The historical origins of this project were actually chemical analysis of burning things in flames (like the original Bunsen and Kirchoff experiments) so the equipment was made as small and portable as possible (the whole apparatus, including all the optics and camera, is easily hand in one hand). It was built to be held on a small camera tripod so we found it could easily be held on a small Vixen mount. Here is a photo of the instrument in it's most recent incarnation:
- SHG
- grism-helio.jpg (252.99 KiB) Viewed 2245 times
The ASI 120MM camera, used in combination with FireCapture, turns out to be very good. First of all, it is relatively sensitive in the blue. The pixel size (3.75 microns) and chip size (1280 x 960 pixels) are also helpful. But most importantly, when using ROI, the video capture speed is very good. It seemed to top out at around 120 fps on my old WindowsXP generation laptop. For the shutter speed of 3.51 ms I was using, FireCapture said the theoretical maximum speed was 285 fps. I'm not sure if the bottleneck is UBS 2.0 or the hard disk write speed. But in any case, 120 fps is still very good. Here is an example of a video captured (it has been compressed for presentation so does not contain all the spectral or temporal information). The ROI was restricted to 384x464 pixels to just focus on the two Calcium lines at 3934 and 3968 angstroms. The hand controller of the Vixen mount was used to manually scan back and forth in RA at the maximum manual rate of 32x sidereal. Normally, the Sun will scan itself over the slit in 120 seconds (on a stationary mount). At 32x, it takes 3.75 sec. As we scan a little further on each side, a full scan is actually about 5 seconds. In a 30 second video, we can make 5 full scans, back and forth. At 120 fps and 3.75 sec, we take about 450 horizontal lines across the Sun's surface. With the optics used here, the Sun's full diameter is projected across approximately 375 pixels, so the image is oblong by a factor of 1.2.
Now I rely on the guidance of Wah to process the data. The original AVI from Firecapture is put into a program called VirtualDub (a very flexible but slightly over featured piece of software). I used this to convert from 120 fps down to 25 fps (I'm not sure I need to do this but I wanted to use other video editing software as well) and also to edit out some of the "black" areas (where there was no Sun visible, at the end of the scans -- also not an essential step). For some reason, the output AVI from VirtualDub was 1.22 GB vs 596 MB for the input (even though parts were edited out). This new video I put into Wah's program SpecLineMerge (in the Solar Reference Library). This is a very simple and easy to use program. I figured out how to use it fairly quickly. Only later I discovered that there is a kind of "Users Manual" (see "SHG imaging & processing" in the Library). But any program you can use without reading the manual must be a good one.
I chose a bandwidth of 4 pixels, which in my case I believe corresponds to 0.8 angstroms. I pushed the button for "Merge" and SpecLineMerge fairly quickly created an image file. After applying "Auto Contrast" in Photoshop Elements, out popped this amazing image:
- merged
- Calcium_K_4.jpg (97.7 KiB) Viewed 2245 times
- sun cak
- 4ps.jpg (26.55 KiB) Viewed 2245 times
This really was the very first attempt. The things we need to do to improve things (apart from those mentioned above already): 1) Expose longer to achieve better contrast inside the spectral lines. I made the mistake of exposing as if to capture nice lines, not the structure within the lines. 2) Use a bigger "telescope". I think even moving to a 200mm f4 camera lens (50mm diameter) will improve resolution considerably. 3) Try to double fps to 200 (maybe smaller ROI will accomplish this easily).
I am also intrigued by the idea of "stacking" the SHG images. At 4 sec per scan, 64 frames could be captured within 4.25 minutes, which should increase S/N by a factor of 8. Maybe Wah can figure out how to automatically stack the output of SpecLineMerge (like the above image with 5 disks). I made a quick attempt with AS2.3 and Reg6 but the results were not better than a single image.
