Quote:
Originally Posted by Peter Ward
|
thanks Peter - very useful summary and insight into SBIG philosophy and it has crystallised a number of ideas that had been previously unsupported by data.
However, I found that it warranted some very careful reading to be sure what it is saying and what it is not saying and to understand what the data actually represents. Overall (and perhaps rather surprisingly), I think that it confirms most of what has already been said on this thread:
1. The measured tracking data shows the expected high speed seeing fluctuations, (uncorrelated over large angles) plus an underlying slow variation that is widely correlated. The author assumes that the slow component is due to some form of seeing, but seems to dismiss the (more obvious?) possibility that it is underlying residual mount error - not sure why.
2. Regardless of what the slow component is, the “fast” simulation in figure 6 shows that the performance drops when two star tracking is used – let’s say that again: the data clearly shows that the stars would be bigger if you turned on an AO under these conditions. This is explained in terms of the system chasing the fast seeing – exactly the problem identified earlier in the thread.
3. The “slow” simulation data in figure 7 shows that if you slow down the AO update rate, you can integrate the seeing and get some gain from using an AO. ie, there is likely to be an optimum update rate that will vary with mount characteristics and seeing – faster is not necessarily better.
4. Both simulations show that multi-star guiding is best – this was indicated as a possibility in earlier posts, but primarily if the AO was being used to correct mount errors. In any case, as far as I know multi-star is not an option at present, so the main performance gains shown in the figures refer to unavailable technology - looks like it would be a good idea for the future though.
5. The simulation graphs 6 and 7 look like they are star profiles and you could easily assume that to be the case – however, as the author states later in the text, that is not what they are. Star profiles will only be obtained by convolving these tracking functions with the seeing function. The measured track functions all have quite small FWHM. So the effect of AO correction for the test system will at best be only slightly noticeable in the star profiles, and then only if the seeing is very good (eg below about 2 arc seconds FWHM). AO correction will not be of any consequence at all for worse seeing, where the tracking will be completely overshadowed by the seeing. Looks like Peter’s theory is on the money.
6. It appears that SBIG is working on a complicated way to incorporate multi-star tracking and to have the tracking stars in the imaging field of view using a separate guidescope – clearly they have identified limitations in the current approach.
In summary, based on this paper:
- AO will tidy up mount tracking/slow seeing, but only if the AO update rate is at low enough, or based on multiple stars.
- There is likely to be an optimum update rate based on seeing and mount characteristics and it will probably be between 1Hz and 10Hz, based on the limited data available.
- AO will not tidy up conventional fast seeing outside of the isokinetic patch. If the update rate is too high it has now been demonstrated that it will actually make matters worse (by a factor of up to 1.4x from the theory).
- There is a slow component to the tracking error that is correlated over large angles and time. It could be either residual mount error or some (previously undocumented?) form of slow seeing – the verdict is still out - but whatever it is, this is the only part of the tracking error that current AO will correct.
- If you have a good quality mount, AO correction will probably only have a noticeable effect in very good seeing – it certainly will not fix up bad seeing, but it could fix up a lower quality mount and help with a good mount if used with care.
