reported earlier AIM-120 can still be defeated by barrel rolls at high altitudes

[Default774]

The AIM-120 is still very vulnerable to barrel roll manoeuvres at higher altitudes. A well timed barrel roll defeats the missile every single time.

120_barrelroll_BVR.trk 120_barrelroll_HALFRANGE.trk 120_barrelroll_NEZ.trk 120_barrelroll_BVR.acmi 120_barrelroll_HALFRANGE.acmi 120_barrelroll_NEZ.acmi

Edited December 7, 2022 by Default774

[Moonshine]

^ this

[_SteelFalcon_]

 

[Ironhand]

Hmmm…looks more like an aileron roll…which makes it even worse…

[Default774]

In game perspectives of all tracks

NEZ
https://streamable.com/yf3mrd

Half range

https://streamable.com/7rwaw3

BVR

https://streamable.com/tg5awm

Edited December 7, 2022 by Default774

[Coxy_99]

18 hours ago, Ironhand said:

Hmmm…looks more like an aileron roll…which makes it even worse…

 Agree ive tried barrel rolling without adding aileron and it hits everytime.

[Маэстро]

1. This was reported eralier

2. That's not the simple aileron roll  - I see g-load between 10-11g in every track, so it's the high-speed high-g barrel rolling and that's the worstest type of maneuver for missile in fact.

Missile flies to the PIP(predicted intercept point) and this PIP moves with very high speed in such type of maneuver. The level of target's g-load defines how far the PIP should shift from its current position. If target performs rolling PIP moves in circle around target direction of flight. The higher g-load is the bigger radius of that circle. The higher roll rate is the faster PIP moves along the circle. The linear velocity of PIP on circle is proportional to the product of radius by roll rate(for ex. 100m radius and 5 radinas per second roll rate (~3/4  of revolution per sec) produces 500m/s linear velocity). That forces missile to perform very tight spiral with g-load at the edge of missile capabilities(at close range g-load required to intercept may be out of missile capabilities).

3. We already made several changes in our internal version and missile will behave better in such conditions, but there will be no 100% hit rate of course. In attached tracks missile hits 2 of 3 times now. One miss happens in "halfrange" track because missile has no enough velocity to reach PIP in time.  These changes will be avaiable in the next update.

[Rhayvn]

I don't know what you -think- you fixed, but 120s are incredibly broken after the patch.  Tested AI against AI and got a ~25% hit rate against Mirage 2000s, Mig-29As and SU-27s.  When testing as a player against AI Mirage 2000 and Mig-29S, I managed to get the hit rate to 30% against the mirages but under 20% against the Mig-29S. 

Not.  One.  Single.  Initial.  Shot.  Hit.  Over 10 tests with player against AI.

This is with a 50 mile initial setup and most engagements starting at 25-35K for both aircraft at high speed and then dropping towards the deck after defending against initial shots (First shots from both sides at 22-28 miles).  This was a common engagement type pre-patch and was reliably first shot kill 50-75% of the time.  The missiles are not low on energy when they get to the target.  They are simply tracking poorly in terminal.  Either flying a pure pursuit curve or appearing to drop track as soon as the AI does a roll while notching.

This is the same experience every pilot in my group is having post patch.  I setup the tests to control for other variables we were experiencing in missions and the tests proved consistent.

If your goal is to have AIM-120s with about a 25% hit rate against any reasonably maneuverable target, you succeeded in admirable fashion.  I am sure all the people who have extra time to use DCS over the holidays will appreciate the gift.

Edited December 17, 2022 by Rhayvn

[Default774]

14 hours ago, Rhayvn said:

I don't know what you -think- you fixed, but 120s are incredibly broken after the patch.  Tested AI against AI and got a ~25% hit rate against Mirage 2000s, Mig-29As and SU-27s.  When testing as a player against AI Mirage 2000 and Mig-29S, I managed to get the hit rate to 30% against the mirages but under 20% against the Mig-29S. 

Not.  One.  Single.  Initial.  Shot.  Hit.  Over 10 tests with player against AI.

This is with a 50 mile initial setup and most engagements starting at 25-35K for both aircraft at high speed and then dropping towards the deck after defending against initial shots (First shots from both sides at 22-28 miles).  This was a common engagement type pre-patch and was reliably first shot kill 50-75% of the time.  The missiles are not low on energy when they get to the target.  They are simply tracking poorly in terminal.  Either flying a pure pursuit curve or appearing to drop track as soon as the AI does a roll while notching.

This is the same experience every pilot in my group is having post patch.  I setup the tests to control for other variables we were experiencing in missions and the tests proved consistent.

If your goal is to have AIM-120s with about a 25% hit rate against any reasonably maneuverable target, you succeeded in admirable fashion.  I am sure all the people who have extra time to use DCS over the holidays will appreciate the gift.

 

Would you mind sharing tracks/tacviews from your testing sessions?

[Rhayvn]

Here are a some that cover Blue vs. SU-27.  Blue vs. Mig-29S (With AIM-120Bs) and one with the same scenario but the blue units are JF-17s with SD-10s.  Notice the much better hit rate of the SD-10.

Tacview-AIM120B2.zip.acmi Tacview-AIM120B1.zip.acmi Tacview-SD101.zip.acmi Tacview-SU272.zip.acmi Tacview-SU271.zip.acmi

[Rhayvn]

Two more for Mirage and Mig-29.

The initial shot hit rate is pretty poor mostly due to the targets defending and going cold.  But not all and certainly not the follow up shots.

Some of Player F-16 vs. Ace AI Mirage2000 and Mig-29S.

 

Tacview-Mig291.zip.acmi Tacview-Mirage1.zip.acmi Tacview-PlayerMig2.zip.acmi Tacview-PlayerMig1.zip.acmi Tacview-PlayerMirage2.zip.acmi Tacview-PlayerMirage1.zip.acmi

[Маэстро]

On 12/18/2022 at 3:41 AM, Rhayvn said:

I don't know what you -think- you fixed, but 120s are incredibly broken after the patch.  Tested AI against AI and got a ~25% hit rate against Mirage 2000s, Mig-29As and SU-27s.  When testing as a player against AI Mirage 2000 and Mig-29S, I managed to get the hit rate to 30% against the mirages but under 20% against the Mig-29S. 

Not.  One.  Single.  Initial.  Shot.  Hit.  Over 10 tests with player against AI.

This is with a 50 mile initial setup and most engagements starting at 25-35K for both aircraft at high speed and then dropping towards the deck after defending against initial shots (First shots from both sides at 22-28 miles).  This was a common engagement type pre-patch and was reliably first shot kill 50-75% of the time.  The missiles are not low on energy when they get to the target.  They are simply tracking poorly in terminal.  Either flying a pure pursuit curve or appearing to drop track as soon as the AI does a roll while notching.

This is the same experience every pilot in my group is having post patch.  I setup the tests to control for other variables we were experiencing in missions and the tests proved consistent.

If your goal is to have AIM-120s with about a 25% hit rate against any reasonably maneuverable target, you succeeded in admirable fashion.  I am sure all the people who have extra time to use DCS over the holidays will appreciate the gift.

 

If your goal is to have bugs fixed it would be better to add several tracks instead of quips.
(Yes, we need exactly your tracks(not acmi) to run them under debugger and see that happens in every single case)

[MARLAN_]

In my tests (35K vs 35K, Ace AI 2 ship MiG-29 Mach 1.0 vs 1.2, 30 degree aircraft loft shot at 35nm) I've seen 17/20 hits, of those 3 misses 2 were kinetically defeated and 1 was not supported to pitbull due to the look down penalty (hopefully this will be fixed as part of the F-18/F-16 radar improvments) causing drop track during a/c loft.

Personally I'm very happy with it so far although that's not to say it's perfect or the test cases Rhayvn performed aren't accurate or anything

When I was testing notching though, the AMRAAM often failed to intercept in close when it had sufficient energy, I'm assuming it's a guidance issue where it's not pulling enough lead.

[Rhayvn]

3 hours ago, MARLAN_ said:

In my tests (35K vs 35K, Ace AI 2 ship MiG-29 Mach 1.0 vs 1.2, 30 degree aircraft loft shot at 35nm) I've seen 17/20 hits, of those 3 misses 2 were kinetically defeated and 1 was not supported to pitbull due to the look down penalty (hopefully this will be fixed as part of the F-18/F-16 radar improvments) causing drop track during a/c loft.

Personally I'm very happy with it so far although that's not to say it's perfect or the test cases Rhayvn performed aren't accurate or anything

When I was testing notching though, the AMRAAM often failed to intercept in close when it had sufficient energy, I'm assuming it's a guidance issue where it's not pulling enough lead.

I assume you were using 29As since locking a 29S at those ranges is not possible if they are jamming.  Can you tell me what BVR weapons they had?  (I also assume you are running the latest open beta version.)

I setup an Ace AI 18 against an Ace AI 29A at 35k as in your test and the attached was the result.  No 120 hits and the 18 killed by R-60 and then by R-27ER.  0-5 on the 120s.

I would attach the track files but I discovered I turned them off at some point in the past.  They are back on now and any follow ups will have track files attached as well.

Tacview-18-29-2.zip.acmi Tacview-18-29-1.zip.acmi

[MARLAN_]

17 minutes ago, Rhayvn said:

I assume you were using 29As since locking a 29S at those ranges is not possible if they are jamming.  Can you tell me what BVR weapons they had?  (I also assume you are running the latest open beta version.)

I setup an Ace AI 18 against an Ace AI 29A at 35k as in your test and the attached was the result.  No 120 hits and the 18 killed by R-60 and then by R-27ER.  0-5 on the 120s.

I would attach the track files but I discovered I turned them off at some point in the past.  They are back on now and any follow ups will have track files attached as well.

Tacview-18-29-2.zip.acmi 165.64 kB · 1 download Tacview-18-29-1.zip.acmi 179.07 kB · 1 download

MiG-29A, shooting RS-AA-10c (R-27ER), latest open beta version

I am flying the F-18 as a client vs the AI MiG-29A's though, not doing AI vs. AI tests, so that may be where the discrepancy is originating.

Edit: In both of your Tacviews, the AMRAAM does not appear to be guiding at all, only flying to last known intercept (and surprisingly, doesn't seem to acquire the contact at all even when it looks like it's in the seeker FOV, and the contact isn't maneuvering) it may be worth it to try repairing your DCS before follow on tests.

Edited December 19, 2022 by MARLAN_

[Rhayvn]

8 minutes ago, MARLAN_ said:

MiG-29A, shooting RS-AA-10c (R-27ER), latest open beta version

I am flying the F-18 as a client vs the AI MiG-29A's though, not doing AI vs. AI tests, so that may be where the discrepancy is originating.

Edit: In both of your Tacviews, the AMRAAM does not appear to be guiding at all, only flying to last known intercept (and surprisingly, doesn't seem to acquire the contact at all even when it looks like it's in the seeker FOV, and the contact isn't maneuvering) it may be worth it to try repairing your DCS before follow on tests.

 

I usually run the AI tests first as they are quicker to get a larger sample size.  I am doing the player ones now.  Though I can't seem to get tracks activated again despite removing the disable_write_track=true from the autoexec.cfg.  Edit:  I see I can manually save the track if I exit instead of hot key restarting.

Edited December 19, 2022 by Rhayvn

[Rhayvn]

19 minutes ago, MARLAN_ said:

MiG-29A, shooting RS-AA-10c (R-27ER), latest open beta version

I am flying the F-18 as a client vs the AI MiG-29A's though, not doing AI vs. AI tests, so that may be where the discrepancy is originating.

Edit: In both of your Tacviews, the AMRAAM does not appear to be guiding at all, only flying to last known intercept (and surprisingly, doesn't seem to acquire the contact at all even when it looks like it's in the seeker FOV, and the contact isn't maneuvering) it may be worth it to try repairing your DCS before follow on tests.

 

Also, regardless of the initial shot, the follow up shots show the issue rather well.  Your thoughts on those?

[MARLAN_]

13 minutes ago, Rhayvn said:

Also, regardless of the initial shot, the follow up shots show the issue rather well.  Your thoughts on those?

Here's my first test from today, one missile is kinetically defeated, the reattack shot connects.

Tacview-20221219-134216-DCS-Test_BVR.zip.acmi

[Default774]

5 minutes ago, MARLAN_ said:

MiG-29A, shooting RS-AA-10c (R-27ER), latest open beta version

I am flying the F-18 as a client vs the AI MiG-29A's though, not doing AI vs. AI tests, so that may be where the discrepancy is originating.

Edit: In both of your Tacviews, the AMRAAM does not appear to be guiding at all, only flying to last known intercept (and surprisingly, doesn't seem to acquire the contact at all even when it looks like it's in the seeker FOV, and the contact isn't maneuvering) it may be worth it to try repairing your DCS before follow on tests.

 

I'm pretty convinced this is more of an AI issue than an AIM-120 issue. The current BVR AI is superhumanly able and completely incompetent at the same time. it will gladly snap into the notch perfectly, and then refuse to defend an incoming missile until it comes within 8nm of them. 

I've done some testing of player vs AI, and employing 120s in anything but slow max range launches results in a near 100% Pk on AI, mostly because the AI is incompetent(ai_player.zip). Miss in 1202.trk is due to my STT lock coasting and providing false datalink updates to the missile, taking control of the track and demoting to SAM track makes the missile hit.

ag1-3.trk(ai.zip) show a basic test setup. MiG-29A(2-4-0) vs F/A-18C(0-2-6) @ 35K M1.0, Ace skill level, Random launch range, 50nm separation. Most if not all F18 deaths can be blamed on the AI being completely incompetent at BVR and either not firing a missile or not defending. All in all the result is mostly the same in all runs. The F-18 usually wins unless the AI decides to sleep at the wheel and fly into the enemy missile.

Most of the misses in your tacviews seem to be caused by:

• The AI losing track before the missile acquires the target
• The AI having a superhuman notching ability
• The AIs sixth sense of the incoming missile's energy state and defending just barely enough to make the missile miss by a couple hundred feet

 

ai.zip ai_player.zip

[Aquorys]

On 12/9/2022 at 9:44 AM, Маэстро said:

[...] Missile flies to the PIP(predicted intercept point) and this PIP moves with very high speed in such type of maneuver. [...]

None of that is relevant in the final phase of an AMRAAM (in the real world), because the target proximity is ascertained by the target detection device, not the guidance section.

[Rhayvn]

24 minutes ago, MARLAN_ said:

Here's my first test from today, one missile is kinetically defeated, the reattack shot connects.

Tacview-20221219-134216-DCS-Test_BVR.zip.acmi 283.76 kB · 1 download

I am still getting completely different results.  Especially when there is any jamming from the target.  29As are at the bottom of the more modern BVR threats.  Try this against a 29S/27/30/33 or if you really want to see the difference, against a JF-17 with SD-10s. 

That said, here are two tracks with 29As and two with SU-27s.  Watch what the 120s do in most (Though not all) terminal phases.  They are not flying lead and are overreacting to target pitch changes.

Tacview-P18-27-2.zip.acmi Tacview-P18-27-1.zip.acmi Tacview-P18-29-2.zip.acmi Tacview-P18-29-1.zip.acmi P18-27-1.trk P18-27-2.trk P18-29-1.trk

[Rhayvn]

Default774, if by "Random launch range" you mean the line in the ME that is "AA Missile Attack Ranges=RANDOM BETWEEN MAX RANGE AND NO ESCAPE ZONE LAUNCH", I always remove that.  For the most part, it just prevents the AI from engaging before it has to defend.  Resulting in them not doing A-pole or descending after their own launch.  It has the base effect of making nearly every AI target much easier to kill.  Especially with the inherent range advantage the 120 already enjoys.

Also, do you have the jamming enabled for the 29S.  You were locking it in a 16 at ~30 miles.  Usually that's not possible if jamming is enabled.  Also, note that it never fired on you.  Generally, the 29S will fire a 27ER or, more rarely, a R-77 before you can burn through the jamming.

Edit:  One additional track file the previous post didn't have the room to add.

P18-29-2.trk

Edited December 19, 2022 by Rhayvn

[Aquorys]

On 12/18/2022 at 12:41 AM, Rhayvn said:

 When testing as a player against AI Mirage 2000 and Mig-29S, I managed to get the hit rate to 30% against the mirages but under 20% against the Mig-29S. 

Not.  One.  Single.  Initial.  Shot.  Hit.  Over 10 tests with player against AI.

This is with a 50 mile initial setup and most engagements starting at 25-35K for both aircraft at high speed

The AI still does very weird things for sure, especially Ace AI, but I got much better results when I tried to reproduce your scenario. AI set to "Launch by target threat estimate" though. However, that would typically make it more of a threat, not less.

50 nm initial, FL300, M1.15.

#1 Target M2000C, it fired first, AMRAAM launch at 25 nm, hit
#2 Target M2000C, I fired first, AMRAAM launch at 35 nm, hit
#3 Target M2000C, I fired first, ARMAAM launch at 29 nm, hit

#4 Target Mig-29S, I fired first, AMRAAM launch at 29 nm, hit
#5 Target MiG-29S, I fired first, AMRAAM launch at 35 nm, hit
#6 Target MiG-29S, I fired first, AMRAAM launch at 34 nm, miss, 2nd launch high to low, missed due to terminal guidance failure, 3rd launch high to low, hit, but I ate an R-77 that was fired by the MiG-29S 45 degrees off-bore-sight 6 seconds after its radar could technically have the ability to scan the area of space I was in. That's a bit fast for a MiG-29S pilot to defend against an AMRAAM, then readjust the radar, wait for the radar to find a target, select the target and fire a missile. I had previously evaded two R-27ERs.
#7 Target MiG-29S, I fired first, AMRAAM launch at 33 nm, missed because it could not find the target despite it being ~20 degrees off-bore at ~5 nm and much less before that. I defended against an R-27ER, the rest turned into a mess of flying projectiles, with the remaining R-27ER, 2x R-77, 2x R-73s, 3 AMRAAMs and 2x AIM-9X fired, none of which did anything (mine not tracking, the opponent's kinetically defeated), and it ended up becoming a dogfight in which I had the MiG-29S defensive until its pilot ejected, probably due to running out of fuel.
#8 Target MiG-29S, I fired first, AMRAAM launch at 31 nm, R-27ER inbound at 28 nm, another AMRAAM launch at 23 nm, the first AMRAAM failed in the same way as the one in engagement #7, the second one hit.

My initial launches were around M1.1 to M1.3. The M2000C went crazy fast, launching at M2.00 in the first engagement. Good for missile performance, but bad for maneuvering.

I had previously tested against a MiG-29A and got first-launch hits in all 3 engagements.
 

On 12/18/2022 at 12:41 AM, Rhayvn said:

and then dropping towards the deck after defending against initial shots

Not sure about the exact situation, but in most cases, that's a bad idea. Exceptions being mostly radar performance on some aircraft and attempting to merge due to an asymmetric weapons situation.

Edited December 19, 2022 by Aquorys

[Rhayvn]

42 minutes ago, Aquorys said:

The AI still does very weird things for sure, especially Ace AI, but I got much better results when I tried to reproduce your scenario. AI set to "Launch by target threat estimate" though. However, that would typically make it more of a threat, not less.

50 nm initial, FL300, M1.15.

#1 Target M2000C, it fired first, AMRAAM launch at 25 nm, hit
#2 Target M2000C, I fired first, AMRAAM launch at 35 nm, hit
#3 Target M2000C, I fired first, ARMAAM launch at 29 nm, hit

#4 Target Mig-29S, I fired first, AMRAAM launch at 29 nm, hit
#5 Target MiG-29S, I fired first, AMRAAM launch at 35 nm, hit
#6 Target MiG-29S, I fired first, AMRAAM launch at 34 nm, miss, 2nd launch high to low, missed due to terminal guidance failure, 3rd launch high to low, hit, but I ate an R-77 that was fired by the MiG-29S 45 degrees off-bore-sight 6 seconds after its radar could technically have the ability to scan the area of space I was in. That's a bit fast for a MiG-29S pilot to defend against an AMRAAM, then readjust the radar, wait for the radar to find a target, select the target and fire a missile. I had previously evaded two R-27ERs.
#7 Target MiG-29S, I fired first, AMRAAM launch at 33 nm, missed because it could not find the target despite it being ~20 degrees off-bore at ~5 nm and much less before that. I defended against an R-27ER, the rest turned into a mess of flying projectiles, with the remaining R-27ER, 2x R-77, 2x R-73s, 3 AMRAAMs and 2x AIM-9X fired, none of which did anything (mine not tracking, the opponent's kinetically defeated), and it ended up becoming a dogfight in which I had the MiG-29S defensive until its pilot ejected, probably due to running out of fuel.
#8 Target MiG-29S, I fired first, AMRAAM launch at 31 nm, R-27ER inbound at 28 nm, another AMRAAM launch at 23 nm, the first AMRAAM failed in the same way as the one in engagement #7, the second one hit.

My initial launches were around M1.1 to M1.3. The M2000C went crazy fast, launching at M2.00 in the first engagement. Good for missile performance, but bad for maneuvering.

I had previously tested against a MiG-29A and got first-launch hits in all 3 engagements.
 

Not sure about the exact situation, but in most cases, that's a bad idea. Exceptions being mostly radar performance on some aircraft and attempting to merge due to an asymmetric weapons situation.

 

Are your targets jamming?  You generally can't lock up a jamming target at those ranges unless they launch, since their signature increases at that point.
You are seeing the issue in more detail on the ones where your first shot does not hit.  When the target is low and notching, the problem is most prevalent.
This is different than pre-patch.

"AI set to "Launch by target threat estimate" though. However, that would typically make it more of a threat, not less."  Not really.  It still means they hold shots longer in some cases rather than always making use of their range.  I can mean it would be harder for you to defeat their missiles kinetically in rare cases where they drive closer to no escape range or have inferior weapons.  But those aren't the case when you are using 120s.

"Not sure about the exact situation, but in most cases, that's a bad idea. Exceptions being mostly radar performance on some aircraft and attempting to merge due to an asymmetric weapons situation."  Descent is to counter incoming missiles.  Even with descending, you will often have a fairly narrow window between an ER hitting you and your missile going active.  The AI always does this, so the AI vs. AI tests, it wasn't something that could be controlled.

If you have tacview files, I would be interested in viewing them.

[Rhayvn]

I think Default has an excellent point here too in that it is mostly an AI problem.  I do not see the same problem at lower AI skill levels.  Almost entirely at Ace level with some Veteran in higher tier ACs.  Also, many aircraft don't generate the issue.  Mig-23s, for instance, don't manifest the terminal guidance issues.

But, a high performance AC with high level AI can expose a problem with the missile.  Note that the changes in the last patch were based on a specific thing players were doing.  If the fix for that enabled the AI to do something similar that is outside of the normal effectiveness for missile defense, then that needs to be fixed too.