Jump to content
Sign in to follow this  
Guest dwhitley

PMDG Autopilot

Recommended Posts

Guest dwhitley

I have posted this topic in another area of the AVSIM Forum, but thought the subject might draw a better answer here in the Specific PMDG forum.Apparently, PMDG knows something about Autopilots that the other AC designers do not know.The issue is the autopilot effectiveness. Specifically Roll in and out of headings, nav capture, and nav tracking. In my case, the PMDG 737 series behaves just like the real aircraft autopilot. It will promptly roll to and from a "Bugged" heading without lag, and capture the localizer (or any nav course) precisely at just about any intercept angle.The approach to the correct design on the PMDG is apparently in the heading control. For example, compare a PMDG to just about any other FS aircraft. The PMDG will perform a precise 180 degree turn in 2 minutes. It does this by promptly rolling to the turn, and promptly rolling out of the turn. Other aircraft will slowly roll into the turn, continue the turn at a 3 degree per second rate, and then slowly roll out of the turn, thus taking about 2 minutes and 30 seconds to complete. This slow roll in and roll out process is what causes a lag in nav intercept, and thus causes hunting down the pathway (localizer, nav, etc).The PMDG design apparently has accomplished the correct autopilot operation in areas other than the AC config file. I have compared the PMDG config file with other AC config files, and the flight tuning and autopilot code is virtually the same. Many replies in my other posts have suggested changing various numbers in the aileron_effectiveness, nav_proportional_control, and nav_integrator_control. These changes are basically cheating the system. It does create a little better capture rate, but affects the way the aircraft handles under manual pilot control.If anyone wants to better understand the problem, fly a test with an FS factory aircraft and then a PMDG aircraft. Simply launch and level off somewhere at 3000 feet. Engage the AP on a heading, and then command a few turns. The problem will immediately become evident. Notice specifically how quickly the PMDG will roll into the turn, and how quickly it will roll out of the turn. Then, notice with other FS AC, how slowly the AC rolls in, and very slowly it rolls out. Timing the turns makes the problem even more evident. Time a 180 or 360 degree turn. Notice that the PMDG will make a 180 deg turn in 1 minute, and a 360 deg turn in 2 minutes. Pretty much standard for a real AC. Then, notice how long it takes the average FS AC to make the same turns. You will notice the longer time evolving from the very slow roll outs. So, it is very evident that PMDG AC are designed differently in how the AP works. As mentioned earlier, I do not see any differences in the config files, so they are accomplishing the task through other methods.It appears that MSFS is not the problem (as many have claimed) if one vendor can accomplish the correct procedure.That said, PMDG may have proprietary coding in the autopilot area. However, if not, they are accomplishing the tasks in other areas (other than the AC CFG file) of their package. So, if anyone knows where the fix may lie, or how PMDG accomplishes the correct autopilot operation, it would create better experience for those of us who fly in the real world.This may all seem a bit picky, but I use MSFS to assist my real world flight students in instrument flight instruction. The error within the autopilot is very annoying and causes some confusion. PMDG had raised the standard of MSFS by being more of a real simulation experience vs a game. Thus, if anyone knows of a resolution to the above issue, then we are accomplishing a better environment for the RW pilots.Thanks Dennis

Share this post


Link to post
Guest Steven Stratton

Dennis, I think it it fair to say that there is very little in common code-wise with the default autopilot system and the PMDG autoflight system, and I am thankful for that! :(

Share this post


Link to post

Geeze, really? Using MSFS for instrument instruction?? I think you should look at ASA On-Top for this purpose (see: http://www.asa2fly.com/On-Top-95--P1012_product1.aspx ). I have a copy of On Top (several versions and years old) that impressed me with its flight "mechanics." This is a strictly no-eye-candy instrument trainer that if you had the necessary accessories becomes an FAA approved PC-based trainer. It costs more than MSFS but not by as much as the difference in performance.


Dan Downs KCRP

Share this post


Link to post

Hi Dennis,This was explained to me once by the PMDG programmers during the beta testing of the 747. Much went over my head, but from what I can gather there is basically a "virtual autopilot/FCC" which sits running outside the FS programme itself and then injects control commands into the simulator model. It was explained that the software worked on the same principles and methods as the real world FCCs. I was amazed at the shear amount of work and detail thats gone into the product to make it work realistically,yet most of the time the things are so subtle they wont even get noticed.Thats why you wont spot anything in the config files.CheersJon B


787 captain.  

Previously 24 years on 747-400.Technical advisor on PMDG 747 legacy versions QOTS 1 , FS9 and Aerowinx PS1. 

Share this post


Link to post

Yeah that's the basic idea - we don't use FS's default autopilot coding at all. (most developers use what's called a "wrapper" around the default FS autopilot, they can do modes like FL CH etc that way, but it's not mathematically correct, they're essentially continuously varying the VS mode of the default AP to produce something that looks like FL CH but isn't etc) Our AFDS code runs outside of FS's functions, and is actually simulating the real thing very closely - this is why for instance you see the yokes in our VC actually moving - that's because our code is commanding the control surfaces, not commanding the default FS autopilot (one for sure way to check for wrapper code is to see if everything's being done using trim commands - many addons use trim to control pitch when on AP, that's an FS default AP thing and it's not realistic)A couple other planes that use realistic custom APs outside of FS like this are the LDS 763, the Project Tupolev Tu-154, and the upcoming Airsimmer A320...


Ryan Maziarz
devteam.jpg

For fastest support, please submit a ticket at http://support.precisionmanuals.com

Share this post


Link to post
Guest dwhitley

Thanks for the info. I figured as much about the PMDG AP being a lot more sophisticated. The AP does a fantastic job. You can tell that a lot of hard work and many many hours went into the project. Congrats to the team that designed it !!!A lot of folks in other forums came back with some tweaks for the cfg file, but it only "cheated" a little on the flight control surfaces to make the flight dymanics a little more snappy:) However, the cfg changes only screw up the way the manual flight controls behave.Coupling up on a nav signal, or GPS requires some heavy duty math. Especially when you start tracking procedure turns, holding patterns, climb and descent profiles etc. I never expected MSFS to achieve such tasks. But, I would have expected the basic heading control to behave a little more rigidly. However, in the outcome, it's basic heading that controls the aircraft's path anyway.The SIM aricraft that I mostly fly is Eaglesoft's CJ-1. They do pretty good work, and I hear that their Cirrus is a great AC. However, I don't know if any of their other projects AP's fly better or not. But it appears the ES CJ uses the MS AP, and behaves as most other typical FS aricraft do. I have writtin in their forum, but have yet to get a reply back.I also tossed this disscussion out into Peter Dowson's Forum. I thought maybe something in the FSUIPC may be causing the AP issue. Control Spiking etc. I also thought "if" there was a resoultion / fix for the AP, that maybe the FSUIPC could incorporate such a fix to correct the AP's roll rates etc. But typically, in the tone that Peter usually replies with, all I got back was how I should go out and build my own. Enough said :)Anyway, thanks again for the input.PS, I am very familar with ASA's products. There are much better, and, much worse programs out there. Nothing can substitute for the real thing, but I'm working on a safety program to address the expected drop in flying experience due to the current econmy. MSFS is on our list of simple and very econimical "stay sharp" ideas.Nite all ,D

Share this post


Link to post
this is why for instance you see the yokes in our VC actually moving - that's because our code is commanding the control surfaces
Is it possible to smooth out the "jerkiness" of the yoke (and throttle) animations? Maybe in the upcoming SP?

Matthew S

Share this post


Link to post
The approach to the correct design on the PMDG is apparently in the heading control.
To me the most obvious telltale sign of MSFS autopilot is how it (mis)handles glide slope and strong cross winds on ILS approach.

Share this post


Link to post

"PS, I am very familar with ASA's products. There are much better, and, much worse programs out there"I would be very interested in a much better sim product than the ASA On-Top. Can you recommend any?


Dan Downs KCRP

Share this post


Link to post
Is it possible to smooth out the "jerkiness" of the yoke (and throttle) animations? Maybe in the upcoming SP?
A lot of times this is actually caused by jitter in your joystick... Aside from that I believe it's just a limitation of what FS can display when animations are tied to an external process like that.

Ryan Maziarz
devteam.jpg

For fastest support, please submit a ticket at http://support.precisionmanuals.com

Share this post


Link to post

Dennis,As the author of the AP system thanks ! And I in turn have to thank Rob Randazzo for the patience and time it takes for each plane to complete this project to a good standard (most vendors would have gone: finish it now in whatever state and lets sell "something" to the unsuspecting public). The system is indeed completely independent from any FS logic (unlike most other add ons which feature a "wrapper" around FS control logic). It is fascinating to "engineer" and implement real world industrial control logic to a sub-1000 dollar product. A three-axis FCC moves your controls as a "virtual pilot" based on computed flight director commands (this in turn has its own completely independent logic). I believe it is impossible to drive accurately a high performance jet with default FS stuff. For example the pitch channel is FS driven AP's only uses elevator trim for pitch adjustments and has only one set of control constants for all modes. Can you imagine flying an aircraft using only trim ?Having said that the NG has the v1.0 AP. I look first at the 744 and then at the MD11 and see how much development has come through the last 5-6 years. Accordingly there is still scope for improvement.Vangelis


====================================

E M V

Precision Manuals Development Group

====================================

Share this post


Link to post
Guest dwhitley

I wish I understood how the process was accomplished in the sim. In a RW AC, (a simple autopilot), there is a blind turn indicator and a coil surrounding the heading bug. Simply put, if the bug is not on the desied heading, there is a flow of current throught the coil, and voltage is applied to the roll servo, thus making the AC Turn. The rate of roll is limited by the physical design of the turn rate gyro. Once the bug starts to center, the current through the coil starts to null, thus stopping the roll. If the heading varies either side of the bug, the approproaite voltage is supplied to the roll servo to keep the heading as that the bug has commanded.As for the nav stuff, there are still only two wires, ((a left and right)) wire that feeds the physical autopilot servo from "whatever box" such as GPS FMS etc. There is still a turn rate gyro to keep everything "governed" at a standard 3 degree per second roll rate, otherwise the AC would roll itself over trying to manage a turn. The pitch control works much in the same manner, but usually governed by an altimeter or VS device.The problem with the MSFS SIM I think lies in the AP portion of AC control, not the GPS or NAV sections. But then agian, I don't know how all of this works in the SIM. But as as for GPS and FMS control, I think the majic box does all of the math and then feeds the AP.Anyway, what I think you are saying is, you have re-built a new AP from scratch. So there are actually two processes going on. The AP,,, and the Nav, which are two different components. Take the PMDG and do a 180 or 360 turn with the heading bug. The PMDG AC will quickly roll to it's bank angle, and manitain the 3 deg per sec roll rate, then, promptly roll out about 5 degrees before the bug. This is why the PMDG AC wil capture a localizer of Glide slope precisely from jsut abut any intercept.Now, try it on the average MS AC, and you will notice how slowly the AC turns into the roll, and how very slowly the AC rolls out of the turn. This is what is causing the slow capture and tracking of an ILS. It's not a function of the nav equipment, it the function of the actual autopliot that causes the problem. Another test is to time a 180 or 360 deg turn. The PMDG will do them pretty much standard rate of 1 min and 2 min respectively. THE MSFS AP will take about 1:20 and 2:30 respectively.Anyway, I know it's a pretty complicated process designing an AP in the SIM. I was hoping someone like FSUIPC would have found a crutch / fix for the sluggish AP problem in MSFS. But then again, who but a few, probably notice an issue other than it won't capture and track and ILS :) I believe the fix is more than just a tweak or cheat on the flight control surfaces. I wish I had the knowhow to tackle the proble, but I don't have the time nor the knowledge. I just drive them, not build them. :(Thanks again,D

Share this post


Link to post
A lot of times this is actually caused by jitter in your joystick... Aside from that I believe it's just a limitation of what FS can display when animations are tied to an external process like that.
I'll do some more testing later tonight, but I'm sure it happens with both my joystick and CH yoke. It definitely doesn't happen with the LDS 767 which very smoothly animates the throttle and yoke (i think) movement that are being commanded by the AT/AP.

Matthew S

Share this post


Link to post
As the author of the AP system thanks ! And I in turn have to thank Rob Randazzo for the patience and time it takes for each plane to complete this project to a good standard (most vendors would have gone: finish it now in whatever state and lets sell "something" to the unsuspecting public).
Your work is excellent, thank you!One little thing that bugs me about the MD11X is that the AP seems to be a little aggressive when initiating turns or pitch changes. It's as if the planes inertia is not taken into account (by your custom AP) when initiating these changes. This is one area where the default FSX AP seems to be smoother. Even the custom LDS767 AP is much smoother in this respect.Also I if set bank limit to 5 degrees and then command a turn 90 degrees to left (or right) the AP will bank past the 5 deg bank limit before settling back down to 5 deg. Additionally, I find the throttle/yoke movements that are being commanded by the AP are a little "jerky" in their animation. It's as if you need to "sample" the commanded position more often.These are just little things I know, but they do detract a little from the feeling of immersion.Thank you again! The MD11X is an amazing achievement.

Matthew S

Share this post


Link to post
Guest dwhitley

Just for grins, does anyone know if my original topic had the same issue in FS X ?D

Share this post


Link to post

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  
  • Tom Allensworth,
    Founder of AVSIM Online


  • Flight Simulation's Premier Resource!

    AVSIM is a free service to the flight simulation community. AVSIM is staffed completely by volunteers and all funds donated to AVSIM go directly back to supporting the community. Your donation here helps to pay our bandwidth costs, emergency funding, and other general costs that crop up from time to time. Thank you for your support!

    Click here for more information and to see all donations year to date.
×
×
  • Create New...