3 hours ago, fsiscool said:

@Alec The big question is the expectation. If the expectation is wrong, you can also bin the conclusion. So if you say "unexpected results", what is your expecation based on?

Shouldn't you first figure out, how the pitch response of a 172 looks like? Until now you only argued with terms like "twitchyness", "rubber band feeling", which are completely nonscientific buzzwords. 

In correct terms, your video is showing the short period mode pitch response, as visualized here:

Until now, you have not prooved at all whether the new or the old sim is better in simulating the default 172. Because all is based on an expectation which is only an idea in your mind.

But how, if we search for real world data?

The best diagram, I can find for the curve that describes the correct expectation is this:

 

(from here Airplane Stability and Control (Flying Qualities) | Academic Flight)

In the diagram, the red curve shows hardly any overshooting after the initial large peak. After just 0.5 seconds, the angle only changes in the sub-0.03..0.05° range anymore.

And now, based on this data, you can review again the two aircraft and conclude which one is better.

My conclusion: the new one is better.

This is also no surprise, because Seb explained, that they have instrumented real aircraft with dataloggers to base their flight modeling in MSFS 2024 on hard, real world data.

So your expectation was not solid enough to proove errors in the MSFS 2024 flight model, but it was actually the MSFS 2024 flight model, that highlighted errors in your expectation.

It is not out of complete luck that I did the exact same methodology for testing short period response than your articles linked do. These "nonscientific buzzwords" you mention is explained in one of my first posts, I dont like to use scientific name dropping to make points that are gonna leave many in the dark. when those same words can be easily translated to terms we can all relate to, pilots, aero engineers or just simmers.

Now, you have very useful valuable data in there, about the C172. Thanks for bringing that into this discussion of ours. 

I have taken the time to go 0.25x timerate and logging the times of hitting the High and Low points of each Cycle Amplitudea after each Elevator Impulse. This took some time, but with your data I felt it was worth doing.

Here are the raw data and my Half Cycle Time calculation between High and Low point of the Period: The Original Timecode is in Minute:Seconds:Frames format, and have to be converted from 60fps into milliseconds, which is a hardwork I have already done for all 😉

MSFS2020

Cycle one

Low - 1:17:47

High - 1:18:19

Neutral - 1:18:59

HalfCycle Time Period: 534 milliseconds

---------------------------------------

Cycle two

 

Low - 1:20:23

High - 1:21:04

Neutral - 1:21:34

HalfCycle Time Period: 684 milliseconds

-------------------------------------------

Cycle Three

 

High - 1:22:56

Low - 1:23:29

Neutral - 1:24:04

Half Cycle Time : 550 milliseconds

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

Now MSFS2024

Cycle One

 

Low - 1:54:44

High - 1:55:10

Half Cycle Time: 434 milliseconds

-----------------------------------------------------

Cycle Two

 

High - 1:56:21

Low - 1:56:43

Neutral - 1:57:14

Half Cycle Time: 367 milliseconds

------------------------------------------------------------

Cycle Three

 

Low - 1:58:26

High - 1:58:53

Half Cycle Time: 450 milliseconds

 

Now lets analyse your Real Life Graph of what the Half Cycle should be:

 

You have just proved my point.  The Real Life C172 Frequency Between High and Low points of the Cycle match very MSFS2020, and the trend is clearly showing it is too fast in MSFS2024. That would be what I continue to call Rubber Band Feeling, as I still dont see any point into using Engineering terms into this discussion while we can perfectly get understood in a way everyone could follow.

 

Thanks for that data! Really helped my point to show more clear than I could have alone 🙂

Edited October 15, 20241 yr by Alec

23 minutes ago, fsiscool said:

 

Therefore my assesment stands: The OP had wrong expectations. The quicker pitch settle time of MSFS 2024 and the fewer oscillations are not less realistic but actually closer to the expectation curve.

I never said anything about Pitch Oscillations. I said:
" the twichyness is much more noticeable. This makes a considerable impact on the flying qualities of the planes. "

My mention of oscillations was in the directional testing. 

I dont know where you are getting this talk about me having wrong expectations, as I made my points perfectly clear on what I was looking for and what I experienced in MSFS2020 vs MSFS2024

"So, I did some Flight Testing Maneuvers that are used in real life to determine Airplanes Stability, such as Fast Kick of the Rudder, and Fast Elevator Impulse, and immediately release all controls to neutral,  to determine not only the amplitude of the movement, but most importantly the frequency between cycles and how much does it take to dampen the oscillations back to equilibrium. I did so in the only planes I could, XCub and Cessna 172."
 

So, I just proved my point above, and I dont really see what you are going on about me having wrong expectations

9 hours ago, fsiscool said:

But how, if we search for real world data?

The best diagram, I can find for the curve that describes the correct expectation is this:

Is there a text explanation of the test that produced the graph?

Also @Alec, do you know if the top of the curve in all of your tests was 0.5 degrees of AoA?

edit 2: looking at the pitch ladder in the garmin, it seems not😅

I think for a 0.1 / 0.2 second difference in poorly matched data, I’ll go ahead with the pilot reports of feeling🤩

Edited October 15, 20241 yr by scotchegg

4 hours ago, Alec said:

You have just proved my point. 

Sorry you are making basic errors.
1. The time between the highs an lows as you call it is near meaningless (see the geogebra link).
2. The settling time of a step response is defined as the time until the response stays within a reasonable error band (for that even wikipedia is sufficiant).
3. The settling time starting point is not the peak of the response but the control input.
4. What you measure is the dampening or ringing frequency, which is near meaningless.
5. Even if you consider your wrong numbers, you are fooling yourself (and others) if you argue that a difference of roughly 150ms and standard error bands which almost overlap at the edges is making a notable difference in the sim.

The correct settling time taken from the diagram is ~0.5s (assuming a reasonable error band; a reasonable error band is defined as "errors inside the error band are not noticed by the pilot anymore").

I did just quick attempt to get the response time correctly out of your video and stopped right after I found, that MSFS2020 has a response time of 2..3 seconds when measured correctly. MSFS 2024 must be (much) closer to reality

So, no, you are absolutely not correct in your assessment.

Not only was your expectation wrong by a pretty large factor, also your analysis of the time response is flawed.

1 hour ago, scotchegg said:

Erm, the video is about pitch response, not rudder kicking. I can’t see if the graph is latitudinal or longitudinal.

I did not talk about anything else than pitch response/longitudinal stability. The diagram is also about pitch response. His 172 video does show the rudder too but I concentrated on the pitch, because he called that "much worse" while it actually improved in MSFS 2024.

 

2 hours ago, scotchegg said:

I’ll go ahead with the pilot reports of feeling🤩

This. 

11 minutes ago, fsiscool said:

Sorry you are making basic errors.
1. The time between the highs an lows as you call it is near meaningless (see the geogebra link).
2. The settling time of a step response is defined as the time until the response stays within a reasonable error band (for that even wikipedia is sufficiant).
3. The settling time starting point is not the peak of the response but the control input.
4. What you measure is the dampening or ringing frequency, which is near meaningless.
5. Even if you consider your wrong numbers, you are fooling yourself (and others) if you argue that a difference of roughly 150ms and standard error bands which almost overlap at the edges is making a notable difference in the sim.

The correct settling time taken from the diagram is ~0.5s (assuming a reasonable error band; a reasonable error band is defined as "errors inside the error band are not noticed by the pilot anymore").

I did just quick attempt to get the response time correctly out of your video and stopped right after I found, that MSFS2020 has a response time of 2..3 seconds when measured correctly. MSFS 2024 must be (much) closer to reality

So, no, you are absolutely not correct in your assessment.

Not only was your expectation wrong by a pretty large factor, also your analysis of the time response is flawed.

I did not talk about anything else than pitch response/longitudinal stability. The diagram is also about pitch response. His 172 video does show the rudder too but I concentrated on the pitch, because he called that "much worse" while it actually improved in MSFS 2024.

 

I have no idea what you are talking about. What is meaningless is your description of mistakes I made, which I have no idea what you are coming with that. Makes completely no sense to me sorry.

I am done with this discussion, I have proven your point, and how you say 150ms is fooling myself. haha nice strategy when you post your own proof that I show how proves my point. Sure OK

I will return in November when I can plot my graphs, and as always, I will show what I mean only to have people come and say how the feeling is great and I must be wrong. Got used to that, I was an Alpha for MSFS2020 and heard a lot of that, already used to it.

34 minutes ago, fsiscool said:

I did not talk about anything else than pitch response/longitudinal stability. The diagram is also about pitch response. His 172 video does show the rudder too but I concentrated on the pitch, because he called that "much worse" while it actually improved in MSFS 2024.

Sorry, yeah, I noticed that after and cheekily deleted that part😅

29 minutes ago, Alec said:

I will return in November when I can plot my graphs, and as always, I will show what I mean only to have people come and say how the feeling is great and I must be wrong. Got used to that, I was an Alpha for MSFS2020 and heard a lot of that, already used to it

Actually, I don’t think your graphs will add much to the conversation unless you can match everything in the sim version with the exact 172 and it’s environment used in the test that produced that graph (distance of travel of ailerons, condition of the hydraulics, atmospheric conditions, etc. etc. etc. etc.).

Wouldn’t you say that getting within 0.2 seconds in both versions in what you must accept are hardly scientific conditions in your testing is, as it’s put in academic parlance, ‘not bad’?

Edited October 15, 20241 yr by scotchegg

we can put all the numbers and whatever, in my opinion there is something weird in MSFS both 20 and 24 for the limited time testing, about how to plane respond to inputs.

That's the main thing that steers me away from MSFS, because flying planes every 2 days, when firing up something in MSFS it just screams wrong in my brain, when i'm doing whatever inputs the reaction is not what my brain is expecting, something like you would drive your car and when turning the wheel left, you go right .... quite extreme example, but to illustrate the feeling.

 

And we've not even started with how turbulences affect the plane...

https://tenor.com/fr/view/mickey-mouse-toy-cute-bobblehead-gif-17414052

 

You can dial in a very good FM that match real world numbers, stuff like pitch/power, climb rates, cruise etc etc, ppl have been able to do this properly since a long time ago, even on older sim(with look up tables and such etc etc ....), not that part of FM i'm referring.

 

I'm still going to wait for release, maybe 3rd parties will be able to work some magic. The good thing is that ground handling is miles better than 20, and that's a big win! But i still think there is something wrong at core....

Edited October 15, 20241 yr by Muds

20 hours ago, cyril972 said:

it feels like MSFS2020 was a proof of concept for MSFS2024.

In many ways, yeah.

Something I don't think people are amazed enough about is that ACES made all the other flight simulators until Microsoft unceremoniously fired them all. Even when the code changed in a major way, they still had years of flight simulator development experience to draw on.

The outfit that made MSFS made one game before they started, and it was a 3rd person action RPG about surviving a plague 555 years before the first airplane was invented. From day one with MSFS I was amazed at how much they got right. Now they have years of experience under their belts and a much better idea of how to build a sim. I'm not surprised at all that they wanted to do a new version.

 

4 hours ago, scotchegg said:

I’ll go ahead with the pilot reports of feeling🤩

Likewise 🙂 ... for non-pilots like us, apart from spotting obvious differences or issues we can only go by IRL pilots' opinions. .. and there too we need to hear from multiple pilots given that opinions can vary, and see what the general trend is. And so far the majority of IRL pilots feel that the MSFS 2024 C172 flight dynamics feel better than MSFS 2020 C172 from all what I can see, and a vast majority feel that ground handling is greatly improved over 2020.

Eagerly awaiting the full release of 2024 so we can get a more solidified sense of where things stand. At least that will have to be based on the default fleet.. then we'll see what 3rd party devs make of the new capabilities (and hopefully that won't take eons).
 

Edited October 15, 20241 yr by lwt1971

17 minutes ago, lwt1971 said:

Likewise 🙂 ... for non-pilots like us, apart from spotting obvious differences or issues we can only go by IRL pilots' opinions. .. and there too we need to hear from multiple pilots given that opinions can vary, and see what the general trend is. And so far the majority of IRL pilots feel that the MSFS 2024 C172 flight dynamics feel better than MSFS 2020 from all what I can see, and vast majority feel that ground handling is greatly improved over 2020.

Eagerly awaiting for the full release of 2024 so we can get a more solidified sense of where things stand. At least that will have to be based on the default fleet.. then we'll see what 3rd party devs make of the new capabilities (and hopefully that won't take eons).
 

I'm not an airline pilot, but I do fly. There is definitely a step above most other sims including 2020, when it comes to flight dynamics and ground handling. There was a slight issue with the twitchiness of the elevator inputs in the tech alpha, but sensitivities somewhat put a band-aid over it. To be blunt, I didn't set up my whole honeycomb peripherals and just used to a controller to take a look at the tech alpha. 

The ground handling was really good, the plane felt like it had some weight behind it. I'm really excited to see what 3rd party devs are going to do with their aircraft, I'm sure Fenix will have a blast with this new sim. 

To think we tested an outdated version of the sim, and the adjustments and updates we missed out on, I'm excited for next month. 

Edited October 15, 20241 yr by rick celik

1 hour ago, scotchegg said:

Actually, I don’t think your graphs will add much to the conversation unless you can match everything in the sim version with the exact 172 and it’s environment used in the test that produced that graph

I recall one of the preview talks where either Seb or Jorg mentioned the monitoring equipment they used to create the updated flight models will be documented in the SDK, giving anyone with access to a plane and said equipment the ability to compare real world measurements with how those measurements translate into the sim. I'm sure some ambitious developer will rise to the occasion!

37 minutes ago, eslader said:

The outfit that made MSFS made one game before they started

Asobo contributed to a lot more than A Plague's Tale before MSFS's release: https://en.wikipedia.org/wiki/Asobo_Studio

4 hours ago, fsiscool said:

Sorry you are making basic errors.
1. The time between the highs an lows as you call it is near meaningless (see the geogebra link).
2. The settling time of a step response is defined as the time until the response stays within a reasonable error band (for that even wikipedia is sufficiant).
3. The settling time starting point is not the peak of the response but the control input.
4. What you measure is the dampening or ringing frequency, which is near meaningless.
5. Even if you consider your wrong numbers, you are fooling yourself (and others) if you argue that a difference of roughly 150ms and standard error bands which almost overlap at the edges is making a notable difference in the sim.

The correct settling time taken from the diagram is ~0.5s (assuming a reasonable error band; a reasonable error band is defined as "errors inside the error band are not noticed by the pilot anymore").

I did just quick attempt to get the response time correctly out of your video and stopped right after I found, that MSFS2020 has a response time of 2..3 seconds when measured correctly. MSFS 2024 must be (much) closer to reality

So, no, you are absolutely not correct in your assessment.

Not only was your expectation wrong by a pretty large factor, also your analysis of the time response is flawed.

I did not talk about anything else than pitch response/longitudinal stability. The diagram is also about pitch response. His 172 video does show the rudder too but I concentrated on the pitch, because he called that "much worse" while it actually improved in MSFS 2024.

 

 

Just chiming in. Essentially it is this right?  Alec is looking at the purple and you are saying the green box is what really matters?

Edited October 16, 20241 yr by Gelrd

As @Alec says, access to the SDK and being able to record & plot flight data is the way to go.

For me, the differences can be small between 2020 and 2024, but the base problem is both leave a lot to be desired in terms of ground physics, static stability, ... 

The rather ridiculous nervous oscillation that all aircraft, from the lightest GA to the 748, exhibit when there's wind variability &  turbulence, show that no matter how good the intentions are in the way ASOBO designed the flight dynamics, there is still a lot to be fine tuned ... To be honest I by far preferred MS FLIGHT in various aspects of "feel of flight", and it's FDM was much closer to FSX than after all MSFS is with it's overlapped "CFD" approach.

Alec at least finds that ground physics in 2024 feel a lot better than in 2020... I honesty saw no difference, other than that baked by the "tricks" the flight dynamics file can include in order to fake the physics and say : "hey, don't worry... x-wind will not bother your tail until up to X knot taxiing speed...", but not only... 

I hope it's just because it's Alpha...

This being said I am sure MSFS 2024 will be my goto airline simulation, and I'll love it, just NOT for the flight dynamics and some aspects of weather modelling...

Edited October 16, 20241 yr by jcomm