Thank you for the new FM

[pierrewind]

Hello, I finished my calcs and plotted them over the real MiG-21bis Cl table.

 

The shaded area is the buffeting zone, and upper that the plane will stall, downwards, all must to be ok.

 

* Blue line: The Cl with 33º on the UUA-1 indicator

* Orange line: The Cl with max possible pitch

* Yellow line: The CL with 21º on the UUA-1 indicator

 

[1] EDIT: Possible DCS bug related to the atmospheric lift, compare the two attached images, one at 760mmHg and other at 900mmHg, also check the corrected rho value. For this maybe the current FM Cl is ok for < 0.6M

[2] EDIT: Using the ARU-3V in manual forced at full left the FM seems a lot better! (attached img2)

 

As you can see actually the M3 MiG-21 has three issues:

1) the UUA-1 indicator don't work as the real one.

2) the FM at max pull over-performs the pre-stall MiG-21 capabilities under 0.6M (check the possible DCS bug[1])

3) the FM at max pull under-performs the MiG-21 capabilities over 0.6M[2]

 

As conclusion, the FM is near to real on 0.6M and under[1], and the biggest issues are the under-perform over 0.6M using the ARU-3 in auto[2] that it's very noticeable and the UUA-1 issue that at this point makes this instrument totally useless and difficult to compare the exact non-buffeting values.

 

Anyway, thank you a lot for this update, that did a better FM than before and set it on the right way!

 

UPDATE: The tables was updated with ARU-3V in manual and auto.

 

The chart and tables are attached as libre office calc and an image.

 

Thanks to you for this great piece of research.

 

Assuming this computation is correct as is my reading, I am forced to disagree with your observations.

 

-According to the orginial chart, below mach 0.6, @33 pull, You should be obtaining a stable CL @0.8. From your testing, it is shown at a value well above 1.2. This equals to a 50% performance gain. To me this is very very far from being negligeable.

 

-I also think you somehow missed a few important details. That is perfectly normal when working on such advanced computing. But, you cannot put a curve For UUA @ 33 without this value being maintained. I understand this was not possible to test at low level as you would have gone well above max G. Should you wish to continue your blue curve on chart 1 past mach 0.6, you need to fly higher, this will have the effect of reducing the air density and therefore allow you to perform the tests correctly.

 

-I know there is some very complex wording and formulas involved. When you are saying max pull, I assume you are talking about maximum stick input. This maximum stick input doesn't show anywhere on the original chart. You shall plot the max stick input based on the associated UUA, speed and CL, according to my observations, you do so solely based on a combination of speed and CL which is not scientificaly correct. What happens at this point is the g protection kicks in before you are given the chance to reach max CL. This is somehow solved in chart 2 by smartly saying telling R2D2 to sleep but, climbing would work as well.

 

-For your testing, The UUA 21 chart seems to be using the perfect technique. Unfortunately, it shows again a gap to the actual chart between about 50% and 30% gain. IRL that would likely equate from 7g instantanuous being improved to about 9.8g. As you can see this is a very big difference to the actual aircraft.

 

The variable that is not explicitely stated in the original chart is the altitude. The test pilots have had to fly at much higher altitude in order to obtain the values without putting their lives at risk (like pulling 20Gs :cry: ). In theory, this entire chart could be ploted flying at 1g. (very stupid to do IRL; probably even impossible.)

The great values you obtained should ideally be looked at in their entirety. I stated on different points that the CL was up to 50% higher. This difference could also be explained by unprecise weight, wing area, density. But, if the entire curve manages to have a similar shape, only a coherent g load could validate the entire chart.

 

But then, the given lift coeficient for a given aircraft is just one part of the equation. Drag, and above all controlability are critical factors to consider. According to the actual flight manual, the current flight model does not exhibit any of the caracter the real life mig 21 should have at high AOA. That's what I miss the most with the current mig21 flight model.

 

I really thank you for your amazing dedication and work! :joystick: Unfortunately, to me all it proved if it can prove anything is that the flight model is very very far from the actual aircraft at the moment.

 

By the way, well done on passing my reasonability check, it's not every day you see some coherent data!:book:

[riojax]

Well yes, it is possible that the UAA vs actual AoA is wrong somehow, but still both your CL max and CL 33 lines are way above the highest line on the original chart. CL=1.25 should be the highest at some extreme AoA, yet you have measured 1.39.

 

Yes, this is that I said at first as a FM issue, but testing the current air density effect on lift on all DCS modules gave me to think that probably this is the culprit of this slight difference on the limit of the stall and buffet. Think that you can pass my line without stall and is near the real stall point.

 

Also, thanks, you gave me a really good idea and was to calculate the real LF value for the UUA-1 33º and later get the AoA in DCS using the current FM. The difference using this method for some values must to be enough to tune the current UUA-1 implementation.

 

P.S. I updated the original post.

[pierrewind]

Yes, this is that I said at first as a FM issue, but testing the current air density effect on lift on all DCS modules gave me to think that probably this is the culprit of this slight difference on the limit of the stall and buffet. Think that you can pass my line without stall and is near the real stall point.

 

Also, thanks, you gave me a really good idea and was to calculate the real LF value for the UUA-1 33º and later get the AoA in DCS using the current FM. The difference using this method for some values must to be enough to tune the current UUA-1 implementation.

 

P.S. I updated the original post.

 

To me I don't see any issue in respect to UUA vs AOA... it seems to be linear as expected within acceptable error margin.

[riojax]

Thanks to you for this great piece of research.

Wow, thank you for your time and great response.

 

-According to the orginial chart, below mach 0.6, @33 pull, You should be obtaining a stable CL @0.8. From your testing, it is shown at a value well above 1.2. This equals to a 50% performance gain. To me this is very very far from being negligeable.

Yes, but the problem is that the UUA-1 is wrong and their values can't be usable to match the graph.

The only usable value at now is the stall point, as it don't depends of instruments readings. You can see the new update with the second image yellow line using calculated and simulated values that can be used to fix the UUA-1 bug.

 

-I also think you somehow missed a few important details. That is perfectly normal when working on such advanced computing. But, you cannot put a curve For UUA @ 33 without this value being maintained. I understand this was not possible to test at low level as you would have gone well above max G. Should you wish to continue your blue curve on chart 1 past mach 0.6, you need to fly higher, this will have the effect of reducing the air density and therefore allow you to perform the tests correctly.

You are right, mess the maximum values with the 33º was an error, but it's fixed now.

About the air density, the bug is not about that, the bug is that in DCS changing the air density and by that later the rho according the value, the LF changes a lot more than it do in real. And by that bug, all FM can have a bigger/lesser Cl than the real plane (but the plotted line form must to be the same)

 

-I know there is some very complex wording and formulas involved. When you are saying max pull, I assume you are talking about maximum stick input. [...]

No, sorry it was my fault, I mean maximum pull until stall.

 

-For your testing, The UUA 21 chart seems to be using the perfect technique. Unfortunately, it shows again a gap to the actual chart between about 50% and 30% gain. IRL that would likely equate from 7g instantanuous being improved to about 9.8g. As you can see this is a very big difference to the actual aircraft.

Right, I'm not saying that the current FM is perfect, only it's in the good way to be :)

 

The variable that is not explicitely stated in the original chart is the altitude. The test pilots have had to fly at much higher altitude in order to obtain the values without putting their lives at risk (like pulling 20Gs :cry: ). In theory, this entire chart could be ploted flying at 1g. (very stupid to do IRL; probably even impossible.)

The great values you obtained should ideally be looked at in their entirety. I stated on different points that the CL was up to 50% higher. This difference could also be explained by unprecise weight, wing area, density. But, if the entire curve manages to have a similar shape, only a coherent g load could validate the entire chart.

Ok, those charts can be difficult, I attached the 2nd at more detail.

The lime line is the stall point, passed it the plane will stall.

The shadowed is the buffet zone, and the green is the lift limit for 0m altitude. As you can see the blue line must to be in the 0m and not in 10km as now, but it can be done.

 

But then, the given lift coeficient for a given aircraft is just one part of the equation. Drag, and above all controlability are critical factors to consider. According to the actual flight manual, the current flight model does not exhibit any of the caracter the real life mig 21 should have at high AOA. That's what I miss the most with the current mig21 flight model.

Sadly I never flyed an MiG-21... for this is better to listen real pilots trying it on DCS.

 

I really thank you for your amazing dedication and work! :joystick: Unfortunately, to me all it proved if it can prove anything is that the flight model is very very far from the actual aircraft at the moment.

You're welcome, also thank you for this great and detailed response. About the FM, yes... it needs a lot work but I think that this update is in the good way to do that.

 

By the way, well done on passing my reasonability check, it's not every day you see some coherent data!:book:

Oh yeah! I want a logo for this unlocked achievement :)

[riojax]

To me I don't see any issue in respect to UUA vs AOA... it seems to be linear as expected within acceptable error margin.

Look at the second image, the yellow table shows the calculated LF, speed, altitude, etc. for the UUA-1 33º value, now try this parameters in DCS and look at the UUA-1 values. This is the error.

[pierrewind]

Look at the second image, the yellow table shows the calculated LF, speed, altitude, etc. for the UUA-1 33º value, now try this parameters in DCS and look at the UUA-1 values. This is the error.

 

I am really not quite sure to understand what the issue is.

 

Just because it probably is something trivial, let's go back to basics.

 

On your said calculated chart (not actual data if I understand correctly), I see mostly everything being coherent (more on that later).

 

On a given wing, at a given angle of attack, the CL should remain constant. That's actually true for low speeds, but as we get into compressability issues (higher mach), the CL can change slightly hence the bump on the chart.

 

As this angle of attack should be "fixed" I do not see why we have a static UUA and a variable AOA. This just doesn't make sense to me.

 

As such, I would tend to think that something is leading you to believe the AOA should change. I think If you could tell me more as to why this aoa value is changing, we could nail this data.

 

Based on my very limited testing in DCS, I see perfect correlation between UUA and AOA irrespective of the speed. As it should be according to the DCS mig manual.

 

 

As for the achievement logo... hum... something like a kerbal test pilot would make sense I think :P

[Frederf]

I think UUA and AOA relationship in reality is highly variable while in DCS module it is a fixed relationship. CLmax values seem mostly OK but they are always occurring at the same UUA 33 the same at 0.9 and 0.5 Mach. UUA is effectively a true AOA gauge in funny units. But that is not really a FM issue but an instrument one.

 

CL-max graph with fixed tail plane arm shows very good adherence. It also suggests that when tail plane is ARU-controlled showing much difference that ARU logic is not good. Again not FM issue but system logic one.

[pierrewind]

https://forums.eagle.ru/showpost.php?p=3457419&postcount=2

 

On page 54/55, there is the description Of the UUA.

 

It clearly states that the UUA "IS" the AOA and it is not variable it just is data as is the speed, altitude... It makes sense to find a correlation between the two in DCS!

 

I will try to get some DCS data tomorrow and make it stick together. But right now I think we are creating issues with that UUA that are not actually present.

 

Edit: Due to the positioning/Flow error on the vane, there is a discrepency between actual AOA and UUA indication. Similar to the difference between Calibrated and true airspeed. This difference doesn't prevent one given unit UUA to allways be equal to a same AOA.

 

https://forums.eagle.ru/attachment.php?attachmentid=105667&d=1412951059 This article might help shed some light.

Edited April 24, 2020 by pierrewind

[Argo Navis]

Very interesting discussion gents.

 

This graph was posted here a couple of years ago in a similar topic, and I have been looking for the book it came from for a while now. No luck so far.

I have a bunch of original manuals in different languages, including Russian ones, but none of these manuals have this exact graph. A simplified version of this graph is there in every book I have, but not this one. This graph has much higher data resolution and more detail. My suspicion is that this page is from a book that goes into much more detail about the aerodynamics of the Mig-21 than the common flight manual. Since it is a pretty good quality scan, I suppose the whole book is out there somewhere with even more graphs about the aircraft that might help to improve the FM further.

Does anybody here have the whole book by any chance?

[riojax]

It clearly states that the UUA "IS" the AOA and it is not variable it just is data as is the speed, altitude... It makes sense to find a correlation between the two in DCS!

 

Nope. This post explain it perfectly:

https://forums.eagle.ru/showpost.php?p=3424565&postcount=130

[pierrewind]

Some quick testing on my side. (see attached Sheet).

 

I yeld very similar results to those found by Riojax, which should be considered as a good sign :P .

 

Based on actual Charts, it shows a significant increase in CL anywhere between 40% and 80% Outch...

 

It shows a clear downward slope that shouldn't exist.

 

I kept the data above mach 0.9 for reference. However my flying wasn't precise enough to consider these values as accurate. I was mostly pulling too much as the speed decreased. This would tend to confirm a drop too charp in the current flight model as it should mostly remain constant below mach 1.1-1.2. However At these speeds, we actually get closer to the expected flight performance.

 

Now for the AOA issue. I must say, I'm getting down the rabbit hole at this point.

 

On the attached document this is what is being said for a configuration similar to the one used in the mig 21 (page 11)

 

"The tests were conducted over a Mach number rsnge of 0.6 to 1.02.

Sample calibrations of the installation at M= 0.6, 0.81, and 0.92

me given in figuree 13. These data, as well as those at the other test

speeds, show a linear vsriation of local angle of attack with true angle ‘-

of attack. The data also show that for constant angle of attack the

local angle of attack decreases as the Mach number increases. In addi- -

tion, the local angle of attack is shown to chsmge about 1.6° for each

degree change in true angle of attack. Faother locations of the sensor

on the nose and for other fuselage nose shapes, the variations of locsl

singleof attack with true angle of attack will, of course, be different."

 

The variations are mostly due to mach effects in this document. This is expected; compressibility is Bad!

 

But what is really important is the corelation between actual AOA and measured AOA is maintained. This perfectly applies to the Mig 21 and the UUA.

 

Based On my data, I show that as the UUA is maintained (within 1 unit below mach 0.8 ) it doesn't maintain a stable AOA. We are talking about 1-2 degrees difference; this isn't very significant, but that is noticeable. This could be explained by Bad uua logic, Bad AOA logic (not as if there are no bugs in DCS :cry: ), or mach effects/actual probe errors.

 

This makes the calculations potentially out of place. However, As the UUA is maintained, the CL increases with a reduction of speed; it shouldn't. But if we look at this through the AOA, as the AOA reduces, the CL Increases; it shouldn't. So wichever way we look at this data, it just doesn't work. Which would tend to confirm the current flight model Is way off.

 

There is a last note I have to point out. During testing, I found a discrepancy between the reported fuel level between DCS and the one reported by tacview. I used the one reported by tacview. It was showing more than the one reported by DCS; therefore it is likely to contribute toward the higher found CL. 6284.4 was FF with tacview; 5027 was FF with DCS. This could possibly contribute to a false reported 6% performance gain.

 

Edit: added UUA 28 info. Similar result found.

DCS Mig 21 FM testing.zip

Edited April 25, 2020 by pierrewind

[pierrewind]

Got to play arround With AOA vs UUA.

 

Based on data obtained, the linearity of the the relation is confirmed. (At 0 UUA, AOA was extrapolated). This shows a correct behaviour in regards to the variation of UUA and AOA.

 

The mach effects shows up with an almost perfect predictability. It seems to be maybe a tad too strong, but it is coherent to the papers shared in this topic.

 

With this data, the behaviour of the UUA probe is to be deemed reasonable. :pilotfly:

Edited April 25, 2020 by pierrewind

[Moody1204]

Thanks you for this robust approach and your interesting testing.

This proove that was stated before you clarify the topic and put an end to this discussion.

 

Hope this will stop some people from thinking that you can pull as much as you want of the stick without being worried about stalling.

 

 

Cheers.

[Fri13]

I just cannot understand that we keep getting such major FM changes with the MiG21Bis, that it doesn´t land at a final FM with some fine tuning here and there. I believe that M3 owes an explanation here as to what is going on, especially to all of us who have been there from the start and before.

 

That is my problem, as it escapes the logic that the MiG-21Bis has received since the release major or even extremely radical flight modeling changes, that has gone not just from one direction to one, but it is zig-zagging to all directions.

 

And when someone is jumping all over places, it is a sign of something....

[rinkerbuck]

yeah I'm rather annoyed that they have changed FM qualities at least twice in just a couple of months on such an old module, and currently it's drastically different than it was just a few weeks ago, and apparently is just broken until the next patch., This on top of the FPS killing radar.

[riojax]

Thanks you for this robust approach and your interesting testing.

You're welcome.

 

This proove that was stated before you clarify the topic and put an end to this discussion.

Sadly this didn't ended the most tricky topics, for example, the actual UUA-1 behavior seems to not match with the real one and it's difficult to know how the electronics send the data from the sensor to the UUA-1 indicator and how it's modified. Anyway this peer reviewed data throws light on the FM behaviour.

 

Hope this will stop some people from thinking that you can pull as much as you want of the stick without being worried about stalling.

This is a false dilemma fallacy. No-none talked about "pull as much as you want". We are talking about that the current FM have CLmax issues. Some problems shows a possible slight gain on <0.6M (but also we discovered a possible DCS bug that can trigger this problem) and a gain on >0.9M at ground level using the ARU-3V on manual but a huge performance loss using the ARU-3V on auto.

 

About your feeling of "pull as you want" it's more an actual ARU-3V issue than a FM one. Test it yourself, the data is here and peer reviewed.

[riojax]

I yeld very similar results to those found by Riojax, which should be considered as a good sign :P

Nothing more to say, your data and approach seems right to me too.

[riojax]

Based on data obtained, the linearity of the the relation is confirmed. (At 0 UUA, AOA was extrapolated). This shows a correct behaviour in regards to the variation of UUA and AOA.

As I know the DUA-3 have an electronic system that manage the raw AoA and sends it to the UUA-1 indicator. This data can be modified using an unknown algorithm and it wouldn't be linear... who knows.

If someone can get some cockpit videos showing the UUA-1, speedometer, etc. maybe we can try to extrapolate.

[riojax]

There is a last note I have to point out. During testing, I found a discrepancy between the reported fuel level between DCS and the one reported by tacview. I used the one reported by tacview. It was showing more than the one reported by DCS; therefore it is likely to contribute toward the higher found CL. 6284.4 was FF with tacview; 5027 was FF with DCS. This could possibly contribute to a false reported 6% performance gain.

Oh! I used the raw F2 DCS view data...

I must to review that. Thank you!!

[Enduro14]

so following this thread is interesting but dizzing at same time. What is the cut and dry answer, is the flight model incorrect or is it actually closer to reality via those charts? The way some folks where waving the pitch forks makes it seem for outsiders that the Mig21 is a screwed up right now...?

 

:joystick:

[Moody1204]

You're welcome.

 

 

Sadly this didn't ended the most tricky topics, for example, the actual UUA-1 behavior seems to not match with the real one and it's difficult to know how the electronics send the data from the sensor to the UUA-1 indicator and how it's modified. Anyway this peer reviewed data throws light on the FM behaviour.

 

 

This is a false dilemma fallacy. No-none talked about "pull as much as you want". We are talking about that the current FM have CLmax issues. Some problems shows a possible slight gain on <0.6M (but also we discovered a possible DCS bug that can trigger this problem) and a gain on >0.9M at ground level using the ARU-3V on manual but a huge performance loss using the ARU-3V on auto.

 

About your feeling of "pull as you want" it's more an actual ARU-3V issue than a FM one. Test it yourself, the data is here and peer reviewed.

 

 

 

 

From the begining I stated that the problem is the FM, I don't mind if the problem comes from the ARU-3V or the UUA indicator.

 

Concerning the ARU-3V, you can use manual mode and you'll still be able to pull as many AOA, Gs you want' you'll never stall.

 

 

 

Before last update you could "feel" when you approached the limits now you can't (because it appears there are no limits at all).

 

 

You're not suppose to pull as much without buffeting/stallin.

 

You're not suppose to do barrel roll on final prior to touch down, this is wrong that's it (try you'll be fine).

Edited April 29, 2020 by Moody1204

[riojax]

so following this thread is interesting but dizzing at same time. What is the cut and dry answer, is the flight model incorrect or is it actually closer to reality via those charts? The way some folks where waving the pitch forks makes it seem for outsiders that the Mig21 is a screwed up right now...?

The FM is pretty good and accurate, of course it have issues to be fixed and some inaccuracies, but the F-5E for example is a lot worse[1]. The two big issues regarding that are the UUA-1 indicator behavior that at now is incorrect and the ARU-3V that is buggy a this moment.

About the pitchforks... think that the FM changed a lot times and the people trends to think that the correct one is the first that they tried, or the most familiar. Anyway the current data shows that the current FM is on the good way, with issues but nothing really serious.

 

[1] See the attached image.

Edited April 29, 2020 by riojax

[riojax]

From the begining I stated that the problem is the FM, I don't mind if the problem comes from the ARU-3V or the UUA indicator.

This topic is more complex than it. The FM seems acceptable as you can see on the CLmax graphs, but the UUA-1 is broken and you can't to test a known UUA-1 value to fully ensure the FM quality.

 

Concerning the ARU-3V, you can use manual mode and you'll still be able to pull as many AOA, Gs you want' you'll never stall.

No, you can pull until the stall barrier, but never cross it, due that you will not stall. Not is a "pull as many AOA, Gs you want" issue, is more related to the wing form, the real CL parameters and the ARU-3V. Of course it must to be tuned and you must to be able to pass those limits on manual, but in automatic the behavior must to be similar to the current one (without the current CLmax underperforming)

 

Before last update you could "feel" when you approached the limits now you can't (because it appears there are no limits at all).

Ask a real pilot about that, think that the old behavior can be wrong. With only data on the hand the current one seems more real-like, anyway only a real pilot can say you more on it, until that, the data is the only that we have.

 

You're not suppose to pull as much without buffeting/stallin.

The red lights are indicating you the buffeting, and yes, I miss a bit more vibrations, but who knows, maybe the real air-frame is more stable that we think...

 

You're not suppose to do barrel roll on final prior to touch down, this is wrong that's it (try you'll be fine).

Wow, I will try it, but yes, on a big AoA and almost without energy if you do a barrel roll it must to end bad :P

Edited April 29, 2020 by riojax

[LupinYonder]

This has turned in to a fantastic discussion with data collected an analysed. I have to imagine the Devs went through a similar process.

 

Why are they so silent! It's taken 8 pages ( not counting many many other threads ) and many hours of hard work by Mig 21 fans to discover the FM properties when the Devs could have given us this information or discussed the changes and their intentions with us openly, making a lot of this discussion redundant.

 

If we had a more open dialogue with the devs, a lot of the mysteries of the FM would be answered and a lot of the disagreements over realism and accuracy settled. I mean, they don't even mention their changes in the patch logs!

 

It lands on Hiromachi to try and put out the forum fires while having no real information to offer us. It's very frustrating.

[pierrewind]

To Me, there are currently some big flaws to the flight model.

 

1) Too much perfromance. The Wing has a CL too high between 40% and 80%. This CL is directly responsible for a increased instantaneous turn rate by about the same number...Bad

 

2) High alpha flight caracteristics. As you get into High AOA, according to the mig flight manual, you should start to get into buffet.

 

3) It will not stall

 

4) As per the mig-21 flight manual "If the control stick is not set neutral immediately after onset of stall, during 3-4s, the aircraft will start oscillating briskly and unsteadily about its three axes, with high lateral g loads and rudder forces", "rocking of wing to wing". But since the current mig-21 is not stalling, it's not really relevant...

 

Therefore to me, at the moment the flight model is badly flawed.

 

As for the UUA-1, my personal position is quite different from Riojax.

My testing showed

-Perfect linearity between AOA and UUA at all speeds

-Mach effects showing with a good predictability and in accordance with papers from the NASA.

 

As such, to me, the behaviour of the probe is to be deemed accurate. Now, Riojax seems to be convinced that this probe should not display any errors due to non referenced calculators. However, to my knowledge no such system exists on the mig-21; the probe is only calibrated to show correct LOCAL angle of attack based on ACTUAL LOCAL angle of attack. I see 0 mention of any compensation between measured (actual) LOCAL AOA and actual AOA. Therefore, to me the UUA-1 Probe is behaving as it should.

 

I'd love to hear more from Riojax as to why the UUA-1 probe does not behave correctly, but I need this info to be based on facts and available documentation.

 

PS: Riojax, I think when you're saying the the behaviour of the UUA probe is incorrect. I think you say that based on Max CL coherent, then as you pull to a specific UUA, the CL doesn't stay coherent. You assume this error is due to a faulty UUA behaviour. However, based on my personal testing, I would tend to say the behaviour of the probe is correct. Therefore to me the only variable left is the CL i think the CL coded by magnitude is wrong. Your point of view is that the probe is wrong and the flight model correct.

I want to hear more from you as to why this UUA sensor is bad as that could clear up some misconceptions from anyone.

Edited April 29, 2020 by pierrewind