reported STT track memory F-15 incorrect and barely functional

[KenobiOrder]

The document quoted in this post IS NOT ITAR Controlled or Classified. It contains no distribution notice etc. It is from 1979.

 

source: https://www.scribd.com/doc/297530454/TO-1F-15C-34-1-1-Nonnuclear-Weapon-Delivery-Manual-Air-to-Air-USAF-Series-F-15C-and-F-15D-Aircraft-Change-6-15-Mar-1981\

 

 

 

According to the F-15 -34 that can be found publicly online from 1979, the F-15 track memory in STT works as follows:

 

At ranges over 10 miles:

 

-If signal is lost for 3 seconds, the radar goes into memory.

-It stays in memory for 3 more seconds

-If it does not find the target in this time, it starts a 6 bar +-3 degree scan (HIGH/MED PRF) at the targets extrapolated position for six seconds, and if it finds the target it will continue to track as normal

 

Under 10nm:

 

MEM is not displayed, instead after a 3 second loss the radar goes into the acquisition scan immediately in Medium PRF only for 4.5 seconds, only returning to search if it cant find the target.

 

Below 5nm: The radar enters a 3 second re-acquisition sequence following a 1.5 second signal loss.

 

 

This should be corrected, especially since this is from a 1979 Eagle and we have a more modern eagle since we have TWS and Aim-120 etc. Furthermore, while FC3 is supposed to be feature complete, this is not a request for a new feature. The game already has a track memory function in STT but it does not work properly, so this is merely a request for the function to be fixed. I think this is no different than the request for the radar range to be fixed which ED has already approved and we are just waiting for.

 

In the game right now regardless of range, the radar immediately goes to MEM upon signal loss the instant it hits the STT notch speed of 50knots. There is no 3 second wait before entering MEM, which reduces the time the target must be in the notch by half. Additionally, there is no automatic requisition at the extrapolated targets position. In total, for a target to notch the radar in STT, it would have to remain out of the radars view for 12 seconds, not the 3 we have in game right now.

 

I have enclosed a track that demonstrates the current behavior is not correct.

 

F-15 STT MEMORY.trk

Edited September 19, 2021 by KenobiOrder

[SgtPappy]

This would be awesome to have, as many of my lost targets are right there, near the merge when I've seen cold bandits basically just turn around and instantly break lock during the 0.5 seconds they're in the notch. At such close ranges, the range gating and small res cell should prevent this but anyone who flies the F-15 knows that any bandit can break lock at any range without chaff just by being 90 deg. aspect momentarily. No chaff needed, just turn around. I've died a couple times this way as my boresight has been just a bit off. This tactic should be suicide for a bandit at this range, but it ends up as very deadly (at least against my flying which needs improvement). As mentioned I've noticed at longer ranges (> 5nm or something), the radar will try to extrapolate the target's heading via MEM and if the target comes out of the notch enough, it will reacquire but at short ranges, I've never seen this happen.

 

The high level of automation necessary to make the Eagle a deadly WVR is missing from FC3 and though I understand that it's not a FF aircraft, it would be good to use existing architecture to have some of these function in the background perhaps. 

Edited September 20, 2021 by SgtPappy

[KenobiOrder]

@BIGNEWY

[BlackPixxel]

I think it depends whether you lock them with BVR radar mode or close combat radar modes.

 

Same with the Flanker. Radar extrapolation is only there in BVR mode. Vertical scan, boresight and helmet lock do not extrapolate, instant loss of lock.

 

Obviously a FC3 thing.

[Exorcet]

It would be nice to have this fixed, but this is a FC3 plane. The whole point is that not everything is 100% accurate. If we brought up every inaccuracy or omission with the F-15, we'd end up with a DCS module (I would not complain, but I don't think it's likely to happen this way).

[KenobiOrder]

1 hour ago, Exorcet said:

It would be nice to have this fixed, but this is a FC3 plane. The whole point is that not everything is 100% accurate. If we brought up every inaccuracy or omission with the F-15, we'd end up with a DCS module (I would not complain, but I don't think it's likely to happen this way).

Sure, but this is a core functionality like the radar range changes were getting. And ED has made adjustments to FC3 over time, such as the Flankers data link or the fact that aspect was not always factored into radar range.

 

This is already a feature of the module, it's just way to short to be useful. This is a core function that gives a reason to use STT and prevents casual notching.

[okopanja]

4 minutes ago, KenobiOrder said:

Sure, but this is a core functionality like the radar range changes were getting. And ED has made adjustments to FC3 over time, such as the Flankers data link or the fact that aspect was not always factored into radar range.

Not completely, DL with wingman is missing in MP. In addition ECM is lobotamized with 60s limitation of 15 seconds delay.

 

Lets face it: FC3 is a legacy product in maintenance phase, this means they will not implement feature requests.

 

Today the focus is on single modules(aircrafts, maps, weapon packs) that bring the fresh money. If we see FF among FC3 it will likely be a red aircraft, due to saturation of offer on blue side.

 

[KenobiOrder]

8 minutes ago, okopanja said:

this means they will not implement feature requests.

It is not a new feature, like the radar range changes were are already getting. At the simplest, ED could simply increase the timeout to 12 seconds.

[BlackPixxel]

8 hours ago, KenobiOrder said:

It is not a new feature, like the radar range changes were are already getting. At the simplest, ED could simply increase the timeout to 12 seconds.

 

That would be a very bad approach, because with the current mechanic there is a 100% success of extrapolation mode if the target leaves the notch.

 

Alot can happen within 12 s, and the target might not be anymore where the radar expects it. The search of the radar needs to be modelled which takes time.

[okopanja]

7 hours ago, KenobiOrder said:

It is not a new feature, like the radar range changes were are already getting. At the simplest, ED could simply increase the timeout to 12 seconds.

You are probably right about this from user perspective. It sounds easy, but I do not think this is that easy. I assume they did not have the document you provided at the time they modeled the F-15c radar in BVR in the first place.  Fast-forward today, from their point of view this is actually a new state that needs further modeling. This is the reason I used the word feature.

 

A feature of the software product called FC3, released 2013, 8 years ago. I bought it since it sounded as a great deal: you get A-10, Su-25, Su-27 and F-15 in the package. I started with Su-27 since this is what I was used to 20 years ago with their even older products, with idea to go with F-15 which I considered as more sophisticated. In MP I quickly learned both of these birds are not air superiority fighters anymore: F-15, having the great AIM-120c but with missing a datalink blind as a mole, Su-27 with half-functional datalink and rather unusable medium range missiles. Su-27 had the option to join A-10 and Su-25 and find itself a niche where it can still be successful thanks to DL and EOS. The F-15 can still fly fast and high as an Eagle and hope for the best while flying above the small pond full of crocodiles. With so many RED flying F-16 (which IMHO is totally crazy), it is rather difficult to use its SA of threats.

 

The reason these aircraft get so little love is that major part of development of that project ended in 2013, with maybe some maintenance period of 4-5 years, which is typically how long you can expect a software product to be maintained, unless there is a new release. In this case new release did not occur, instead the company started creating the different eras/modules/maps/weapon packs. This means that most of the company's talent moves to another project with only skeleton crew (typically a single integrator/developer + manager in order to minimize the costs) remains to run the tests needed for the integration of these modules. From personal insight on how things work in software companies, I can tell you the skeleton crew is on average limited in skills (or decides to act like that after their next performance interview), and is increasingly defensive when it comes to the changes.

 

They kept the old product in the game for 2 reasons:

- cache cow with almost no investment, but with small steady income as entry level title.

- these aircraft are still needed in the game, after all all those new customers need intelligent drones out there.

 

I do not see this changing unless ED decides to create FF modules of FC3 aircraft. In this case they already did something similar with A-10 and even offered the upgrade. Bare in mind that when they did this they likely did not earn the same amount of money as for the new module, so likely their management will oppose upgrade path in future. However with all new modules being announced in different eras I do not think any shift in behavior will occur anytime soon.

 

IMHO: they will not fix anything short of causing the issue with active modules. (I wish to be proved otherwise for both F-15 and Su-27).

Personally, I will wait for discounts (black Friday, Christmas) to be offered on some of the western and eastern aircraft that are FF.

[draconus]

4 hours ago, okopanja said:

They kept the old product in the game for 2 reasons: [...]

You forgot that none of the FC3 aircraft are available otherwise and they are very popular. A-10A is close but it is not the same as A-10C.

They don't need much "love" other than maintenance becuase they are considered finished product unless some new tech is required to keep them up to date and working.

[okopanja]

2 hours ago, draconus said:

You forgot that none of the FC3 aircraft are available otherwise and they are very popular. A-10A is close but it is not the same as A-10C.

They don't need much "love" other than maintenance becuase they are considered finished product unless some new tech is required to keep them up to date and working.

Well, I tried to explain software development and lifecycle of the software products, and why introducing an new requirements is not seen as a bug, but rather as a feature request to a legacy product.

 

I must say I can not fully agree that they are popular now, it depends if you measure popularity through sales or by usage on servers. The first one would simple indicate that module is popular as a selling item, but given the limitations I would also assume that what follows is the purchasing of more capable module (ED likes this). On Growling Sidewinder server, I would say they are way down when it comes to popularity. On blue most popular are F-18 (around 50%), F-16 (20-40%). On red side de-facto leader of the pack is F-16, followed by Flanker, JF-17 and Mig-21.

[KenobiOrder]

I don't think this would be much hassle to implement, and its not a new feature really.

 

Delaying the MEM to start after 3 seconds and then adding 3 more seconds on top of that is in effect merely a scalar adjustment to existing modeling. During this period the radar would require if the target appears inside the extrapolated area inside whatever gates the memory allows for.

 

The 6 second requisition scan is somewhat different but also very simple, especially I would think for FC3 radar model. You simply draw a line to the extrapolated target position and lock onto anything that falls within a beam width defined by a 6 bar 3 degree raster, since at that point the purpose of this scan is to capture  a target that is along the extrapolated position but might have move in altitude or range to be outside the previous tracking gates.

Edited September 22, 2021 by KenobiOrder

[nighthawk2174]

Would love to see this fixed like the radar range.  If that's getting fixed I see no reason this shouldn't as well.

[GGTharos]

In the F-15 with sparrow in flight signal loss should dump the system into FLOOD.

[KenobiOrder]

I believe flood only activates if the target is in the flood fov and range.

 

In any case not of much use if using a 120 or just trying to keep a lock for observation or tracking a longer range shot.

 

But it would be nice to have the auto flood function as well.

[GGTharos]

I agree, likewise TWS should present actual tracks but then TWS would need to have its actual modes implemented.

[okopanja]

14 hours ago, KenobiOrder said:

Delaying the MEM to start after 3 seconds and then adding 3 more seconds on top of that is in effect merely a scalar adjustment to existing modeling. During this period the radar would require if the target appears inside the extrapolated area inside whatever gates the memory allows for.

I read the attached document and I am wondering if you will get much out of this change when dealing with the notch.

Consider these conditions:

- you are 40km way of the target,

- target flies 1000 km/h which enters the notch. It travels roughly 277,78 m/s.

- it appears that beam width of F-15 is 2,5 degree. (this is based on the diagram of the scanning pattern).

 

Within initial 3 seconds the target travels 833,34m , meanwhile your radar attempts at tracking as before based on what occurred, with the beam resulting in ~ 872m wide cone at that distance: so far so good you should receive your reflection. In case it continued to notch filter will drop the target and you will not see it (meaning you progress to state 2). If the the target changed the aspect, depending on the change you may get reflection in which case the filter will not drop it: you kept your lock. If the target made more radical move, it exited the cone and therefore you got no reflections, while meanwhile for another 3 seconds your radar still continues to search linearly (already here your chance at retaining the lock drops significantly). In third state for 6 seconds you the radar switch to mini-raster, which expands the search area (horizontally 3 deg, calculated at 4192m). The target meanwhile the target traveled 1666,68m from the point where you lost signal. The pattern may hit it, but its more about weaker probabilities at this point. This is why it transitions to state 4: long range search.

 

Please note that if targets goes in higher speeds it travel longer trajectory, while your beam still has the same width.

 

IMHO: modeling of the above may not be trivial as you believe. I mean you can roll dices (which nobody in forum likes as I gather)

for each state based on calculated probabilities, or make complete calculation of radar cone coverage.

 

@GGTharossince you are for ages here ( ), do you remember if the radar modeling details in DCS were ever published?

[draconus]

3 hours ago, okopanja said:

...while meanwhile for another 3 seconds your radar still continues to search linearly...

Since it's not clear from your description I'd just remind that radar does not search around the point it lost the contact. It searches where it thinks it would be should the target did not change the attitude and speed.

[GGTharos]

5 hours ago, okopanja said:

IMHO: modeling of the above may not be trivial as you believe. I mean you can roll dices (which nobody in forum likes as I gather)

for each state based on calculated probabilities, or make complete calculation of radar cone coverage.

 

We don't know what the radar would do exactly but you can sort of guess and there'll be no need to roll dice:  If you get a reflection within the specified search area (raster size and range gate) you lock it

 

5 hours ago, okopanja said:

@GGTharossince you are for ages here ( ), do you remember if the radar modeling details in DCS were ever published?

They were not but I can tell you two things about them:  a) they basically follow the radar equation and b) for FC3 at least they operate on the game's object ID, not on the basis of reflection.

[okopanja]

5 hours ago, draconus said:

Since it's not clear from your description I'd just remind that radar does not search around the point it lost the contact. It searches where it thinks it would be should the target did not change the attitude and speed.

I meant precisely that.

4 hours ago, GGTharos said:

We don't know what the radar would do exactly but you can sort of guess and there'll be no need to roll dice:  If you get a reflection within the specified search area (raster size and range gate) you lock it

First one is more in line on how a real radar works and whey the whole state machine is executed like this. If we are talking about probability from observer with no prior knowledge on position (like in case 2), the described algorithm bets on the target that more or less moves as before. When this is not the case probability of detecting the signal again drops.

 

My point is that radar may loose signal for many different reasons and not only the notch. E.g. rapid altitude change (after all it was more important to keep much higher 6 bar scan than reduced raster scan of 3 degree), terrain, hardware deteriorating. 

 

The radar in question had low reliability of just 15 hours of MTBF (it must have been really expensive to operate, until they managed to expand that to 120 hours with v1). That is so low that programmer would have to assume that the hardware will not be so reliable.

 

Radar is more likely to keep lock with the target that stayed in the notch longer, provided that it changes expect within interval of 12 seconds. Any other target that moved away from these two search zone (extrapolated 6 second line and 6 second pattern) would simply break the lock, and the longer you wait the worse for you.

 

The main reason you get MEM is to prepare the pilot for the fact that lock is lost and that limited attempt of recovery is in progress. 4.5 seconds may be actually a favor to the DCS pilot to start LRS sooner.

[KenobiOrder]

10 hours ago, okopanja said:

Within initial 3 seconds the target travels 833,34m , meanwhile your radar attempts at tracking as before based on what occurred, with the beam resulting in ~ 872m wide cone at that distance: so far so good you should receive your reflection. In case it continued to notch filter will drop the target and you will not see it (meaning you progress to state 2). If the the target changed the aspect, depending on the change you may get reflection in which case the filter will not drop it: you kept your lock. If the target made more radical move, it exited the cone and therefore you got no reflections, while meanwhile for another 3 seconds your radar still continues to search linearly (already here your chance at retaining the lock drops significantly). In third state for 6 seconds you the radar switch to mini-raster, which expands the search area (horizontally 3 deg, calculated at 4192m). The target meanwhile the target traveled 1666,68m from the point where you lost signal. The pattern may hit it, but its more about weaker probabilities at this point. This is why it transitions to state 4: long range search.

so in 3 second the target movied 833m, then another 833m in the later 3 during M, for 1666m. As your said. Then in another 6 seconds it moves another 1600m. Total is about 3300, so our 4100 meter scan is more than big enough to see the target.

 

Moreover, the target must be moving in angle between the targets extrapolated position and the range of angles that would stay within the Doppler gate for a range of speeds. If the target continues its turn and goes nose cold, it will be rapidly required probably just by memory alone. To stay in the notch, the target must fly a close to a perfect circle because it has to keep a radial velocity that is within the doppler gate. It it turned 90 degrees to the radar and stayed there, it would simply fly out of the notch on its own, it has to keep adjusting its turn to stay in the beam. On top of that, the faster it goes the harder it is to stay in the notch, because this will increase the radial velocity for a given angle to the radar.

 

The point of the raster scan after 6 seconds is that the target might have moved in altitude or in range outside of the target tracking gate. Regarding the range gate @GGTharos I dont think there would be much of one if at all, since at this stage a target might come out the beam (say at 9 seconds) but be further in range than the original tracking loop but still within the line of sight of the raster. There is a risk you might lock onto the wrong target, but its a fairly slim one and worth the risk in this situation.

[okopanja]

41 minutes ago, KenobiOrder said:

so in 3 second the target movied 833m, then another 833m in the later 3 during M, for 1666m. As your said. Then in another 6 seconds it moves another 1600m. Total is about 3300, so our 4100 meter scan is more than big enough to see the target.

Mind that the actual speed might be much higher, then these numbers increase as well.

43 minutes ago, KenobiOrder said:

To stay in the notch, the target must fly a close to a perfect circle because it has to keep a radial velocity that is within the doppler gate.

The volume of the space where this target can move without being detected can be calculated.

[KenobiOrder]

Has anyone from ED looked at this?

[nighthawk2174]

Bump