[DCS BUG] AIM-54

[captain_dalan]

 

That turn seems non-sensical in the context of breaking a radar lock.

 

 

Breaking radar lock may have been successful wish some older radar in this case, but I doubt it would be easy to do it without ECM and/or chaff.

What if a hard break is executed to maneuver the plane out of the missile's radar FoV? I mean, the shot was made sub 10NM, that usually means active off the rail. So self guided. Just speculating here.

 

 

The funny thing is offline I have 0 problems dodging the A or C fired by "ace" pilots. Online its a different story using literally the same tactics/maneuvers.

 

Not funny at all. At least not in my own experiences. AI seldom follows the same rules as human controlled stuff in DCS does. But even if it wasn't for that, i can tell you that even a random (but proficient RIO) makes a hell of difference in DCS. I usually fly MP on dogfights servers only, but on few occasions i've visited massive MP servers. On one such occasion, i just couldn't hit anything. And i mean anything. Mind you, i always support my missiles till some 20NM away. But that day i had 10 launches, all well within desirable parameters and not i single hit. Then this random dude asks if i want him to be my RIO. I politely explain i have no voice com to communicate with him. He says that's ok, i should just fly the right way and he will do the rest. So we take off, i fly to the bulls, he picks up and sorts contacts on radars, launches and bang. One launch one hit. No imagine if we were actually a team. A good RIO can do wonders. With Jester...... dunno. I tired last week, he was all over the place. Switching radar modes, dropping contacts. Hectic and random. Nothing i did mattered. I can't tell if it was a bad night, or has AWG-9 been borked lately.

 

 

Also doesn't say if the 14 fired from low or high or what. That matters. Is that quote all that it says?

 

"This firing occurred on 2 February 1973. The target was a QF-86 (drone version of an old Air Force fighter) cruising at an altitude of 17000ft and a speed of mach 0.7. "Smoke's" airplane was cruising at an altitude of 17000ft and a speed of mach 0.8. The target was acquired and a Phoenix was launched at a range of 10NM. Shortly after the launch the drone's controller rolled the QF-86 into a 5 g descending turn and made a 6 g pullout, at 9000ft. The missile arrived just at the pullout scoring a range hit. The F-14's radar was directed away from the target after initial positioning because this was to be an ACM active launch. In this mode, when the trigger is pulled, the Phoenix's radar comes on immediately and locks on the first target in its FoV......"

 

From a different book though.

Still, not only shows the missile could hit a maneuvering target, it was also tested specifically for one such scenario. And this is the A. So...... yeah, built for bombers.......that pull 6g evasive maneuvers.

[Skysurfer]

What do they say, "you dont know what you don't know". :smilewink:

[FWind]

 

 

What if a hard break is executed to maneuver the plane out of the missile's radar FoV? I mean, the shot was made sub 10NM, that usually means active off the rail. So self guided. Just speculating here.

 

 

 

Not funny at all. At least not in my own experiences. AI seldom follows the same rules as human controlled stuff in DCS does. But even if it wasn't for that, i can tell you that even a random (but proficient RIO) makes a hell of difference in DCS. I usually fly MP on dogfights servers only, but on few occasions i've visited massive MP servers. On one such occasion, i just couldn't hit anything. And i mean anything. Mind you, i always support my missiles till some 20NM away. But that day i had 10 launches, all well within desirable parameters and not i single hit. Then this random dude asks if i want him to be my RIO. I politely explain i have no voice com to communicate with him. He says that's ok, i should just fly the right way and he will do the rest. So we take off, i fly to the bulls, he picks up and sorts contacts on radars, launches and bang. One launch one hit. No imagine if we were actually a team. A good RIO can do wonders. With Jester...... dunno. I tired last week, he was all over the place. Switching radar modes, dropping contacts. Hectic and random. Nothing i did mattered. I can't tell if it was a bad night, or has AWG-9 been borked lately.

 

 

 

"This firing occurred on 2 February 1973. The target was a QF-86 (drone version of an old Air Force fighter) cruising at an altitude of 17000ft and a speed of mach 0.7. "Smoke's" airplane was cruising at an altitude of 17000ft and a speed of mach 0.8. The target was acquired and a Phoenix was launched at a range of 10NM. Shortly after the launch the drone's controller rolled the QF-86 into a 5 g descending turn and made a 6 g pullout, at 9000ft. The missile arrived just at the pullout scoring a range hit. The F-14's radar was directed away from the target after initial positioning because this was to be an ACM active launch. In this mode, when the trigger is pulled, the Phoenix's radar comes on immediately and locks on the first target in its FoV......"

 

From a different book though.

Still, not only shows the missile could hit a maneuvering target, it was also tested specifically for one such scenario. And this is the A. So...... yeah, built for bombers.......that pull 6g evasive maneuvers.

 

7:20

[captain_dalan]

Good finds! We were looking for some AIM-54A specific cases though, hence those book references.

[Harlikwin]

 

 

 

"This firing occurred on 2 February 1973. The target was a QF-86 (drone version of an old Air Force fighter) cruising at an altitude of 17000ft and a speed of mach 0.7. "Smoke's" airplane was cruising at an altitude of 17000ft and a speed of mach 0.8. The target was acquired and a Phoenix was launched at a range of 10NM. Shortly after the launch the drone's controller rolled the QF-86 into a 5 g descending turn and made a 6 g pullout, at 9000ft. The missile arrived just at the pullout scoring a range hit. The F-14's radar was directed away from the target after initial positioning because this was to be an ACM active launch. In this mode, when the trigger is pulled, the Phoenix's radar comes on immediately and locks on the first target in its FoV......"

 

From a different book though.

Still, not only shows the missile could hit a maneuvering target, it was also tested specifically for one such scenario. And this is the A. So...... yeah, built for bombers.......that pull 6g evasive maneuvers.

 

Interesting test case, so from the description, they start Co-alt, and the drone executes a break and dive so a basic evasive maneuver of sorts, and I've never had issues with the kinematic capabilities of the missile. However, that maneuver as described wouldn't have notched the missile IMO, and of course it has a doppler seeker. So missile still tracks because it doesn't filter out the target. Presumably much like the Iranian engagements. But again, exactly how notchable is the missile?

 

Do we know anything of seeker gimbal limits and track/search rates or patterns? The seeker is based on the AN/DPN-53.

[Skysurfer]

Reference the below image. The seeker is also giballed so once the missile has a lock or is looking in the general direction you won't create enough LOS rate change to escape its FOV.

[SgtPappy]

I posted in the other thread (link: https://forums.eagle.ru/forum/englis...bility-of-kill) before I realized there's a discussion on the AIM-54A specifically here. Firing from above, down onto the bandit, the AIM-54 is very easy to notch. However, it becomes pretty god-like when looking up, even with chaff, beaming, etc. so that tactic should be used when attacking enemies. I am however not 100% sure this should be the case. Tacviews attached here for reference.

 

Based on some simple back of the envelope calculations (also posted in the previous thread), I would expect the AIM-54A to be fairly chaff-hungry at the appropriate ranges since the resolution cell is so big and I don't believe such an old missile with mech scan had any ability to beam sharpen. For convenience, here's the calculation and please if anyone has more to add to this or if it's just plain wrong, speak up:

 

The AIM-54 radar diameter is slightly smaller than the missile diameter at about d = 0.380 m. At X-band, let's say carrier frequency, f = 9GHz, then wavelength, λ = 0.033 m the angular resolution would be θ ~= 0.033/0.380 = 0.08766 radians = 5.022 deg. This is pretty big, and so the cross-range of the resolution cell at, say 5nm would be R = θ*5nm = 0.08766*5 = 0.438 nm = 0.813 km which is pretty huge. With such a large cell, I would expect chaff from, say, the standard USAF RR-170 cartridge (3 million dipoles, maybe 25% of which are X-band) would give an RCS of ~80 or more m^2 and that would be enough to pull the RCS centroid away from the aircraft at that range while the defender is beaming, rolling, pulling G's to create more RCS variation and defeat the missile predictive track.

 

The AIM-54 is a very early mech-scan (i.e. no beam shaping) seeker so I find it a bit hard to believe that it won't eat chaff at that range. Not sure if this is the intent of the current missile modeling. But I'm up to be proven wrong, this is just my feeling based on what I know.

AIM-54A tests.zip

[Skysurfer]

I havent found 54's to be harder to notch in a lookup. In DCS this really doesn't seem to matter. In a lookup, getting the right angle might simply be harder. I'll take a look at your tacviews later on.

[draconus]

I havent found 54's to be harder to notch in a lookup.

 

There's no notching in the lookup by definition. You could be doppler filtered but you'd have to be faster than the missile... no chance :)

[Skysurfer]

 

There's no notching in the lookup by definition. You could be doppler filtered but you'd have to be faster than the missile... no chance :)

 

IRL there might be no such thing, in DCS there 100% is. It simply does not matter what is behind you.

 

EDIT: Ok, just watched your tacviews, you are NOT notching or defending correctly in most cases.

[SgtPappy]

 

IRL there might be no such thing, in DCS there 100% is. It simply does not matter what is behind you.

 

EDIT: Ok, just watched your tacviews, you are NOT notching or defending correctly in most cases.

 

Thanks Skysurfer. Do you have any tips? I have been doing this exact maneuver the whole year and it worked every time until the recent API change patch for the AIM-54. It still works vs the S530 and AIM-7.

[Skysurfer]

 

Thanks Skysurfer. Do you have any tips? I have been doing this exact maneuver the whole year and it worked every time until the recent API change patch for the AIM-54. It still works vs the S530 and AIM-7.

 

For a notch in DCS, you need to do two things for it to be effective. Slow down to < M0.4, reduce closure to a minimum (put the missile on your 3-9 line) and preferably drop chaff. See, the way proportional navigation and missile guidence works in DCS, missiles will take lead if a target is flying fast or beaming the missile with a high angular velocity in order to achieve the desired intercept point. This very "lead" the missile takes creates such a geometry that now makes it very hard for the defending aicraft to create the desired 0-closure rate aspect. The slower you are in the notch the less "lead" does the missile pull and the easier it is to achieve the desired effect. Older Fox-1's in DCS will usually easily go for chaff and/or break the lock in a semi-decent notch unless you revert to something like flood mode or pulse etc. Active missiles can re-acquire you at any time if you leave the notch too soon or are simply too fast, creating said offset geometry. It would also be desireable to test this with another human player in a MP environment since the AI doesn't really use the same missiles and code more often than not.

[SgtPappy]

Understood, that sounds very risky. I can see why it's much more preferable to notch per what I've heard from retired fighter pilots - at least against older threat radars. Thanks!

[Noctrach]

 

For a notch in DCS, you need to do two things for it to be effective. Slow down to < M0.4, reduce closure to a minimum (put the missile on your 3-9 line) and preferably drop chaff. See, the way proportional navigation and missile guidence works in DCS, missiles will take lead if a target is flying fast or beaming the missile with a high angular velocity in order to achieve the desired intercept point. This very "lead" the missile takes creates such a geometry that now makes it very hard for the defending aicraft to create the desired 0-closure rate aspect. The slower you are in the notch the less "lead" does the missile pull and the easier it is to achieve the desired effect. Older Fox-1's in DCS will usually easily go for chaff and/or break the lock in a semi-decent notch unless you revert to something like flood mode or pulse etc. Active missiles can re-acquire you at any time if you leave the notch too soon or are simply too fast, creating said offset geometry. It would also be desireable to test this with another human player in a MP environment since the AI doesn't really use the same missiles and code more often than not.

 

This is mostly true, but I'd like to add some extra nuance.

From a combination of LUA file reading and very extensive testing with my old SATAL team: two things need to happen

 

1) Missile needs to be looking below its own horizon (so any pitch down greater than 0 in magnitude will suffice)

2) Closure rate needs to be lower than the LUA defined notch gate. This is unique for each missile and radar.

 

Slowing down makes this easier, but you can do it at mach 2+ if you have a steady hand and good timing.

With the old API the missile would execute a scripted behaviour where it would pull 13G off the target for a second or so, which meant even a 0.1 second notch would spoil the shot.

With the new API, the notch is only valid as long as you keep the closure below the defined number and the missile no longer goes haywire immediately.

[Skysurfer]

 

This is mostly true, but I'd like to add some extra nuance.

From a combination of LUA file reading and very extensive testing with my old SATAL team: two things need to happen

 

1) Missile needs to be looking below its own horizon (so any pitch down greater than 0 in magnitude will suffice)

2) Closure rate needs to be lower than the 's LUA defined notch gate. This is unique for each missile and radar.

 

Slowing down makes this easier, but you can do it at mach 2+ if you have a steady hand and good timing.

With the old API the missile would execute a scripted behaviour where it would pull 13G off the target for a second or so, which meant even a 0.1 second notch would spoil the shot.

With the new API, the notch is only valid as long as you keep the closure below the defined definition and the missile no longer goes haywire immediately.

 

Copy. I will do some tests with the new API missiles and see for a fact if lookup is still a thing.

[SgtPappy]

 

This is mostly true, but I'd like to add some extra nuance.

From a combination of LUA file reading and very extensive testing with my old SATAL team: two things need to happen

 

1) Missile needs to be looking below its own horizon (so any pitch down greater than 0 in magnitude will suffice)

2) Closure rate needs to be lower than the 's LUA defined notch gate. This is unique for each missile and radar.

 

Slowing down makes this easier, but you can do it at mach 2+ if you have a steady hand and good timing.

With the old API the missile would execute a scripted behaviour where it would pull 13G off the target for a second or so, which meant even a 0.1 second notch would spoil the shot.

With the new API, the notch is only valid as long as you keep the closure below the defined definition and the missile no longer goes haywire immediately.

 

Do you think this look down is accurate? I always believed that you really had to be close to clutter but it seems look down at all spoofs the missile even if you're 10,000 ft above the ground or water.

[Noctrach]

 

Do you think this look down is accurate? I always believed that you really had to be close to clutter but it seems look down at all spoofs the missile even if you're 10,000 ft above the ground or water.

 

We know for a fact it isn't tbh :book:

Most, if not all radars since the 60s had some form of range gate processing. The MiG-19 in DCS iirc has a radar that didn't do this and as a result it is completely blind in a look-down scenario under something like 3-4 km altitude.

Similar to doppler gating, you have different "bins" for range, if you detect a target at a slant range "5 to 6 miles" and there's clutter on a slant range "7 to 8 miles" the radar will just ignore that.

To really be able to benefit from background clutter you'd have to get the ground inside the range resolution of the radar you're trying to spoof.

 

I have a tacview somewhere from the old API where I demonstrated someone how to notch a near co-altitude AIM-120C by doing a lazy turn without chaff at 30,000 feet.

I don't think the new API changed this, but who knows what the future holds.

 

Personally I always just share this under "ECM would take care of providing my clutter" in my headcanon.

There's plenty other reasons why a missile would miss in real life, but complete loss of contact through notching at 20,000 feet would not be one of them.

 

Edit: there it is.

Bear in mind this took advantage of the scripted 13G pull behaviour. In the current API the missile would still be notched, but would continue to my last known position and pick me up again 9 times out of 10.

There's no way I'd do able to do this right now, least of all at Mach 1.2 :P

HighAltitudeNotchNoChaff.zip.acmi.zip

[dundun92]

Slightly curious, where are the luas with this stuff? Unless its hidden in the missiles_data.lua somewhere? Or is it in the lua for AI radars?

[SgtPappy]

 

We know for a fact it isn't tbh :book:

Most, if not all radars since the 60s had some form of range gate processing. The MiG-19 in DCS iirc has a radar that didn't do this and as a result it is completely blind in a look-down scenario under something like 3-4 km altitude.

Similar to doppler gating, you have different "bins" for range, if you detect a target at a slant range "5 to 6 miles" and there's clutter on a slant range "7 to 8 miles" the radar will just ignore that.

To really be able to benefit from background clutter you'd have to get the ground inside the range resolution of the radar you're trying to spoof.

 

 

That's exactly what I thought, but I still need to look into a basic calculation for the range binning that the AWG-9 and likely the AIM-54 will use - something tells me it's a lot smaller than 10,000 ft! Are there further API improvements meant to address this?

 

And damn, I don't think I'd ever be brave enough to try what you did in the tacview!

[Noctrach]

@dundun92 I'm pretty sure the notch values are defined in missiles_table.lua, by

seeker = { height_error_max_vel }

in KPH for "half" the gate.

 

This is would result in +-32 knots for the AIM-120B, +-27 knots for the AIM-120C, +-75 knots for the AIM-54 (not sure if this is the HB one, most likely is)

Correlates pretty well with what I've seen in TacView

[dundun92]

Thanks! However, that field isnt on the old API missiles (like R-77). Do you know where it would be for them?

[Noctrach]

@dundun92 That would be missiles_data.lua, P_77, but since most of that is in cyrillic, it ends up gibberish for me.

[dundun92]

In notepad++ it comes out as расчет времени полета, which translated to "flight time calculation", and it shows up for IR missiles as well.

[Noctrach]

Yeah I noticed it was off, the values on the AIM-120 and AIM-54 wouldn't make sense either. So no idea. It might be something related to the Fi/phi values which iirc relate to sensor tracking but I'm just guessing at this point.

[dundun92]

The Fi values and OmViz stuff are related to seeker gimbal limits, LOS rate limits, etc. So yea no clue. Guess its testing time lol.