Re: HD 145913

[Michael Koppelman <lolife@bitstream.net>]

On Thursday, May 8, 2003, at 10:38 PM, Arne Henden wrote:

>   Once you get to this signal level, you need to worry about systematic
> effects such as how good your flatfielding is.  Also, the quoted errors
> tend to be ~Poisson, so your comp star can look good but your variable
> can be quite a bit worse if it is fainter, on a bad part of the chip,
> etc.  The flickering can be real, but when you start making assumptions
> about low-level variations, you have to test those assumptions in many
> different ways to ensure they are correct.  Kato-san already mentioned
> how the power spectrum of flickering should look like, for example.

I'll have to look that up. I hear what you are saying, though. Just 
because Mira is stating low errors doesn't mean that I'm not 
introducing more noise or uncertainty somewhere else. I want to be able 
to trust my data. I think by reduction process is pretty good. I'm 
doing twilight flats when possible and t-shirt flats but I'm using a 
dimmer source and taking a little bit longer exposures. I'm 
auto-guiding so the stars stay in roughly the same place. They do 
wander around a couple of pixels. My average FWHM is 3.1. My filters 
are almost parfocal, so the focus isn't perfect on all the filters at 
all temperatures but it's pretty good. My site generally sucks, so the 
atmosphere may be a limiting factor for me. That's why I like to look 
at the comp star deviation as my reference. I use multiple comp stars 
always, no matter what anyone says at the AAVSO. ;)

Anyway, I'd like to think that with a nice strong signal like this I 
could trust that a 3-sigma error of 0.01 on the comp star would be a 
good indicator that I could trust a 0.04 variation in the program star. 
It does look like noise to me and I don't know if the seemingly 
significant period of 13 minutes is just an alias or something. My 
exposures were 35 seconds apart.

It's probably noise but I'll give 'er another night or two when it 
clears.

Thanks for your help!

Michael Koppelman