Applying DeadTime corrections to SANS

[Gesner Passos]

Further to our Mantid SANS meeting the week before last I have been doing some reading around on deadtime corrections.

The Mantid algorithm ApplyDeadTimeCorr, which I think was written for Muons?, is the appropriate correction to apply for a Poisson counting process.

http://www.mantidproject.org/ApplyDeadTimeCorr

There are however 2 caveats, and I recommend that the algorithm description page be updated accordingly:

i) That the second term in the denominator is <=0.1 (ie, 10% deadtime); this is because the correction is based on a Taylor expansion which increasingly diverges with increasing deadtime.

I would further recommend that the algorithm be modified to report to the Mantid Results Log whenever the deadtime exceeds 10%.

ii) That the detector itself is ‘non-paralyzable’; this means, in essence, that the detector will always return counts that increase with the incident count rate no matter what the incident count rate is.

(A detector that actually saturates and stops counting is ‘paralyzable’; a detector whose observed count rate goes through a maximum with increasing incident count rate is termed ‘semi-paralyzable’. The algorithm in ApplyDeadTimeCorr must not be applied to such detectors)

References ‘Image quantificaton with a large area multiwire proportional chamber positron camera’. SR Cherry, PK Marsden, RJ Ott, MA Flower, S Webb, JW Babich. Eur J Nucl Med, (1989), 15, 694-700.

‘Two-Dimensional X-ray Scattering’. BB He. Wiley, 2009.

​http://en.wikipedia.org/wiki/Dead_time

Thus ApplyDeadTimeCorr would seem to be quite easy to incorporate into our reduction process, it simply requiring a table of deadtimes for each detector (ie, pixel). This table could be an external file loaded like the flood files and displayed on the Reduction tab in the GUI with a check box to specify whether or not to apply it. The default can be ‘on’ and a default table need only have deadtimes of 0 seconds (the situation we assume at present).

So the question we then need to answer is, what do we think the deadtimes of our detectors are?

And might we need to use some wavelength-weighted average for this (eg, our gas PSD’s are hit by a much higher flux of short wavelengths but these are less-efficiently detected, and vice versa)?

Steve King - email 28/10/2013

[Nick Draper]

Ownership handed over to Peter

[Stuart Campbell]

This ticket has been transferred to github issue ​9937