Thursday, June 28, 2012

NuSTAR first light

The press release, with the pointing and metrology-corrected data with two orbits of integration can be found here. In this post, you get to see the instant-gratification version of the first light image.
NuSTAR consists of a pair of co-aligned x-ray telescopes. It's sort of like the Chandra and Newton observatories, but also different because it has graded multilayer mirrors that allow it to reflect much higher energy photons than those other two x-ray telescopes (they reflect up to ~15 keV, while NuSTAR is good up to 80 keV). My graduate work, which I started in 1995, involved making the algorithm for multilayer design optimization. The code I developed for the balloon-borne HEFT mission's mirrors was also used for NuSTAR's optics.
Another difference between NuSTAR and Chandra & Newton is the detectors -- NuSTAR uses CdZnTe detectors to catch the higher energy x-rays. The energies we can see with NuSTAR would pass right through Chandra's and Newton's Si CCDs.
Now, under "normal" circumstances, we (the NuSTAR team) get a 12 minute ground station pass every ~90 minutes, and the data processing pipeline takes ~30 minutes to turn packets of data into something comprehensible to astronomers. I was looking for instant gratification, so I talked to Rick Cook (who designed the NuSTAR readout ASICs) about getting a text output map of the focal planes during the ground station pass. We needed it to be as fast as possible, so Rick decided to pack two pixels per ascii character, allowing 3 bits per pixel (ASCII characters are 7 bits). I am a big logarithmophile, so I probably pushed for some form of log scaling. Rick gave me a scheme where 0 counts = 0, and otherwise, the output value is [floor(log_2(counts))+1]. I wrote the program to decode Rick's output stream in real time to generate an log-scaled image of the intensity on the detector. So this is what we saw in the instrument room at Space Sciences Lab, Berkeley about one minute before we lost ground station contact:
Note that this is just a 5 second integration with no pointing/metrology corrections applied (though one may argue that an integration that short shouldn't need much correction. The press release has a much nicer looking image, but this isn't really all that different. In the PR, the scaling is [counts^2], which emphasizes the peakiness of the distribution. Log scaling always shows you all the warts, and as you can see, NuSTAR is (knock on wood) pretty wart-free!


Post a Comment

<< Home