SANS: bug in SaveCanSAS1D

[Anders Markvardsen]

Steve writes (note the below was the result of analysing sans2d00011585.nxs):

My colleague at ILL has just found what appears to be a bug in SaveCanSAS1D.

So if you process the attached SANS2D file and write the 1D as CanSAS output the file reports SASdetector_SDD 11.827

But if you examine the file the rear detector was at 3.024m (and the front was at 1.419m).

[Anders Markvardsen]

Observed so far:

Looking at the code it is hardcoded to output information about the first detector in the workspace only, which for both LOQ and SANS2D is a monitor – so 11.827 below refers to a monitor to sample distance.

Note this is somewhat indirectly reflected in the SaveCanSAS1D output, which for the SANS2D run 8672 is:

<SASdetector>

<name>monitors</name> <SDD unit="m">12.098</SDD>

</SASdetector>

emailed steve

[Anders Markvardsen]

Removed hardcoding of SASdetector element in savecansas1d. re #5268

Changeset: c927b81edd0204e002af039dc04c1872dbbf5806

[Anders Markvardsen]

Removed hardcoding of SASdetector element in savecansas1d. re #5268

Changeset: 06f53c519fba8295f6b2d777b7430547c49963e9

[Anders Markvardsen]

To test:

1. Do any isis sans 1d reduction

2. Go to Geometry tab and take a note of the 'rear-detector' and 'front-detector' positions

3. Go back to the 'Run Numbers' tab and tick 'CanSAS' and click 'Save Results'

4. open the xml you saved and find the section:

<SASdetector>

<name>front-detector</name> <SDD unit="m">1.87565</SDD>

</SASdetector> <SASdetector>

<name>rear-detector</name> <SDD unit="m">6.00861</SDD>

</SASdetector>

The distances here to should equal those in the Geometry tab as 0.001*sqrt(X^2+Z2)

[Martyn Gigg]

Trying with Mask file: systemtests/Data/SANS2D/MASKSANS2D.091A and run 992 (in systemtests Data) I can't get the numbers to agree in the geometry tab vs the file.

On the geometry tab I have:

rear-detector:
 X 100.1
 Z 3999.1
front-detector:
 X -1050.0
 Z 2501.1

Is it just a precision thing?

[Anders Markvardsen]

Not I am just putting in a comment for now this before I loose this information (and my mind).

From debug I get from the SaveCanSas1D.cpp sample position at

sample z = 19.280999999999999

Front det z = 21.755391989560000 y = 0.17649999999999999 x = 0.98915719301400007

Rear det z = 23.338072000000000 y = 0.19650000000000001 x = -0.084199999999999997

which gives e.g. the sample to detector distances using the cpp line in SaveCanSAS1D.cpp:

double distance = comp->getDistance(*m_workspace->getInstrument()->getSample());

(which looks perfectly OK)

Front Det - Sample = sqrt((21.755391989560000-19.280999999999999)^{2+0.17649999999999999}2+0.98915719301400007^{2) = 2.6706}

Note the Beam Centre is (84.2, -196.5).

For now this comment is just for info

[Anders Markvardsen]

Further tests shows for LOQ the match is OK. For example LOQ run 54432 in Test/systemtests/Data/LOQ.

SANS2D does not appear to match. One explanations for this may be: 1) that X and Z coordinates are shown (which at same level appear to be based on log values), but in fact there may also be a none zero Y value...

A follow up ticket to investigate this has been created #5698

[Martyn Gigg]

True. The file is showing the correct values though, which is most important.

[Stuart Campbell]

This ticket has been transferred to github issue ​6114