RE: Data Disk Format

["Gutzwiller, Michael" <mgutzwiller@lanvision.com>]

The version of Image Scientist Tom is using ignores the BSCALE and BZERO
values and displays the actual data values.  That's why Tom's numbers are
signed integers.

Thanks,

Mike G.


-----Original Message-----
From: Creager, Robert S [mailto:CreagRS@LOUISVILLE.STORTEK.COM]
Sent: Friday, December 22, 2000 1:50 PM
To: 'Tom Droege'
Cc: 'Tass Mailing List'
Subject: RE: Data Disk Format



Hey Tom,

Raw-to-fits is adding in a offset of 32767 in the fits header, not the data,
and that explains the difference in means.  Programs designed to read the
fits header will add that value to each point.  ImageScientist must not be
looking at the fits header values, or maybe it's an option.  What I was
trying to figure out is how the sigma was so different between FTOOLS and
ImageScientist.  Adding (or not) the 32767 has no effect on sigma.  When I
look the data for a dark and subtract 32767 from every pixel, the mean I
calculate is -25119, but the sigma is still 716...

Cheers,
Rob

> -----Original Message-----
> From: Tom Droege [mailto:tdroege@veriomail.com]
> Sent: Friday, December 22, 2000 11:25 AM
> To: Paul Repacholi; tass@listserv.wwa.com
> Subject: Re: Data Disk Format
> 
> 
> I am just trying to point out that there is a format error 
> somewhere.  I am 
> writing 16bit signed integers in the raw data file.  RawToFits does 
> something with them, I do not know what, and previously this 
> has not seemed 
> to be a problem for anyone.  At this point, I do not care 
> that the .fts 
> header is wrong.  It is no doubt wrong.  The problem of the day is to 
> figure out how to use existing programs to analyze existing data.
> 
> I am doing my best to use the full range of my 16 bit signed 
> ADC.  So I 
> have set zero at about -25000 counts.  I am not claiming to be a 
> spotlight.  ;^)
> 
> The real data then goes from the vicinity of -25000 counts for the 
> pedestal, to 32767 for saturated full scale.
> 
> Tom Droege
>