Hunt For Earth-Like Planets Delayed

An anonymous reader sends along this excerpt from Nature News: "Kepler, NASA's mission to search for planets around other stars, will not be able to spot an Earth-sized planet until 2011, according to the mission's team. The delays are caused by noisy amplifiers in the telescope's electronics. ... The problem is caused by amplifiers that boost the signals from the charge-coupled devices that form the heart of the 0.95-metre telescope's 95-million-pixel photometer, which detects the light emitted from the distant stars. Three of the amplifiers are creating noise that compromises Kepler's view. The noise affects only a small portion of the data, Borucki says, but the team has to fix the software — it would be 'too cumbersome' to remove the bad data manually — so that it accounts for the noise automatically. He says that the fix should be in place by 2011." Mindful of Halloween's approach, NASA has put up a piece looking at some of the already-known exoplanets that wouldn't be very friendly to human life.

Re:head scratch...

[cheesybagel]

Yep, the issue is bandwidth. The downlink does not have enough capacity to send the daily produced photographs for ground processing, so the satellite does on-board image processing on the photographs and just sends the results back to Earth. Consider the SuperWASP [wikipedia.org] system generates 100 GB per night and you will get an idea of the amount of data being processed for this kind of application.

Re:head scratch...

[bughunter]

I used to design these kinds of cameras, and there are at least two potential reasons why this can't be done on the ground:

Firstly, and most likely, there's an essential step that needs to be done in the camera hardware. Perhaps something related to Correlated Double Sampling [ccd.com] or Pixel Binning [ccd.com] needs to be adjusted. In the first case, the signal and reset measurements need to be done as close together as possible to reduce 1/f noise, which can quickly dominate the noise side of the SNR expression, and it may be the timing of these measurements that is at fault. In the latter case, there would be a sqrt(N) penalty for measuring the charge on each pixel and then adding the N pixels together. Conversely, reading each pixel multiple times may be necessary to overcome an unexpected noise source - a sqrt(N) improvement in readout noise can be had by measuring the pixel N times and calculating the average of the measurements. All of these adjustments can only be made at the camera; there's no way to accomplish this after the data has been digitized and radioed to Earth.

2 - Bandwidth. There's just no bandwidth available to ship down the raw data so it can be processed on Earth. The spacecraft must send down reduced data and derivative results. Therefore, these corrections need to be made onboard.

These reasons aren't necessarily exclusive, either... both could be true.