2011-11-03 Email from Judit:
I know, I expected that there will be more detected photons at the point out there at the telescope.
I am sending a pdf of the notebook, assuming you do not have Mathematica installed on your computer. There are enough comments in the file, I think, so one can follow how the calculations are made.
2011-11-09 Email from Judit:
Here I am sending some more graphs:
The graph entitled “Transmittance” is to illustrate how we would loose transmittance in front of the WFS as aluminum coated surfaces age. I am showing the 4 old - 2 new combination too, because on the two mirrors after M4 we may put enhanced coating, which could maintain higher reflectance longer.
On the other three graphs for three spectral types you can see two lines: the lower line was calculated assuming 6 old mirrors, the other one 6 new aluminum mirrors. The number of detected photons as a function of V magnitude is expected in the band between the two lines on each graph.
As for comparing results to real life; this routine predicted for the old TT system 7 detected photons in 1 ms for a V=10 G0 star. I think, that is close to what we are getting.
In the original routine the aluminum transmittance was taken as flat 80%. Here I am using spectral reflectance curves from TR91, as noted in the pdf. A single “old” coating in my routine is slightly higher than 80% on the red side, and less on the blue side.
Transmittance.pdf B0.pdf F0.pdf M0.pdf
Comment from Theo: This still seems very low, but I can't fault the math. How does this compare with the estimates contained in the proposal itself? Steve, can you remind us of where the original estimates came from? Clearly, we will need all of the visible light for this.
2011-11-11 Email from Steve:
I see from TR91 that the visible reflectance for old aluminum was extrapolated from much to the red, and also the comment that cleaning techniques can bring old aluminum reflectivity to near the new value. After some hunting, I found something more direct on this. I'm attaching a spreadsheet made at Kitt Peak when monitoring telescope primaries. I see that the reflectivity does seem to stabilize at considerably better than the “bad” case in TR91. From this spreadsheet, it appears that the “fresh” reflectivity is more representative of optics that are cleaned regularly than the “old” reflectivity. The old case in TR91 seems to be pessimistic.
Sample of Kitt Peak Primary Mirror Reflectivities.xls
2011-11-11 Email from John:
I agree. I measured reflectivity of ridiculously old and dirty aluminum coatings at iota with better values than that. I will have to dig out my iota memo…
According to the spreadsheet, Steve sent today, aluminum coatings can be well represented simply by a factor of 0.9. I added this approximation to the graphs, and the new graphs are zoomed in to the fainter side of V magnitudes. The thicker line on each graph came from a calculation where the aluminum coatings were represented as flat 0.9 reflectance. As expected, on each graph they basically coincide with the line representing the “new aluminum” curve.
Please notice that I changed the title of the graphs, because strictly speaking this calculation gives the number of photo electrons. These are not necessarily detected, they could be lost in the electronic noise of the camera. So when the numbers are on the low side, actual detection will depend on the quality of the camera.
In the end I think, the new thick line on the graph can be taken as upper limit for detected photons.
B0zoom.pdf F0zoom.pdf M0zoom.pdf
Judit 2011-11-23:I made two new plots, which I think could be useful:
One to show the difference between the QE of the currently used tip/tilt CCD and two of the curves from Xiao's presentation.
The other plot is showing how the number of photoelectrons, this time on a single pixel assuming 25 actuators and 4 pixels for each sub-aperture, is dropping as a function of V magnitude. I calculated this using the QE curve labeled BV for three spectral types, F0 doing the worst and M0 is the best. These values are the maximum number of photoelectrons, assuming that all the light up to 1000 nm is reflected toward the WFS. phelperpix.pdf
Note that the slightly better QE curve does not make much difference compared to the QE of our current CCD. The significant difference from
previous plots is a division by 100, as this is for a single pixel when we distribute light to 100 pixels.