For the past year or so I’ve been measuring the times of minima (ToM) of eclipsing binaries and submitting them to var.astro.cz so they can analyze the O-C curves.
A couple of days ago I was measuring FQ Lyn based upon their predicted ToM. However, even though I started an hour preceding the ToM, My light curve capture a maximum instead. It looked like the epoch they were using was not correct.
For the past few days I was lucky enough to have clear skies and the timing was such that I could capture both the primary and secondary minimum (I still don’t have the secondary maximum).
I then added the ASAS-SN measurements to the plot to see what I would get. The period finder in Peranso couldn’t find a good period but when I specified to use the known period I got the following
I interpret this to mean the epoch was different between the two sources. It seems the same period of 0.569226 days works for both data sets. I did do a heliocentric correction on my data before comparing them.
The ephemeris from VSX also shows incorrect minimums, off by several hours.
The question is what to do with this. At the least I should probably tell var.astro.cz to update their epoch so people can accurately measure the ToM. They also say the secondary minimum has the same brightness as the maximum brightness. That is also not correct.
“I interpret this to mean the epoch was different between the two sources. It seems the same period of 0.569226 days works for both data sets.”
I guess you only used ASAS-SN V data and not contemporaneous g observations.
I don’t think it is an issue with the epoch but just the period not being accurate enough. It will look good enough for each dataset but if you used ASAS-SN data several years old and current data (yours), then a small difference in the period value will result in a phase shift between the two datasets. You just have to adjust the period until you get a good fit.
It is weird that Peranso couldn’t find a period, EWs are very easy to solve. Of course, the period thrown by the software will be half the orbital period, so you have to multiply it by two.
“The question is what to do with this.”
You can improve the period (at first try the VSX epoch because it might be okay) and submit a revision to VSX, with a phase plot (with the epoch in the table as phase 0.00) as supporting evidence.
You can also improve the range. Use the ASAS-SN zero point.
“They also say the secondary minimum has the same brightness as the maximum brightness. That is also not correct.”
Probably they meant that the secondary minimum has the same brightness as the primary minimum. It doesn’t make any sense to compare a minimum with a maximum.
I guess you’re right that the period was not quite accurate. However, in order to get a good result I had to specify the search range to be 0.56 to 0.57 and upped the steps to be 10,000. It did then come up with 0.56921865 which yielded this phase plot using the ANOVA option.
Not a lot of difference between that and the published 0.569226 but I guess with a timespan of 5095 days between ASAS-SN’s data and mine, a small discrepancy can add up.
I did originally use the VSX period and that is what produced the shifted graph in my first message. I will see if I can figure out how to submit a revised period.
Another way to refine an assumed period that is close to the true period if there are data spanning many (and I mean a very large number of) cycles is to plot an O-C diagram to obtain the cycle count, then plot an ephemeris (the actual times of minimum against the cycle number). Provided that the period is constant the slope is the period and regression analysis will give the error of the estimate.
If the period is changing at a constant rate the O-C plot will be a parabola (of course, enough data have to be present to do this accurately). A quadratic function can then be fitted to the ephemeris plot. The slope of the parabola at zero epoch will be the period at that epoch. This slope is the coefficient of the first differential of the 2nd order term of the fitted function. The cycle number can be shifted arithmetically to make any epoch of interest zero.
A simpler solution, if the period is changeing slowly, is to treat the ephemeris plot as linear (it is, after all, a straight line) and assume the slope is the period.
Your epoch of primary minimum appears to be incorrect in all your plots. You should be able to set your epoch to the middle of the primary minimum and get the period “correct” topether.
If I remember peranso had the Kwee & van Woerden method to determine minimum time. The AAVSO software repository also used to have a Kwee & van Woerden tool.
You are correct that the epoch was not set to the primary minimum. The purpose of the graph was just to show the overall light curve. It wasn’t until refining the period that I got the two data sets to align.
