[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: GJD to HJD
Robert C. wrote to me:
>So, before going beyond calculating WS statistics, (like to orbital
>determinations), the calculations should be done with HJD?
Yes. For every datapoint you plot, you have two types of
errors: the photometric ones, that move a point up/down on most
plots, and the timing ones, that move a point left/right. That is,
you are dealing with at least two dimensions.
If you don't include HJD so that the light-travel-time error
is removed, then you have a systematic error in the time
dimension. This can be very important, depending on the application.
>We _can_ know the shutter open/close times to a few milliseconds
>Tom's data may not be that good but it is very easy to make it
>that good. Microseconds are hard but anyone with a modem
>connection to the Internet can get milliseconds. If you record the
>time of the STAMP's confirmation message for the open and close
>commands and subtract the transmit time (at 9600 baud) you can get
>to milliseconds not that you really need it.
Sorry, you are wrong on this one. The eventual problem is
the shutter itself. It takes a finite amount of time after the
command is sent before it starts to move, and before it finishes
moving, whether the command is to open or close. This amount of
time is not necessarily constant, either; for instance, it depends
on how taught the cables are. Likewise, because it is a two-vane
shutter, there is a different exposure time across the field of view.
In the end, such careful knowledge on a several-minute exposure is
not *that* important, since the exposure time smearing itself
is the limiting factor. Where you need such precision is when the
exposure times are short, and I just avoid such short exposures
with ARNE. On the other hand, I do agree with you that those
things that are predictable, such as the difference in timing between
the two shutters, should eventually be removed. Tom sent me one
of the BX24 boards and I must confess I haven't done much with it,
but the intent was to improve the Stamp operation. I'll get to it...
>I'm one of those nuts who thinks that if I can gain 0.02 seconds
>by only the cost of typing in a dozen lines of code, I'll type it
>in. You only do it once.
>So a one second error in HJD conversion _could be_ the largest error
>source in in the time.
My one-second comment was just because I haven't thoroughly looked
at Patrick's implementation. It is certainly much more accurate than
this, but I didn't want to quote a number off the top of my head.
Remember, SLALIB is the astronomical library used by IRAF as well
as STARLINK, and used by many researchers across multiple wavelength
disciplines where accurate timing coordination is important. If
there were significant errors in its routines, we'd know about it.
If you roll your own GJD->HJD routine, you deserve the errors that