[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: No Escape - can we comput an allowable timming error?
I'm thinking out load here. The idea is that there
should be some way to derive a time accuracy requirement
from 1) our self imposed goal of 1% photometry and
2) the highest expected rate of change in brightness of
Let's suppose we have a fast period variable. What is
highest rate of change of brightness expressed in
magnitudes per second? I have no idea what this would
be. I'll guess a binary star could run at 1 mag per second
at the some part in the light curve. If your goal is
1% photometric error then drawing the data point .01 second
"off" on the time axis gives a .01 magnitude "bump" on the
I think the way to look at a 30 second expose is not that
you have a 30 second uncertainty in the instantaneous brightness
but that you know the integral of photon emission over
time for a 30 second integration. For a periodic function
knowing the integral is quite useful.
I think it begins to make
a difference when you see only the part of an eclipse So fractional
seconds may matter if you want to know what fraction of that 30
second expose was during a transit. You may want to know the
time down to 1% of the expose length.
I'm not the person to work out the details but I do think that
given a goal of XX error in a light curve we can deduce an allowable
Arne's 6 second per day drift is typical of PC clocks. They are
awful. A $2.00 watch you buy at the supermarket checkout is
100x better. You have to do something. The simplest thing
is to buy a $2.00 sticky back digital watch and stick it on
your CRT. Then set the time by hand every day or two. Even that
is better then just using the built in clock.
> -----Original Message-----
> From: firstname.lastname@example.org [mailto:email@example.com]
> Sent: Wednesday, March 14, 2001 8:36 AM
> To: firstname.lastname@example.org
> Subject: Re: No Escape
> Hi, Oscar!
> The accuracy needed is often called into question on
> this list. My feeling on time, as I continue to state,
> is that this is one parameter for which we can get really
> good accuracy. So, you should do the best job you
> can within reason. Certainly +/- one second at
> the time you set your PC clock is easily obtainable.
> You should know the drift of your clock
> so that you can continue to keep this one second
> accuracy at any time during the night. If you have
> NTP or equivalent, then the time will be accurate
> to a few milliseconds or better.
> Do we need this accuracy? For the minute-class
> exposures, knowing time to +/- one second is probably
> overkill since you have +/- 30 seconds timing error
> just from the open shutter time. For short exposures
> of a few seconds (possible with the Mark IV), you
> might want even better accuracy than a second. So the
> bottom line is that, since you have control over time,
> make it as accurate as you can so that you can forget it
> as an error source in the future.