|
|
Post by jtyotjotjipaefvj on Feb 1, 2018 16:31:31 GMT
Everyone seems to be under the impression that laser wobble simulates some inaccuracy in ship sensors or turret actuators. This is however not the case. The wobble is used to simulate wider beam sizes. Since the game apparently can't handle damaging an area of armor intersected by the beam, it instead uses Monte Carlo integration ( wiki link) to approximate the wider beam. Instead of dealing damage over an area every frame, the game picks a single random point inside the spot and damages that. When averaged over hundreds of frames, the end result is identical, and you end up with an even damage distribution inside the spot area. You can verify this for yourself by seeing how laser wobble varies with the M² factor of your laser when range stays constant. I ran a quick test myself, with the results below in spoilers. The tests were done with 1 Mm range and a laser setup that's otherwise identical, but with varying M² factor of the laser. M² of 3, wobble size roughly 4 meters:  M² of 100, wobble hits so hard the lasers start missing the ship completely:  If someone who knows how lasers work can verify that these wobble sizes match the actual spot size of the laser I used, we could be certain that the wobble simulates only the spot size and no other inaccuracies of the system. Although since railguns don't suffer from any wobble, I would imagine that spot size is the only cause for laser wobble. I used a modified 100 MW vanilla laser, meaning it has a wavelength of 395 nm, and an aperture radius of 40 cm. Hopefully this also clears up a lot of the confusion regarding laser wobble, as many people seem to think it's unrealistic and lasers should be even more effective than they already are. There's of course still the matter of CDE limiting the M² factor to a minimum of 3 instead of the physical limit of 1, but I don't really know enough to say anything one way or another. |
|
|
|
Post by Kerr on Feb 1, 2018 17:25:37 GMT
Everyone seems to be under the impression that laser wobble simulates some inaccuracy in ship sensors or turret actuators. This is however not the case. The wobble is used to simulate wider beam sizes. Since the game apparently can't handle damaging an area of armor intersected by the beam, it instead uses Monte Carlo integration ( wiki link) to approximate the wider beam. Instead of dealing damage over an area every frame, the game picks a single random point inside the spot and damages that. When averaged over hundreds of frames, the end result is identical, and you end up with an even damage distribution inside the spot area. You can verify this for yourself by seeing how laser wobble varies with the M² factor of your laser when range stays constant. I ran a quick test myself, with the results below in spoilers. The tests were done with 1 Mm range and a laser setup that's otherwise identical, but with varying M² factor of the laser. M² of 3, wobble size roughly 4 meters: M² of 100, wobble hits so hard the lasers start missing the ship completely: If someone who knows how lasers work can verify that these wobble sizes match the actual spot size of the laser I used, we could be certain that the wobble simulates only the spot size and no other inaccuracies of the system. Although since railguns don't suffer from any wobble, I would imagine that spot size is the only cause for laser wobble. I used a modified 100 MW vanilla laser, meaning it has a wavelength of 395 nm, and an aperture radius of 40 cm. Hopefully this also clears up a lot of the confusion regarding laser wobble, as many people seem to think it's unrealistic and lasers should be even more effective than they already are. There's of course still the matter of CDE limiting the M² factor to a minimum of 3 instead of the physical limit of 1, but I don't really know enough to say anything one way or another. A 400mm lens, 395nm M²=3 laser achieves a spot size 7.23m in diameter at a distance of thousand kilometers. There you have it, if M² doubles the spot size will increase it's radius/diameter by two and increase it's area by 4x. |
|
|
|
Post by Rocket Witch on Feb 1, 2018 17:33:27 GMT
The 1.0.8 patch notes explicitly say "added targeting sensor inaccuracy to all weapons fire based on sensor diffraction limits. Has little effect, except at extreme ranges (hundreds of km)". Monte Carlo integration for lasers existed before wobble was added. Where do you get the impression that railguns don't wobble?
|
|
|
|
Post by jtyotjotjipaefvj on Feb 1, 2018 18:36:28 GMT
The sensor wobble seems to be fairly small at 1 Mm, too small for my ghetto test to discern. If it does exist, it's less than 1 meter per Mm of distance. I tested four different lasers with 40 cm aperture, 395 nm wavelength and varying M², as well as two different 50 km/s railguns with varying spread size but otherwise identical performance. The measured hit pattern sizes as well as ones predicted either using the displayed spread angle, or in the case of lasers, computed from the displayed beam power and intensity, are shown in the table below:  Below are in-game screenshots showing how I measured the above results, spoilered to avoid bloat. Ship used as target:  Overview of measurement, range is 1 Mm:  spot of 3 M² laser:  7.35 M²:  15.2 M²:  34.1 M²: (note the extra spot count)  Railgun target:  0.001 degree spread pattern:  0.0001 degree pattern (side view):  0.0001 degree pattern (head-on view):  An interesting find in the 34.1 M² laser: it looks like the game uses more than one MC sample per frame when the spot size gets very large. I can see around a 100 spots coming out from 18 laser emitters. |
|
|
|
Post by captinjoehenry on Feb 1, 2018 19:35:29 GMT
I can confirm turret wobble at extreme range with my 1Mm railgun. It remains laser accurate at Mm ranges but the impact point slowly wobbles a notable amount. I can try and get some images in a bit.
|
|
|
|
Post by jtyotjotjipaefvj on Feb 1, 2018 20:01:27 GMT
It's now clear I should have tested this with only a single laser instead of 18, since each weapon is in a different phase of the sensor wobble pattern. With a single laser you can easily see the difference between sensor wobble and MC integration of spot size, as sensor wobble varies much more slowly. With a single AE laser, you can see that at 1 Mm, the spot size is approximately 10 cm, while sensor wobble has a radius of approximately 1 to 2 meters in radius.
This leads to an interesting implication: if you want to be able to take down lightly armored micromissiles, minimizing spot size actually increases your time to kill. If you have a 100 MW laser with a 2-meter spot size, you will continuously hit a missile with 10 cm diameter with some laser power, whereas lasers like the AE 100 MW one will only rarely hit due to sensor inaccuracy. Since micromissiles usually have practically no armor to speak of, this strategy should prove quite effective against them. This also has the added benefit that larger spot sizes translate to smaller laser apertures, which means your PD laser turret will be far lighter than one intended for anti-capital use. I'll have to test this out to see if it actually works.
|
|
|
|
Post by jtyotjotjipaefvj on Feb 1, 2018 20:35:52 GMT
I just tested my previous idea and boy does it work. A 1 GW laser with ~1 MW/m² intensity at 1 Mm, translating to roughly 3.5 m spot radius is ridiculously effective at taking down micromissiles with 3 mm PE armor. A deep fryer takes 2 minutes 30 seconds to burn through a salvo of 20 micromissiles, and gets the first kill at around 200 km range when starting the approach from 1 Mm. Whereas my PD beamer kills all 20 missiles at 1 Mm in less than one second. This seems to be a strategy that will completely invalidate most current micromissile designs, as you really can't afford to put a relevant amount of armor on them.
I ran another test with a 11 cm thick PE cap inserted to the end of the missile, with the body left unarmored. In that case the Deep Fryer killed the micromissiles only 10 seconds slower than my PD laser - 2:30 versus 2:20 for the whole 20-missile salvo. Still though, it's a faster kill rate with just over half the energy expenditure, and with a laser that only weighs 400 kg opposed to the 33 ton load of lasers the Deep Fryer carries. It would appear that spot sizes under 1 meter in diameter might not make a lot of sense due to the sensor wobble, at least not when dealing with small targets. It might be a different story with capital ships, though even there sensor wobble limits effective spot size to one meter at 1 Mm range.
Edit: the <1 second kill time was due to something else, I can't replicate it on repeat tests. I'll have to see further if there's any sense in this idea.
|
|
|
|
Post by treptoplax on Feb 1, 2018 21:49:44 GMT
THIS IS RELEVANT TO MY INTERESTS (posting to follow)
It's not just micro missiles, but drones. I've found that 20cm diameter railgun drones (it's not a very good railgun at that size, but decent) are very effective against the standard AE library lasers. Do I understand correctly that the less focussed laser also has a smaller mirror, which also reduces the range at which that attacking system is effective?
|
|
|
|
Post by jtyotjotjipaefvj on Feb 1, 2018 22:15:32 GMT
THIS IS RELEVANT TO MY INTERESTS (posting to follow) It's not just micro missiles, but drones. I've found that 20cm diameter railgun drones (it's not a very good railgun at that size, but decent) are very effective against the standard AE library lasers. Do I understand correctly that the less focussed laser also has a smaller mirror, which also reduces the range at which that attacking system is effective? It's more that it makes no sense to make your spot size smaller than the sensor wobble radius, since sensor wobble effectively limits your minimum spot size anyway. No use spending weight to go under it when you can stay at the hard limit with a lighter design. Additionally, if your spot radius is larger than wobble radius, you can ensure a 100% hit rate when facing narrow diameter targets, as the larger spot size means you will hit regardless of the wobble. When you have power to spare, this proves to be an effective strategy. I managed to optimize my PD laser further. Against a salvo of 100 micromissiles with 1mm + 1.5mm PE armor in two layers, the Deep Fryer will only manage to take down 26 missiles before it dies. With 6x333 MW PD lasers with ~2m spot size, the laser star will destroy all missiles before they get closer than 150 km range. At close range, the kill rate of missiles is slightly lower with the unfocused lasers, but the unfocused PD has already killed 41 missiles at 500 km range, whereas the anti-capital lasers only get their first kill at around 480 km. Sadly it's not (yet at least) as good as I'd hoped, but it's still already getting 4 times more kills than an unmodified Deep Fryer. The PD lasers are also far more effective than the 50 km/s railguns, which don't have enough time to kill a relevant number of missiles. I'll include videos of both tests, as well as a screenshot of the missile and laser designs in spoilers below. Laser design. Note the weight.  missile design.  Unmodified deep fryer gets splattered by missiles: Edited deep fryer with 6x333MW pd lasers manages to kill all missiles before it too gets killed. With proper engines, it would still have time to avoid the impact though, so it counts as survival. Funnily enough, the test against the Deep Fryer was also the most efficient Deep Fryer kill I've managed, with ~40 kc cost in missiles. I wasn't aiming for that at all, it just happened. |
|
