SIS problem with XTE update 3.6.1 of FTOOLS

Updated 1997 April 25

We have discovered a problem analyzing ASCA SIS data using XTE update 3.6.1 of FTOOLS, released on 1997 March 4. If you are using this version of FTOOLS, please read on.

The nature of the bug

For some instruments, the first channel in the PHA file is called channel 0; in others, channel 1. We have made a concerted effort to improve the consistency of our files and introduced more rigorous checks in many of our software. There has been an inconsistency in that one tool, rbnpha, called the first channel 'channel 1' in its output pha file regardless of the definition in the input.

A new version of rbnpha has been included in the XTE update 3.6.1, which respect the channel naming convention of the input file. If you are using this version, the value of TLMIN1 keyword in the first extension of the output pha file will have the value 0 (1 for older versions).

However, due to an oversight, one element of the accompanying changes to the SIS response matrix generator, sisrmg, was not included in the same update.

If you extract an SIS spectrum using FTOOLS 3.6.1, generate the response file using sisrmg, and fit the spectrum in xspec, this introduces a spurious one channel shift. The fit will be poorer than it should be and some parameters will be incorrect.

The necessary update to sisrmg has already been made, along with some other changes; the new version is being tested for inclusion in the next full release of FTOOLS, expected in 1997 May.

How can I tell which version I'm using?

You can find out the version of the FTOOLS you are currently running with:

> fversion

However, this will not distinguish between the original v3.6 and the updated v3.6.1.

The best way to find out if you are affected by this problem is to look into the keywords that are in the data and response files (this method works also for figuring out if you need to redo your analysis that you carried out several months ago, if you kept the files).

The spectrum: Find out the value of the TLMIN1 keyword.
```
> fkeyprint sis0.pha tlmin1
```
If the value is 1, the file was produced with v3.6 or older; if it is 0, it was produced with v3.6.1.
The response: Find out the value of the RQFRST and TLMIN4 keywords.
```
> fkeyprint sis0.rmf rqfrst
> fkeyprint sis0.rmf tlmin4
```
RQFRST records the value of the TLMIN1 keyword in the spectrum that was used as input to sisrmg. If this keyword value is 0, TLMIN4 must be present and must be set to 0; if 1, TLMIN4 must be set to 1 or be absent. Other combinations indicate a problem.

The Workaround

As stated above, the next release of FTOOLS will contain a new version of sisrmg with this bug fixed. In the meantime, if you are running xspec 9.02 or later, or if you are willing to upgrade your xspec to 9.02, the you can simply insert the necessary keyword into the rmf file produced by sisrmg as follows:

> fparkey add=yes
Keyword value[] 0
Name of FITS file and [ext#][] sis0.rmf+1
Keyword name[] TLMIN4

(if you called the response sis0.rmf).

If you are using an older version of xspec, there is a different workaround.

Make a temporary copy of your pha file (we shall call this foo.pha below).
Fudge the TLMIN1 and TLMAX1 values in the first extension:
```
> fparkey 1 foo.pha+1 tlmin1
> fparkey 512 foo.pha+1 tlmax1
```
(The value of TLMAX1 depends on the mode: 512 for BRIGHT, 1024 for BRIGHT2.)
Generate the response using sisrmg with foo.pha as the input file.
Delete foo.pha

If you have any questions concerning ASCA, visit our Feedback form.

Return to the ASCA GOF

If you have any questions concerning ASCA, visit our Feedback form.

This file was last modified on Thursday, 24-Feb-2011 17:00:13 EST