Affiliation
American Association of Variable Star Observers (AAVSO)
Mon, 12/29/2014 - 17:37
The telescope I typically use for photometry will have its mirrors recoated in the near future. Obviously, my transformation coefficients will change and must be redetermined. My question is: how stable will the new coefficients be in the months following, as the coatings begin to age?
Hi Tom,
My experience is that the spectral reflectivity doesn't change, but that there is an overall degradation of the reflectivity with time. That might decrease more quickly at the beginning, but it depends on the coating and the environment. I have the most experience with bare aluminum and not any overcoated mirror. With bare aluminum, I didn't see any statistically signficant change in coefficients between a cleaned, but old, mirror coating, and a new mirror coating. The zeropoint changes dramatically as the overall reflectivity improves with the new coating.
You should experiment, though! It would be a good JAAVSO paper.
Arne
I recomend to those observers whom I mentor that they should plan on recalculating their TC's at least once per year, regardless of thier system; the results are seldom the same as the previous year.
Tim Crawford, CTX
Mentoring Team
Time variation of transforms are of great interest to me, and I plan to follow the values closely after the re-coat. I may monitor the transform stability of a second instrument, as a parallel control. The key will be to use the same calibration stars over a period of months.
Looking over the PEP Newsletter archives, I have found two references to significant transform changes on the time scale of months. Attached is a graph of eV for several months after a system had its mirror cleaned (note that the values are x10). Another observer saw his eV change by a factor of two in six months. The changing eVs of the respective observers could not all have been measured with the same stars. I have seen an unsettling variation of eV in my own system with different calibration stars during the same year.
Hi Tom,
I find those numbers very hard to believe, and would take them with a big grain of salt. I've been doing transformations for 40 years with lots of telescopes, and never have seen anything like that.
What "calibration stars" are you using, since you mention you were getting changes depending on what calibration stars throughout the year? I'm wondering if you are using true standards, such as Landolt?
Arne
Arne:
I am using the AAVSO PEP red/blue star pairs (attached), which have hopefully been well-vetted. Can you comment on how much short-term variation in measurements would be expected?
Hi Tom,
The red-blue method is a quick and dirty way to do transformations, as it only requires two stars. Note that this is an "exact" calculation, and not a least-squares calculation, so any error in measurement of the stars, or in reproducing the spectral response of the stars, is not averaged, so there is significant potential for systematic effects. What this means is that, for any pair of stars, you may get great reproducibility in the value of the coefficient, but if you move to another pair of stars, you may again get great reproducibility, but with perhaps a different value than what you got using the first pair.
If you want to quantify any coefficient change, you need to compare apples with apples. You should either use the same pair for all of your measurements over the time interval, or else you need to use many pairs each night and average your result so that systematics are removed.
The red/blue pair list looks like it was originally created from Johnson's 1966 catalog; the magnitudes and colors look good to +/- 0.01mag or so using newer photometry. Again, stars are not black bodies, so the coefficient you get with one pair will usually be different than the coefficient from another pair at some level; you may include a faint companion in your aperture that Johnson did not include in his aperture, etc.
Try to be as systematic as possible in your experiment - I look forward to the results!
Arne
Arne:
Here is an example of variation from the fall of 2014, using the Aquarius red/blue pair. Data were taken on three nights, with the V extinction being 0.18 or less each time.
1. eV = -0.0061, std dev of delta v = 3 mmag
2. eV = -0.0047, std dev 3 mmag
3. eV = -0.0083, std dev 7 mmag
On the same night as (2), I also measured eV using the Pegasus pair as: -0.052, std dev 5 mmag. While the difference between Aqr and Peg may seem absurd, it is consistently absurd with measurements I made in 2013.
The focal lengh of my instrument is in excess of 8000mm, so it seems unlikely I am pulling in faint companions that Johnson did not see.
Hi Tom,
Photometry is an experimental science, not an exact one. You have to understand the potential errors to decide whether such differences are important, and if they are, how to overcome the issue.
In this case, the Aqr pair have a color difference of 0.99. The modern magnitudes are very close to the values in the table. On the other hand, the Pegasus pair have a color difference of 0.44, half of the Aqr pair. At the same time, the modern magnitudes differ slightly, so the magnitude difference between the two stars changes from 0.20mag to 0.18mag. This may seem like a small change, but combined with the smaller color difference, the resultant coefficient determination can easily result in a value of 0.05mag/index. That is why you need to determine your transformation coefficient from several different pairs, not just one, and average the result.
Both determinations are "right", in that they result in values that are appropriate for the input data. At the same time, they are systematically different, which is why you try to randomize everything you can. Or, only use one pair if you are studying long-term behavior of the coefficient as the coating ages.
Arne
Arne:
Could you give us a quick rundown on the potential sources of error in eV determination?