|
|
|
Thread Tools | Display Modes |
#11
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
Chris L Peterson wrote:
On Fri, 09 Jul 2004 05:25:58 GMT, Eric wrote: Use a high contrast region on the surface of the moon. Just like they do with the solar telescopes that use AO. Think Registax... use the same math to point the tip/tilt mirror. Sure, maybe you get 30" patch if you're lucky, but we're talking about high resolution. You have to oversample the telescope to make it worth while. I was thinking more in terms of higher order corrections, which are difficult to do with anything but a point source reference. But I agree, at very high magnifications you could get some benefit from tip/tilt using simple feature correlation for control. If you can build a Shack-Hartman mask, then you get multiple images of the tracking target. Then use the correlation to determine the shifts of each subimage. Then you can reconstruct the higher orders. If you're imaging the Moon or the Sun, you have lots of light. You need field-stops to keep the images from overlaping on the sensor. It takes some real computing power, but I expect you'll be limited by the readout rate of the camera. Eric. |
#12
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
Chris L Peterson wrote: On Fri, 09 Jul 2004 04:15:34 GMT, "Marc Reinig" wrote: The size of the target shouldn't matter, you could improve an image of the moon with AO. I'm not sure I follow. In the visible spectrum, the isoplanatic patch size ranges from 1 to 15 arcseconds. For first order corrections, very short exposure times at 1um, that might go as high as 30 arcseconds. With typical FOVs on the order of 20 arcminutes or more, that essentially means that you have an uncorrected image. Not really uncorrected, just less than optimally corrected. You can still image the lower parts of the atmosphere and parts of the upper atmosphere depending on the angle. And if you can image them, you can correct for them. What kind of reference would you use to correct an image of the Moon? A feature on the moon itself, as Eric mentioned, or a very nearby star at the time of exposure. With the limits of correction as above. Marc Reinig UCO/Lick Laboratory for Adaptive Optics |
#13
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
Eric wrote: If you can build a Shack-Hartman mask, then you get multiple images of the tracking target. Then use the correlation to determine the shifts of each subimage. Then you can reconstruct the higher orders. If you're imaging the Moon or the Sun, you have lots of light. You need field-stops to keep the images from overlaping on the sensor. It takes some real computing power, but I expect you'll be limited by the readout rate of the camera. Technically, if you are using a mask, it would be a Hartmann mask. Shack replaced the holes with lenses so we could collect light more efficiently ;=) Actually, the readout rate shouldn't be a problem. You don't need that much resolution, a 5 X 5 pixel array should suffice and you don't need 100 X 100 sub aperatures, fewer should be fine, so a smaller camera array or binning would do. In any case, I suspect that the processing is what is going to be more of a limit. A Hartmann mask shouldn't be too hard or expensive to build. I haven't heard of any inexpensive Shack-Hartmann arrays available, though, but there may be some coming. However, with the moon or sun I would think it should be possible to use just a Hartmann mask. Has anyone done a simple deformable mirror? I think that might be the real tough one. Marc Reinig UCO/Lick Laboratory for Adaptive Optics |
#14
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
On Sat, 10 Jul 2004 13:55:38 -0700, "Marc R. Reinig"
wrote: Has anyone done a simple deformable mirror? I think that might be the real tough one. I built a proof of concept deformable mirror from a single silicon wafer, with four segments driven by resonant piezo stacks. I've never used it off the bench, and the project is on hold at the moment. Structures like this can be made very inexpensively, although I doubt there is really much of a market. _________________________________________________ Chris L Peterson Cloudbait Observatory http://www.cloudbait.com |
#15
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
Marc R. Reinig wrote:
Actually, the readout rate shouldn't be a problem. You don't need that much resolution, a 5 X 5 pixel array should suffice and you don't need 100 X 100 sub aperatures, fewer should be fine, so a smaller camera array or binning would do. In any case, I suspect that the processing is what is going to be more of a limit. Right. Myabe you only need a dozen or so sub-apertures. Correct for Tip/Tilt/Defocus with four actuators. So I need: beam splitter. 4-hole Hartman mask. 4 small lenses to direct the (roughly) focused images onto the edge of a CCD. read the 8(ish) edge lines of the CCD. get the centroids. do the calc. apply the signal to the actuators. A Hartmann mask shouldn't be too hard or expensive to build. I haven't heard of any inexpensive Shack-Hartmann arrays available, though, but there may be some coming. However, with the moon or sun I would think it should be possible to use just a Hartmann mask. How do you keep the images from overlapping at the CCD ? Has anyone done a simple deformable mirror? I think that might be the real tough one. There were some people making scanning electric-force microscopes using a piezo-speaker to point the probe. Maybe epoxy a mirror to a suitable modified piezo-speaker ? Eric. |
#16
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
Chris L Peterson wrote:
On Sat, 10 Jul 2004 13:55:38 -0700, "Marc R. Reinig" wrote: Has anyone done a simple deformable mirror? I think that might be the real tough one. I built a proof of concept deformable mirror from a single silicon wafer, with four segments driven by resonant piezo stacks. I've never used it off the bench, and the project is on hold at the moment. Structures like this can be made very inexpensively, although I doubt there is really much of a market. Whose piezo-stacks did you use ? ThorLabs ? How high a voltage did you use ? Eric. |
#17
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
On Sat, 10 Jul 2004 21:27:41 GMT, Eric wrote:
Whose piezo-stacks did you use ? ThorLabs ? How high a voltage did you use ? The stacks are integrated into the mirror fabrication. They provide 300 um excursion with a 15V drive signal. I developed these originally to control a tip/tilt prism. They are not simple stacks, but integrated PZT/Si structures with feedback sensors. They are driven with AC and the excursion is adjusted by driving them away from their resonant frequency. They are great controllers for the tip/tilt device, which I've used on a telescope; they appear suitable for a deformable mirror, but since I haven't used this device in a closed loop corrector yet, I don't know what problems might come up. _________________________________________________ Chris L Peterson Cloudbait Observatory http://www.cloudbait.com |
#18
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
Eric wrote:
Marc R. Reinig wrote: Actually, the readout rate shouldn't be a problem. You don't need that much resolution, a 5 X 5 pixel array should suffice and you don't need 100 X 100 sub aperatures, fewer should be fine, so a smaller camera array or binning would do. In any case, I suspect that the processing is what is going to be more of a limit. Right. Myabe you only need a dozen or so sub-apertures. Correct for Tip/Tilt/Defocus with four actuators. So I need: beam splitter. 4-hole Hartman mask. 4 small lenses to direct the (roughly) focused images onto the edge of a CCD. read the 8(ish) edge lines of the CCD. get the centroids. do the calc. apply the signal to the actuators. A Hartmann mask shouldn't be too hard or expensive to build. I haven't heard of any inexpensive Shack-Hartmann arrays available, though, but there may be some coming. However, with the moon or sun I would think it should be possible to use just a Hartmann mask. How do you keep the images from overlapping at the CCD ? As you said, you would still need a field-stop(s) ;=). Marco |
#19
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
There's been some work done without a deformable mirror. You have two
cameras. One is the science camera, the other is a wavefront camera. You take images at the same rate with both cameras, and with short exposures to freeze the seeing. After you have all your data, you can take the best images from the science camea and run them through something like Registax. OR you can use the wavefront images to correct their repsctive frames, and then run everything though something like Registax. You don't need a deformable mirror. You don't need to do the calculations in realtime. You're stuck with bright objects because you need a short exposure science image. You DO need to synchronise the two data streams. I've only seen a few things in EA and JOSA, and mostly simulated data. Anybody head of real data hadled this way ? Eric. |
#20
|
|||
|
|||
Is anyone doing adaptive optics at a non institutional level?
I know of one researcher doing post processing of images using
deconvolution. It's incredibly processor intensive. I don't know if they apply it to a sequence of images, and I don't think it really relies on a guide star at all, though I'm sure it would help. The rest I am familiar with are using it for real time correction of single long exposure images over many, many, many correction cycles, usually using a laser guide star. I'm pretty sure laser guide stars are out of the question for non institutional use, for regulatory reasons, at least in the US G. -Marco "Eric" wrote in message news:vkdIc.32637$eO.16349@edtnps89... There's been some work done without a deformable mirror. You have two cameras. One is the science camera, the other is a wavefront camera. You take images at the same rate with both cameras, and with short exposures to freeze the seeing. After you have all your data, you can take the best images from the science camea and run them through something like Registax. OR you can use the wavefront images to correct their repsctive frames, and then run everything though something like Registax. You don't need a deformable mirror. You don't need to do the calculations in realtime. You're stuck with bright objects because you need a short exposure science image. You DO need to synchronise the two data streams. I've only seen a few things in EA and JOSA, and mostly simulated data. Anybody head of real data hadled this way ? Eric. |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Hubble vs. ground-based adaptive optics | Mike Ruskai | Amateur Astronomy | 16 | February 19th 04 09:37 AM |
Adaptive Optics, amateur level? | Scribe2b | Amateur Astronomy | 22 | December 28th 03 09:11 PM |
Keck Observatory Reaches Major Milestone On Road To Expand Adaptive Optics | Ron Baalke | Astronomy Misc | 0 | October 8th 03 08:18 PM |
Keck Observatory Reaches Major Milestone On Road To Expand Adaptive Optics | Ron Baalke | Technology | 0 | October 8th 03 08:17 PM |