Their LNAV model, while significantly improved recently, does need further improvement.

And don't join a localizer at more than a 30 degree intercept in a non-LNAV mode.  That's entirely unrealistic and not setting you up for any kind of success.  😉

13 hours ago, Stearmandriver said:

Secondly, LNAV has exactly nothing to do with how a 737 (and most other aircraft) handle a localizer intercept.

I threw LNAV and localizer capture in one sentence because since the LNAV update it's capturing the localizer a lot better, so something must have been polished on that front as well and IIRC PMDG said they would bring improvements here in conjunction with the LNAV update.

15 hours ago, Stearmandriver said:

There seems to be confusion here about how automation handles a turn onto final.

Firstly, yes PMDG did recently introduce a new and significantly improved LNAV model.  Gotta keep up.  😉

Secondly, LNAV has exactly nothing to do with how a 737 (and most other aircraft) handle a localizer intercept.  A localizer is a radio beam.  When you're on a heading and you arm Approach or VOR/LOC, you're telling the aircraft to fly straight until it senses the course to that radio beam come alive.  It cannot lead the turn, because it has no way of knowing when to start turning until the localizer is alive and the computer can measure the rate of needle movement.

This is different than flying an LNAV course, which will result in leading turns for a smooth intercept.  It's possible in this case because the FMC is driving and has a known lateral path to follow; it knows in advance where the turns are going to be.  In Approach or VOR/LOC, the FMC is not controlling lateral path.  Thus, even if you've loaded the approach in the box (which you typically would do), the automation has no way to lead the turn.

The reality is that a real 737 will often make a 2 step LOC capture: an initial full bank towards final as the LOC comes alive, and then a shallow out after a second or two of needle movement rate has been sensed, then a steepening bank again to establish on LOC as it centers up.  Sometimes you'll even see an S turn, depending on wind, groundspeed, and intercept angle.  This is the reason many of us will choose to use LNAV to join a localizer if we're on a long final and down low; it'll make the intercept much smoother because the FMC can predict the turn in advance.  That, or I'll often just hand fly it since I can eyeball the ND and predict my turn that way.  Simmers are in general more married to the automation than real pilots; it's perfectly ok to click the magic off and just fly.  😉

This behavior is typical of any aircraft that purely uses localizer sensing to join.  Every airliner I've ever flown behaves this way.  Now, there are aircraft that will use FMC path as a sort of pseudo-LNAV to predict turns even when not in LNAV or "managed" mode; maybe the Airbus does that, I don't know.  

But TL;DR - there's nothing wrong with the way the PMDG captures a localizer, from what I've seen it's realistic.  Their LNAV model, while significantly improved recently, does need further improvement.

And don't join a localizer at more than a 30 degree intercept in a non-LNAV mode.  That's entirely unrealistic and not setting you up for any kind of success.  😉

I totally agree with your assessment, there is no question about it. Until the recent PMDG update, there were issues with the LOC capture regarding intercept angle and closure rate. What surprises me is the "steepening bank angle again to establish on LOC as it centers up", perhaps is a different version of the software? 

Our fleet is quite the opposite, much more shallow and smooth.

 

 From what I see now, it looks like they fine-tuned the roll behavior related to capturing with a positive outcome.

 

Regarding Airbus vs Boeing, there is a total difference in the way systems work and technologies, and not enough space and time to try explaining.  

 

Regarding the LNAV, I guess they are good at playing with the words, and as usual, people fall for this.

 

Again, it's not a "brand new LNAV system implemented" as somebody is mentioning, it is a "tweak" if I may say, where they modified the roll channel of the AP and FD (nothing else), and again the same vague wording makes people believe that we have a brand new LNAV system. Oh yes, still needs lots of work.

 

The brand new one, as they keep on talking about at their forum it might be called LNAV V2.0. 

Edited January 1, 20233 yr by LRBS

5 minutes ago, LRBS said:

Again, it's not a "brand new LNAV system implemented" as somebody is mentioning, it is a "tweak" if I may say, where they modified the roll channel of the AP and FD (nothing else), and again the same vague wording makes people believe that we have a brand new LNAV system. Oh yes, still needs lots of work.

And we know this how? Is the source code available somewhere that we can review?

27 minutes ago, jrw4 said:

And we know this how? Is the source code available somewhere that we can review?

PMDG 737-600 for Microsoft Flight Simulator:
===================================
First up, lets talk about the Babiest of Baby Boeings: The PMDG 737-600!

I mentioned last month that we had decided to use the 600 testing period as an iterative test program to bring in our new Lateral Flight Path model. This decision was made knowing that it would cause a slightly longer development period than the originally planned six weeks for the PMDG 737-600, but it has proven to be a good use of development and testing time. We are just about ready to release the 737-600 and it will include the new Lateral Flight Path model, along a significantly updated vertical path model that is finally able to leverage the improved accuracy of the lateral model into it's own computational process. 

https://forum.pmdg.com/forum/main-forum/general-discussion-news-and-announcements/205179-23jul22-pmdg-737-for-microsoft-flight-simulator-updates-and-general-news-for-your-weekend

Edited January 1, 20233 yr by LRBS

Thank you. I'm particularly interested in how "the new Lateral Flight Path model" and its integration into the vertical path model can be seen as a tweak. How do we know that it isn't a truly new model as the words actually say? Doesn't sound like a tweak to me, but I freely admit I may be wrong. How do we know?

20 minutes ago, jrw4 said:

Thank you. I'm particularly interested in how "the new Lateral Flight Path model" and its integration into the vertical path model can be seen as a tweak. How do we know that it isn't a truly new model as the words actually say? Doesn't sound like a tweak to me, but I freely admit I may be wrong. How do we know?

You are welcome.

I'm afraid that we can't cover here all the issues related to proper LNAV functionality, there are many references of how should work and what kind of problems they have. There are quite a few posts regarding what's wrong and what is expected from real pilots on their forum. If you are not familiar with the correct functionality it might be difficult to see certain discrepancies.

There seem to be honest differences in how individuals familiar with the real 737 perceive the fidelity of the PMDG LNAV system, but that's not what I'm asking about. Nor am I asking what the remaining discrepancies might be. Specifically I asked how do we know that the recently released LNAV update constitutes a "tweak". Barring access to the code, I can't see how one could make that assertion, but am totally open to learning.

1 hour ago, jrw4 said:

There seem to be honest differences in how individuals familiar with the real 737 perceive the fidelity of the PMDG LNAV system, but that's not what I'm asking about. Nor am I asking what the remaining discrepancies might be. Specifically I asked how do we know that the recently released LNAV update constitutes a "tweak". Barring access to the code, I can't see how one could make that assertion, but am totally open to learning.

I can see that the word "tweak" is bothering you.

 

All I can say is that I don't see anywhere mentioning the long waited LNAV 2 to be available with the correct functionality. 

 

Thou, it can be noticed a slight improvement, still a long way from what it should be.

 

My "assertion" of being a tweak is based on day-to-day line operations using the LNAV and how it works.

 

When they gonna completely overhaul this system to correct functionality I will reconsider and for sure I'll be happy for their progress. 

Edited January 1, 20233 yr by LRBS

1 hour ago, jrw4 said:

There seem to be honest differences in how individuals familiar with the real 737 perceive the fidelity of the PMDG LNAV system, but that's not what I'm asking about.

After reading the comments from the 2 pilots, and this statement, it made me wonder if the company procedures being used affects the transition to the localizer?

An interesting experiment would be to have each pilot fly (and record) the exact same approach following their respective company procedures, and then compare the resulting flight profiles to see what differences we might see as a result. 

Edited January 1, 20233 yr by MDFlier