Acceleration comparisons

[FC99]

First of all, it is always appreciated and respected when somebody "do the walk".

Now the bad part.

1. This kind of tests is best done with the utility written for such purpose.
You can find it here.
http://lesnihu.wz.cz/autopilot/autopilot.html
It is easy to set it up and it will execute script which will control the plane instead of you. It is much easier to test planes that way and what is even more important this is way more precise and repeatable test method.

2. Flying the plane is just half of the job, second part is logging the flight parameters. You can do it your way but there is a better and easier way. Use UDPGraph, you can get various parameters on the screen and in the file.
Download it here:
http://avcpage.achilikin.com/il2dl/graph_en.htm

Once you get these two utilities working you will be able to watch TV while your PC is doing the work for you and as a bonus you will get much better results.

[Woke Up Dead]

There's still merit in doing it "by hand" though, Horseback made a lot of useful observations about the difficulty of trimming some planes as they accelerate. I am curious about how much difference trim makes though, maybe someone familiar with that tool could set up a similar test for a couple of the planes that Horseback thought were hard to trim to see how much better the results are.

[FC99]

Originally Posted by Woke Up Dead

There's still merit in doing it "by hand" though, Horseback made a lot of useful observations about the difficulty of trimming some planes as they accelerate. I am curious about how much difference trim makes though, maybe someone familiar with that tool could set up a similar test for a couple of the planes that Horseback thought were hard to trim to see how much better the results are.

It's not about better results ( good pilot will do as good as AP), it's about consistency and ease of testing. And trimming issues are highly exaggerated, you can make good runs without messing with trim much, even in "hard to trim" planes. Just make a plane nose heavy before you start your run and you will not have any problem in making smooth run with minimal altitude deviation.

Originally Posted by horseback

The whole point of this exercise is to do it by hand; if the average player cannot duplicate the results of the tests without a ton of specialized controllers, or be able to overcome some of the problems inherent in flying a given aircraft without many, many hours of practice, meeting the 'book' numbers is meaningless.

"Book" numbers are not what average pilots could do, they are what highly trained test pilots could do.

[horseback]

Originally Posted by Woke Up Dead
There's still merit in doing it "by hand" though, Horseback made a lot of useful observations about the difficulty of trimming some planes as they accelerate. I am curious about how much difference trim makes though, maybe someone familiar with that tool could set up a similar test for a couple of the planes that Horseback thought were hard to trim to see how much better the results are.

It's not about better results ( good pilot will do as good as AP), it's about consistency and ease of testing. And trimming issues are highly exaggerated, you can make good runs without messing with trim much, even in "hard to trim" planes. Just make a plane nose heavy before you start your run and you will not have any problem in making smooth run with minimal altitude deviation.

The trimming issues are generally pretty specific to a given aircraft and seem to be not just a matter of adding nose down trim; certain aircraft will consistently raise or drop their noses abruptly at certain speeds after an extended period of acceleration in level flight, even when you try it flying in the Wonder Woman view (which is the only view option that actually provides consistently accurate and timely climb/altitude and trim data). If you do your runs 'in the cockpit' as I do, shifts in AOA as speed changes make using outside reference points (like the horizon) impractical, and dishonest, inconsistent, illegible or slow Turn & Bank indicators, variometers, altitude indicators and artificial horizons make certain specific aircraft extremely difficult to keep level, when added to their trimming problems. Certain other aircraft of similar performance seem to need much less adjustment and have either consistent or particularly accurate in-cockpit instrument displays, some of them in direct contradiction to reports of the period.

In my opinion, many of the 'hard to trim' class seem to be hypersensitive to minor stick inputs as speed increases; I use the same low stick sensitivities for all aircraft testing, as well as 50% filtering, and attempting to maintain level flight in the 'hard to trim' group with the stick and pedals is just as difficult as trying to add or subtract elevator and rudder trim with button or axis inputs, and sometimes worse.

Originally Posted by horseback
The whole point of this exercise is to do it by hand; if the average player cannot duplicate the results of the tests without a ton of specialized controllers, or be able to overcome some of the problems inherent in flying a given aircraft without many, many hours of practice, meeting the 'book' numbers is meaningless.

"Book" numbers are not what average pilots could do, they are what highly trained test pilots could do.

Test pilots of the 1930s and WWII era were largely self-taught; actual 'training' and schools for test piloting came much later. Military training required a higher standard of precision than general aviation because the military required a level of teamwork and predictability between aircraft un-needed in civilian aviation. Generally, military testing showed much less optimistic results than the manufacturers' in-house tests in the 1930s and early 1940s.

Regardless, the "Book" numbers are a basis of comparison for the average pilots; if plane a can accelerate from 270 to 450 kph in under 40 seconds and plane b takes almost a minute with the same pilot, their "book" numbers should be at least proportional. When other factors intrude or are artificially injected, the proportional differences can get a little lopsided.

cheers

horseback

[MaxGunz]

And when you don't understand what's going on you can believe any conclusion you might come up with.

Now it's time for me to watch the new UFO's from Niburu video.

[sniperton]

Maybe I'm wrong, but it seems to me that you -- horseback and woke up dead on the one side, and FC99 and MaxGunz on the other -- grind in two mills. One issue is the performance of aircrafts optimally exploited by the AI, and another is the ability/inability of the human pilot to achieve that optimum using standard game controllers (i.e. a short stick), watching a monitor less than 90 cm in diameter, and relying on flight data as displayed on the cockpit gauges. These limitations on the human player's side vary from user to user, but still there they are, and should be addressed properly when we discuss 'realism' (whatever it means for us). 'Correct' flight performance is one issue, it's actual 'feasibility' is another. Simply because we don't use the same peripherals as the AI does or r/l pilots did.

[MaxGunz]

You don't represent me at all there.

As far as test pilots, historically less than 1% of all pilots make the cut mostly because of the discipline needed. Read up, a lot of WWII Aces tried for the job and didn't make it. And yeah yeah not all test pilots made good combat pilots if that's what it takes to stop the crying, but it takes the quality of a test pilot to fly the necessary tests. Chuck Yeager made it and was noted as a natural, the two jobs are not exclusive but talent and discipline are.

If during a run the plane goes up and down even 1 or 2 meters that will change the acceleration and trim state. By the time the plane has come back down it's going to look like a sudden change. Sound familiar? Guess why I quit trying to make test runs? I'm not good enough!

From what I have read of the pro books, it takes several flights to get one segment of a test done right and many segments to make a total run. They don't just firewall it from stall and burn till top end a few times then land for beer and number crunching.

Well, what Horseback is doing is still way better than steep dive yanked into climb and then however long it takes to get down to just over 109 steady climb speed is where you call it done - check the height - claim FM bias as suspected not like the "test" wasn't set up to do just that.

[RPS69]

Originally Posted by sniperton

Maybe I'm wrong, but it seems to me that you -- horseback and woke up dead on the one side, and FC99 and MaxGunz on the other -- grind in two mills.

Nope, TD only tries to use a common reference for everybody, and they need to use a criteria and stick to it, even if some other resources says something is different.

Whenever you attack a particular set of aircrafts from a different resources, if they agree to changeit, you are opening the door for someone else to claim another resource for another particular set of aircrafts.

Let them stick to what they decided as a normal path, just try to give support on bizarre things, and improving some effects implementations.

[horseback]

Originally Posted by FC99

First of all, it is always appreciated and respected when somebody "do the walk".

Now the bad part.

1. This kind of tests is best done with the utility written for such purpose.
You can find it here.
http://lesnihu.wz.cz/autopilot/autopilot.html
It is easy to set it up and it will execute script which will control the plane instead of you. It is much easier to test planes that way and what is even more important this is way more precise and repeatable test method.

2. Flying the plane is just half of the job, second part is logging the flight parameters. You can do it your way but there is a better and easier way. Use UDPGraph, you can get various parameters on the screen and in the file.
Download it here:
http://avcpage.achilikin.com/il2dl/graph_en.htm

Once you get these two utilities working you will be able to watch TV while your PC is doing the work for you and as a bonus you will get much better results.

The whole point of this exercise is to do it by hand; if the average player cannot duplicate the results of the tests without a ton of specialized controllers, or be able to overcome some of the problems inherent in flying a given aircraft without many, many hours of practice, meeting the 'book' numbers is meaningless.

This highlights some of the reasons that the ai so consistently outperform human players in certain aircraft and why some aircraft that should have many more users based on their historical records are less successful with occasional users.

cheers

horseback