【全文来自官网https://www.spacex.com/news/2020/04/28/starlink-update】

翻译的肯定不通顺，还请大家见谅。

全文信息量大，首次公开信息较多！

SpaceX正在推出Starlink，以便在全球范围内提供高速、低延迟的网络连接，覆盖光纤互联网覆盖不到的地方。我们还坚信自然夜空对我们所有人的重要性，这就是为什么我们一直在与世界各地的主要天文学家合作，以更好地了解他们的观测和我们可以进行的降低卫星亮度的工程变化的具体细节。我们的目标包括：

1、使卫星在发射后一周内肉眼看不见。我们是通过改变卫星太阳能板的朝向的方式来做到这一点的，这样他们就可以将太阳能板“隐藏”起来。

2、通过使卫星变暗来最小化Starlink对天文学的影响，这样就不会使天文台“心态爆炸”。我们通过在卫星上增加一个可展开的遮罩来阻止阳光照射到航天器最明亮的部分来实现这一目标。下一次发射（Starlink-8，详见https://www.bilibili.com/read/cv5835790）时将会发射一颗带有黑色挡板的Starlink卫星以供测试，到6月份的第9次飞行时，所有的Starlink都会配备这个硬件。另外，关于我们卫星轨道的信息位于space-track.org均有体现，以防意外的闯入天文台的视线。

StarLink轨道

StarLink有三个飞行阶段：

(1)轨道上升

(2)停泊待机轨道，以等待升轨时机(离地380公里)

(3)到达目标轨道(离地550公里)。

在轨道运行期间，卫星使用霍尔推进器在几周内提高高度至工作轨道。有些卫星直接进入目标轨道，而另一些卫星则停留在停泊待机轨道上，以允许卫星进入不同的轨道平面。一旦卫星进入目标轨道，它们就会重新调整姿态和硬件，使天线面向地球，太阳能阵列垂直于太阳，从而能够跟踪太阳以最大限度地发电，并且使卫星尽可能的暗。由于这次机动，卫星变得更暗，因为太阳阵列从地面的能见度大大降低。

目前，419颗卫星（不算DD-A和DD-B）中约有一半在工作轨道（在第一批V0.9的Starlink卫星中，有一颗再入大气层来测试Starlink的自毁效果，测试结果很好，没有留下任何太空垃圾），另一半正在上升或待机停泊。卫星只有一小部分生命将用于轨道提升或维持轨道、进入大气层，并将其生命的绝大部分用于工作中。值得注意的是，在任何特定时间，只有大约300颗卫星将在轨道上升或待机停泊。其余的卫星将在工作轨道上运行。

Starlink卫星

Starlink的飞行高度非常低。也保证了空间安全，并尽量减少了信号延迟。由于低高度，阻力是设计中的一个主要因素。在轨道上升过程中，卫星相对于“风”的横截面积必须最小化，否则拖曳会导致它们从轨道上掉下来。高阻力是一把双刃剑--这意味着卫星的飞行很棘手，但也意味着任何遇到问题的卫星都会在大气层中快速、安全地燃烧。这大大减少了轨道碎片或“空间垃圾”在轨道上的数量。

在升轨中太阳能阵列被平放在卫星前。当卫星上升时，总是保持最小的风阻横截面积，同时保持天线面向地球，足以与地面站保持联系。

当卫星到达550公里的运行轨道时，阻力仍然是一个因素——因此即使完全失控的卫星都会因风阻而迅速降低轨道。但完好的、可以工作的卫星的姿态控制系统能够克服这种阻力。

Starlink的可见度

在日出或日落时，从地面可以看到卫星。这是因为卫星被太阳照亮，但是地面上的人或望远镜却在黑暗中。

这个简单的图表突出了为什么轨道上的卫星比工作轨道上的卫星要亮得多。在轨道上升过程中，当太阳阵列处于平铺状态时，太阳光可以反射出太阳阵列和卫星，并反射到地面。一旦进入工作轨道，底盘部分才能将光线反射到地面，而太阳能板则被“隐藏”了。

卫星物理亮度省略，反正就是Starlink多少都会有影响，后面贴上英语原文

Starlink反光的解决方案

我们已经采取了一种实验和迭代的方法来降低星链卫星的亮度。亮度是一个极难分析的问题，所以我们一直在进行地面和在轨的测试。

例如，今年早些时候，我们发射了一颗“涂黑”的Starlink卫星，这是一颗实验卫星，我们在那里使相控阵和天线变暗，这些设计是为了解决在工作轨道上的的亮度问题。这使卫星的亮度降低了约55%，这是通过将DarkSat与邻近的Starlink卫星进行对比所得的结论。这几乎足够使卫星在工作轨道上时使地面肉眼不可见。然而，黑色表面会更易吸热并反射一些红外光，因此我们将用遮阳面来代替这个方案。这避免了由黑漆引起的过热问题，并预计比“涂黑”的Starlink卫星还要暗，它将阻止所有的光线到达白色的天线。

早期任务的操作

由于遮阳罩并不遮挡太阳阵列，这意味着它将不会给轨道上升带来阻力，但不会减少上升时的轨道亮度。为此，我们正在努力改变卫星的变轨模式。

通过减少接收光的表面积来减少反射到地球上的光。这是可能的，当轨道上升和到达待机轨道，这些未到达工作轨道的卫星不需要提供覆盖互联网用户。然而，有几个微妙的原因，这是很难实现的。首先，将太阳能倾斜于太阳会减少卫星可用的能量。第二，由于天线有时会不朝向地面，因此与卫星的接触时间将减少。

第三，跟踪相机位于底盘的两侧。将太阳能板指向太阳，一颗可以直接指向地球，另一颗直接指向太阳，但这却将导致卫星姿态调整能力更差。

会有一小部分的情况是，由于上述限制因素之一，正在升轨的卫星无法一直把太阳能板垂直于太阳。这可能导致在飞行轨道上偶尔出现一组十分密集的Starlink卫星，这些卫星在某一轨道的某一部分上是可以暂时看到的。

工作轨道的亮度

卫星一生中的大部分时间都停留在工作轨道上，在飞行过程中，它们的太阳能板大多是垂直于卫星的（因为工作的卫星便是太阳能板垂直于卫星）。我们可以调整太阳能板的位置，以反射光从它的太阳能板并远离地球，并把太阳能板隐藏在底盘后面。剩下的主要目标是阻止相控阵和天线被太阳直射。目标是用黑色橡胶塑料泡沫覆盖卫星两侧的白色天线。

利用我们的低轨道高度和平坦的卫星几何结构，我们为卫星设计了一个透明射频（不影响收发性能）的可展开遮挡罩，它阻挡光线到达卫星体的大部分和主体。这个遮阳罩在发射时平放在底盘上，在卫星与猎鹰9号分离时部署。遮阳罩通过完全阻挡光线到达天线来防止光线漫反射到地面。这种方法不仅避免了卫星天线过热，而且对亮度降低也有较大的好处。如前所述，第一颗“VisorSat”（或称“遮罩星”）原型将于5月发射（Starlink-8任务），到6月份，我们将在所有卫星上安装这些黑色遮挡罩（Starlink-10任务）。星通卫星两侧的半球天线也有遮阳罩，使整个卫星大大变暗。

我们一直在与主要天文团体——特别是美国天文学会和Vera C.Rubin天文台合作，以更好地了解天文学界使用的仪器。我们通过与一个天文学家工作组的定期交流通话，提高了我们对整个天文系统的了解，在此期间，我们讨论了技术细节，提供了更新，并致力于保护天文观测。

虽然互相理解是这个问题的关键，但是没有具体的细节，工程问题是很难解决的。Vera C.Rubin天文台多次认为光反射是最难解决的问题，因此我们在过去几个月里一直与那里的一个技术团队密切合作以解决这个问题。在其他有用的想法和讨论中，Vera C.Rubin团队提供了一个目标亮度降低，在我们迭代亮度解决方案时，我们使用它来指导我们的工程工作。

StarLink轨迹是通过Space-track.org和celestrak.com发布的，许多天文学家使用它们来计时观测以避免卫星给天文台拍摄的照片“划道子”。根据天文学家的要求，我们还开始在发射前发布预测数据。这样，观测站就可以在部署的最初几个小时内安排观测时间(当卫星升轨时)。

减少对天文学的影响

Vera C.Rubin天文台(Vera C.Rubin天文台)这样的较大望远镜的巨大收集区域会极为敏感，即使是最黑暗的卫星也是如此，即使是遮罩星也会产生影响，但这是所有卫星不可不免的。要减少卫星的影响，还有很多工作要做，首先要了解天文传感器是如何工作的。

天文界告诉我们他们的成像技术。光学系统使用透镜将光聚焦到成像传感器上。大多数光学天文仪器使用称为电荷耦合器件(CCDS)的传感器，因为遥远的超新星和星系等天文目标在传感器所能探测到的范围内通常是模糊的。在这些应用中，CCDS的低噪点水平使得给定图像的信噪比更高，使人们更容易看到宇宙中非常微弱的特征。

然而，CCDS有一个关键的缺点：与其他常见的传感器相比，比如你手机中的CMOS传感器。如果你把手机对准明亮的光线，你会看到所有的像素都饱和了，并且在亮光源的区域变成了白色。如果你用一个使用CCD传感器的光学系统来观察同一个目标，你会发现这个亮点在图像上产生垂直条纹。

这种差异是由于每个传感器类型读取每个像素的值的方式不同造成的。CMOS传感器本质上在每个像素处都有一个放大器，将采集到的光转换成数字值，而CCD传感器有有限数量的放大器，并将采集到的光(以电子的形式)移动到传感器上，以便数字化。这种机制意味着CCD上的饱和像素往往会从整个像素列中清除数据。

这一效应，通常被称为“开花”，是一个很小但明亮的光源如何影响天文观测的一个例子。这一原则是我们努力的方向核心。虽然不可能创造地球上最先进的光学设备来看不见的卫星，但通过降低卫星的亮度，我们可以使现有的处理类似问题的方法，如帧叠加，提高效率。

未来的Starlink卫星

SpaceX致力于使未来的卫星设计尽可能暗。下一代卫星，旨在利用星舰独特的发射能力，将专门设计为最小化亮度，同时也增加了消费者的数量，提供全球覆盖的偏远地区高速互联网。

虽然SpaceX是第一个解决卫星亮度问题的大型星座制造商和运营商，但我们不会是最后一个。随着发射成本继续下降，会出现更多星座，它们也需要确保卫星不会过度影响地面的问题。这就是为什么我们正在努力使这个问题更容易解决，并提供了先行模板。

END……

英语全文：

STARLINK DISCUSSION NATIONAL ACADEMY OF SCIENCES

SpaceX is launching Starlink to provide high-speed, low-latency broadband connectivity across the globe, including to locations where internet has traditionally been too expensive, unreliable, or entirely unavailable. We also firmly believe in the importance of a natural night sky for all of us to enjoy, which is why we have been working with leading astronomers around the world to better understand the specifics of their observations and engineering changes we can make to reduce satellite brightness. Our goals include:

Making the satellites generally invisible to the naked eye within a week of launch. We're doing this by changing the way the satellites fly to their operational altitude, so that they fly with the satellite knife-edge to the Sun. We are working on implementing this as soon as possible for all satellites since it is a software change.

Minimizing Starlink's impact on astronomy by darkening satellites so they do not saturate observatory detectors. We're accomplishing this by adding a deployable visor to the satellite to block sunlight from hitting the brightest parts of the spacecraft. The first unit is flying on the next launch, and by flight 9 in June all future Starlink satellites will have sun visors. Additionally, information about our satellites' orbits are located on space-track.org to facilitate observation scheduling for astronomers. We are interested in feedback on ways to improve the utility and timeliness of this information.

To better explain the details of brightness mitigation efforts, we need to explain more about how the Starlink satellites work.

Starlink Orbits

Starlink has three phases of flight: (1) orbit raise, (2) parking orbit (380 km above Earth), and (3) on-station (550 km above Earth). During orbit raise the satellites use their thrusters to raise altitude over the course of a few weeks. Some of the satellites go directly to station while others pause in the parking orbit to allow the satellites to precess to a different orbital plane. Once satellites are on-station they reconfigure so the antennas face Earth and the solar array goes vertical so that it can track the Sun to maximize power generation. As a result of this maneuver, the satellites become much darker because the solar array visibility from the ground is greatly reduced. 

Currently, about half of the over 400 satellites are on-station and the other half are orbit raising or in the parking orbit. Satellites spend a small fraction of their lives orbit raising or parking and spend the vast majority of their lives on-station. It's important to note that at any given time, only about 300 satellites will be orbit raising or parking. The rest of the satellites will be in the operational orbit on-station. 

Starlink Satellite

The Starlink satellite design was driven by the fact that they fly at a very low altitude compared to other communications satellites. We do this to prioritize space traffic safety and to minimize the latency of the signal between the satellite and the users who are getting internet service from it. Because of the low altitude, drag is a major factor in the design. During orbit raise, the satellites must minimize their cross-sectional area relative to the "wind," otherwise drag will cause them to fall out of orbit. High drag is a double-edged sword—it means that flying the satellites is tricky, but it also means that any satellites that are experiencing problems will de-orbit quickly and safely burn up in the atmosphere. This reduces the amount of orbital debris or "space junk" in orbit. 

This low-drag and thrusting flight configuration resembles an open book, where the solar array is laid out flat in front of the vehicle. When Starlink satellites are orbit raising, they roll to a limited extent about the velocity vector for power generation, always keeping the cross sectional area minimized while keeping the antennas facing Earth enough to stay in contact with the ground stations.

When the satellites reach their operational orbit of 550 km, drag is still a factor—so any inoperable satellite will quickly decay—but the attitude control system is able to overcome this drag with the solar array raised above the satellite in a vertical orientation that we call "shark-fin." This is the orientation in which the satellite spends the majority of its operational life.

Satellite Visibility

Satellites are visible from the ground at sunrise or sunset. This happens because the satellites are illuminated by the Sun but people or telescopes on the ground are in the dark. These conditions only happen for a fraction of Starlink's 90-minute orbit.

This simple diagram highlights why satellites in orbit raise are so much brighter than the satellites that are on-station. During orbit raise, when the solar array is in open book, sunlight can reflect off of both the solar array and the body of the satellite and hit the ground. Once on-station, only certain parts of the chassis can reflect light to the ground.

Physics of Satellite Brightness

The apparent magnitude of an object is a measure of the brightness of a star or object observed from Earth. It is a reverse logarithmic scale, so higher numbers correspond to dimmer objects. A star of magnitude 3 is approximately 2.5 times brighter than a star of magnitude 4. Based on observations that have been taken by us and by members of the astronomical community, current Starlink satellites have an average apparent magnitude of 5.5 when on-station and brighter during orbit raise. Objects up to about magnitude 6.5-7 are visible to the naked eye (naked-eye visibility is closer to 4 in most suburbs), and our goal is for Starlink satellites to be magnitude 7 or better for almost all phases of their mission. 

There are two types of reflections off of Starlink satellites: diffuse and specular. Diffuse reflections occur when light is scattered in many different directions. Imagine shining a flashlight at a white wall. Specular reflections happen when light is reflected in a particular direction. For example, the glint of sunlight off of a mirror. Diffuse reflections are the biggest contributor to observed brightness on the ground, because diffuse reflections go in all directions. You can see diffuse reflections as long as the satellite is visible. This is why Starlink satellites can create the "string of pearls" effect in the night sky. It's a little counter-intuitive, but the shiny components of the Starlink satellites are a much smaller problem. Whether diffuse or specular, having a high reflectance helps the satellites stay cool in space. When sunlight hits a specular surface of the spacecraft and reflects, the vast majority of light reflects in the specular (mirror reflection) direction, which is generally out toward space (not toward Earth). Occasionally when it does, the glint only lasts for a second or less. In fact, specular surfaces tend to be the dimmest part of the satellite unless you are at just the right geometry.

The biggest contributors to Starlink being bright are the white diffuse phased array antennas on the bottom of the satellite, the white diffuse parabolic antennas on the sides (not shown below), and the white diffuse back side of the solar array. These surfaces are all white to keep temperatures down so components do not overheat. The key to making Starlink darker is to prevent sunlight from illuminating these white surfaces and scattering via reflection toward observers on the ground. While in orbit raise and the parking orbit the solar array dominates due to the much larger surface area. However, once the satellites are at their operational altitude, the antennas dominate because the bright backside of the solar array is shadowed.

Solutions In-Work

We've taken an experimental and iterative approach to reducing the brightness of the Starlink satellites. Orbital brightness is an extremely difficult problem to tackle analytically, so we've been hard at work on both ground and on-orbit testing.

For example, earlier this year we launched DarkSat, which is an experimental satellite where we darkened the phased array and parabolic antennas designed to tackle on-station brightness. This reduced the brightness of the satellite by about 55%, as was verified by differential optical measurements comparing DarkSat to other nearby Starlink satellites. This is nearly enough of a brightness reduction to make the satellite invisible to the naked eye while on-station. However, black surfaces in space get hot and reflect some light (including in the IR spectrum), so we are moving forward with a sun visor solution instead. This avoids thermal issues due to black paint, and is expected to be darker than DarkSat since it will block all light from reaching the white diffuse antennas.

Early Mission (Orbit Raise and Parking Orbit) Roll Maneuver

Since the visor is intended to help with brightness while on-station, it does not shade the back of the solar array, which means that it will not prevent orbit raise and parking orbit brightness. For this, we are working on changing the way the satellite flies up from insertion to parking orbit and to station.

We're currently testing rolling the satellite so the vector of the Sun is in-plane with the satellite body, i.e. so the satellite is knife-edge to the Sun. This would reduce the light reflected onto Earth by reducing the surface area that receives light. This is possible when orbit raising and parking in the precession orbit because we don't have to constrain the antennas to be nadir facing to provide coverage to internet users. However, there are a couple of nuanced reasons why this is tricky to implement. First, rolling the solar array away from the Sun reduces the amount of power available to the satellite. Second, because the antennas will sometimes be rolled away from the ground, contact time with the satellites will be reduced. Third, the star tracker cameras are located on the sides of the chassis (the only place they can go and have adequate field of view). Rolling knife edge to the Sun can point one star tracker directly at the Earth and the other one directly at the Sun, which would cause the satellite to have degraded attitude knowledge.

There will be a small percentage of instances when the satellites cannot roll all the way to true knife edge to the Sun due to one of the aforementioned constraints. This could result in the occasional set of Starlink satellites in the orbit raise of flight that are temporarily visible for one part of an orbit.

On-Station Brightness

Satellites spend most of their lives on-station, where they will always be in the shark-fin configuration during visible passes. We can adjust the solar array positioning in this configuration to reflect light from its largely specular solar cells away from Earth and to partially hide it behind the chassis. The main remaining goal is to block the phased arrays and antennae from direct view of the sun. The goal is to cover the white phased array antennas and the parabolic antennas on the sides of the satellite.

Using our low orbital altitude and flat satellite geometry to our advantage, we designed an RF-transparent deployable visor for the satellite that blocks the light from reaching most of the satellite body and all of the diffuse parts of the main body. This visor lays flat on the chassis during launch and deploys during satellite separation from Falcon 9. The visor prevents light from reflecting off of the diffuse antennas by blocking the light from reaching the antennas altogether. Not only does this approach avoid the thermal impacts from surface darkening the antennas, but it should also have a larger impact on brightness reduction. As previously noted, the first VisorSat prototype will launch in May and we will have these black, specular visors on all satellites by June. The parabolic antennas on the sides of the Starlink satellite also have visor-like coverings that darken them.

We have been working with leading astronomical groups in this effort—in particular the American Astronomical Society and the Vera C. Rubin Observatory—to better understand the methods and instruments employed by the astronomy community. With AAS, we have increased our understanding of the community as a whole through regular calls with a working group of astronomers during which we discuss technical details, provide updates, and work on how we can protect astronomical observations moving forward. A post on some of our sessions is here. One particularly useful presentation from a member of this working group is here.

While community understanding is critical to this problem, engineering problems are difficult to solve without specifics. The Vera C. Rubin Observatory was repeatedly flagged as the most difficult case to solve, so we've spent the last few months working very closely with a technical team there to do just that. Among other useful thoughts and discussions, the Vera Rubin team has provided a target brightness reduction that we are using to guide our engineering efforts as we iterate on brightness solutions.

These technical and community discussions are paired with our existing efforts to make the satellites easier for astronomers to avoid. Starlink trajectories are published through Space-track.org and celestrak.com, which many astronomers use in timing their observations to avoid satellite streaks. We've also started publishing predictive data prior to launch at the request of astronomers. These allow observatories to schedule around the satellites in the first few hours of deployment (as satellites de-tumble and enter the network).

Vera Rubin has been described as the limiting case for Starlink, due to its enormous aperture and wide field of view. These two characteristics work in concert to produce the perfect storm for satellite observations. Most astronomical systems look at an extremely small section of the sky (less than 1 degree), which makes it exceedingly unlikely that a satellite will cross in front of the imaging system in a given observation. On the other hand, systems with very large fields of view normally aren't extremely sensitive, meaning that, while streaks will occur, they will have a small impact on the overall data collection. This is why we've been working so closely with the team at the Rubin Observatory. In fact, despite its wide field of view, the Vera C. Rubin Observatory is sensitive enough to detect a sunlit golf ball as far away as the Moon.

So what can we do to mitigate our impact on these edge cases of wide, fast survey telescopes?

Minimizing the Impact on Astronomy

The huge collecting area of a larger telescopes like Vera C. Rubin Observatory leads to a sensitivity that will render even the darkest satellites visible.They are so sensitive that it won't be possible to build a satellite that will not produce streaks, in a typical long integration. There is much that can be done to reduce the impact of satellite streaks, and that starts with an understanding of how astronomical sensors work.

The astronomical community has done a great job of educating us on their imaging techniques. Optical systems use mirrors or lenses to focus light onto an imaging sensor. Most optical astronomy instruments use sensors called charge-coupled-devices (CCDs) as their detectors because astronomical targets, such as distant supernovae and galaxies, are generally dim–at the limit of what can be detected by a sensor. For these applications, the lower noise level of CCDs allows for a higher signal-to-noise ratio for a given image, making it easier to see very faint features in the universe.

However, CCDs suffer from a key drawback: when compared to other common sensors, like the CMOS sensor in your cell phone. If you point your cell phone at a bright light, you'll see all the pixels saturate and turn white in the region of the bright source. If you look at the same target with an optical system that uses a CCD sensor, you'll notice that this bright spot extends to create vertical stripes on the image.

This difference is due to the way each sensor type reads the values for each pixel. While a CMOS sensor essentially has an amplifier at each pixel that turns the light collected into a digital value, a CCD sensor has a limited number of amplifiers and moves the collected light (in the form of electrons) across the sensor, to be digitized. This mechanism means that a saturated pixel on a CCD tends to wipe out data from an entire column of pixels.

This effect, commonly referred to as 'blooming,' is one example of how a very small but bright source of light can impact an astronomical observation. This principle is the core of our mitigation efforts. While it will not be possible to create satellites that are invisible to the most advanced optical equipment on Earth, by reducing the brightness of the satellites, we can make the existing strategies for dealing with similar issues, such as frame-stacking, dramatically more effective.

Future Satellites

SpaceX is committed to making future satellite designs as dark as possible. The next generation satellite, designed to take advantage of Starship's unique launch capabilities, will be specifically designed to minimize brightness while also increasing the number of consumers that it can serve with high-speed internet access. 

While SpaceX is the first large constellation manufacturer and operator to address satellite brightness, we won't be the last. As launch costs continue to drop, more constellations will emerge and they too will need to ensure that the optical properties of their satellites don't create problems for observers on the ground. This is why we are working to make this problem easier for everyone to solve in the future.