Quote:
Originally Posted by Peter Ward
I remain puzzled as to why you are seeing better performance with the SX AO turned off, as opposed to it being active. My intention was not to bag SX, but your data does not reflect the results Alan Holmes has published, nor my own qualitative results or that seen by many other (SBIG) AO users.
Alan's data shows a clear improvement & correlation, in both Dec and RA that slowly decreases with angular distance from the guide star ( if only mount errors were being corrected, that would not expect a Dec correlation).
I think it incorrect to say, the physics predicts a first order AO will make things worse.
Quite the contrary.
If the atmosphere's isoplanic patches on the night are at least as large as the telescope aperture, tip-tilt or first-order corrections, provided they are fast enough, will be beneficial....this is exactly how professional systems work prior to applying complex wavefront corrections for higher order components via deformable optics.
On nights when higher order turbulence dominates the image, sure, I'd expect zero improvement...but imaging during tragic seeing, regardless of how good your optics/mount/etc. are, is a futile exercise IMHO.
Suggesting residual mount errors/wind buffeting is all that is being corrected by the AO doesn't offer much insight ( with a system similar to mine )as my mount's PE is around 1-2 arc sec and, being in a dome, there is zero buffeting. Begging the question: if not isoplanic waves, what is being corrected ? That said, I am now curious.
It will be interesting to gather some data, with and without high-speed first-order corrections ( in my case being applied by an AO-8 and AO-X) and similarly quantify the results. I'll happily upload the results when done.
|
Maybe Alan's own words can convince you:
"Figure Six shows the star image that would result if the errors were only due to stellar motion. If I didn’t guide at all, but had a perfect drive, I would obtain the result given by the "Slow Guide curve. The higher the peak value, the better the star image. With Two Star guiding, the result actually got a little worse. This is what I mentioned might happen before – the guide system is chasing the seeing, but one frame behind. " ie Alan's published results match very well with what I observed - the star images are "a little worse" with an AO than without. And they were taken in similar conditions of:
"These measurements were made on a night with only a slight wind." Alan then goes on show how reducing the framerate can actually improve performance and then to explain how much better multi-star guiding would be, but that discussion has nothing to do with current systems.
As for mount errors, Alan's analysis method specifically excludes them (he uses differential measures). I excluded them by measuring on a windless night and with short image exposures (10sec) on a pretty good mount. And we came up with the same answer - AO (of any brand) is quite capable of reducing resolution. If you accept this, then the next question is "well why do they help at all?" - what you get probably depends on the spectrum of the seeing error, but I suspect that an AO can tidy up a myriad of mount wobbles, producing a nett benefit only when the improvement from the tidy up is more significant than the increase in seeing noise (ie the seeing must be good to start with). This is pure conjecture though and the complete answer is probably nothing like that simple.
Agree that pro systems must use tip-tilt in conjunction with deformable optics, but I am pretty sure that they would never even dream of trying it with a guide star outside of the isokinetic patch and at bandwidths of 10Hz.
In any case, would really look forward to seeing some hi res FWHM measurements
. I also hope to get some more measurements soon in conditions that are more suited to showing that improvements are possible with AO - not just degradation. In the end, a theory is only of value if it is backed up by measurements - however, if we can get a clearer understanding of what exactly is going on, it may be possible to push the limits a bit harder. In particular, it may be possible to choose an optimum update rate, since faster is not always better.
On isokinetic and isoplanatic patch, I think that they are generally regarded as being slightly different: the isoplanatic patch is that region over which all turbulent effects are correlated. The isokinetic patch is that over which only the low order tip-tilt stuff is correlated - it is bigger than the isoplanatic patch.
Ref:
https://www.sbig.com/about-us/blog/differential/