Get Net Intensity Issues

[John Donovan]

I fixed a minor rounding error in the calculation of net intensities from the Thermo EDS spectra which caused the k-raws for the assigned standard to not be exactly 1.000.  This was pointed out by Ken Severin, thanks!

All fixed in v. 10.2.2 of PFE.

[Probeman]

Just discovered a subtle and hard to find problem that can occur if one is using the RealVNC server for remote access. Typically when VNC installs as a server or even when using the VNC Viewer client, port 5800 is utilized for the communication.

However, as you probably all know, the TE Portal interface that Thermo provides for integrated EDS/WDS also uses port 5800, so when one tries to use the Thermo API, it fails with an

Error 5: Socket Time Out Error

So, finally I realized this and changed the port that VNC uses from 5800 to 5900 (note there are two fields that must be changed, one for VNC communication and the other for Java communication), it all started working fine.

This was on an off-line computer used for re-processing data, but Probe for EPMA still needs to communicate with the Thermo NSS software for obtaining the net intensities.

When this is all working one now gets the proper net intensities as expected:

[Gseward]

John,

Here is a Net Intensity issue that I do not understand. I'm using a Thermo NSS system.

To make for a more efficient spectrometer configuration, I thought I would use EDS for some majors in an analysis setup.

When I looked at the calculated compositions for some unknowns, I was surprised by some erratic totals, so I thought I'd look at the EDS elements.

Here are two EDS spectra from a line scan:




Note that although the spectra and the acquisition parameters are very similar, the net intensities for Si are very different. Mg and Fe look OK. 

To confirm the similarity of the spectra I exported the spectra and plotted the area around the Si peak (log intensity vs eV). I'd say they look pretty similar:



I have to admit that the processing of the EDS spectra is something of a 'black box' to me. So, any suggestion about what I could be doing wrong?

Cheers,

Gareth

[Probeman]

I have to admit that the processing of the EDS spectra is something of a 'black box' to me. So, any suggestion about what I could be doing wrong?

Nice documentation of the issue, but I think I should send a link of this to Thermo.  Keith Thompson or Greg Fritz.     

Why?

Because the Thermo NSS software does the peak stripping for PFE. I send the conditions and the spectrum intensities (and of course the elements to get the net intensities of), and they send me back the net intensities- no error statistics though, and that would be nice!.

But you might also try and get their latest NSS version because there were some issues they fixed I think related to this.

[SteveSeddio]

Gareth,
That's interesting...
If you send the .emsa files to me, I'll happily take a look... And then bother Keith and Greg about the issue if I can't figure it out.

stephen.seddio@thermofisher.com

[John Donovan]

Gareth,
That's interesting...
If you send the .emsa files to me, I'll happily take a look... And then bother Keith and Greg about the issue if I can't figure it out.

stephen.seddio@thermofisher.com

I did get an email from Keith that he thinks it has to do with the auto ID messing up and adding in elements that aren't there. He said they will fix but until then he said to just disable elements that are misidentified.

Odd to me because the NSS auto ID is usually very good!  But it also seems to me that an easy fix would be to have NSS force the auto ID to the elements passed, and then look for other unidentified peaks.

Of course as you know the modeling of the continuum (for getting the net intensities) means that one has to know the composition, therefore the need for auto ID!

[Gseward]

Indeed, I discovered that in this case the variation in Si ka Net Intensity was due to the misidentification of Ta M-lines. After applying a mask for all but the elements that were present, the Net Intensity returned for Si for the two spectra mentioned in the original post were more reasonable:

Line = 258
EDS Spectrum Net Intensities:
mg ka, net intensity=10858.76, net background= 0, net (cps) error= , net error= 37.12778
si ka, net intensity=7633.333, net background= 0, net (cps) error= , net error= 25.07222
fe ka, net intensity=1213.2, net background= 0, net (cps) error= , net error= 5.661111

Line= 260
EDS Spectrum Written to OEM System...
EDS Spectrum Net Intensities:
mg ka, net intensity=10861.81, net background= 0, net (cps) error= , net error= 37.13889
si ka, net intensity=7642.511, net background= 0, net (cps) error= , net error= 25.1
fe ka, net intensity=1216.611, net background= 0, net (cps) error= , net error= 5.666667

That's 5623cps with Ta, 7633cps without. I am a little surprised by the magnitude of the error created by an apparent 0.5wt% Ta?

Anyway, I wonder if it is possible that I am seeing misidentified peaks because I do not have my Thermo system 'optimised' in some way?
However, if misID of peaks is a problem, then it raises some interesting questions about reprocessing data at a later date. Since there is currently no scope for a 'mask' to be sent from PFE vis the NSSPortal, I will be suggesting to my users that they create a new NSS project called PFExyz (with appropriate mask) for every PFExyz.mdb they create, and to make sure that this NSS project is current in NSS before reprocessing occurs.  Another option here would be to save a template.TNP file (with an appropriate name) that could be used to quickly recreate the mask as needed.

John, I wonder (in a world of endless time and resources) whether the:

TEVB ProjectExplorerDisplay

function could be sent the first time EDS data is calculated in a .mdb, to force the user to make a selection of an appropriate project??

From TEPortal doc:
TEVB ProjectExplorerDisplay
This function is used to instruct NSS to display its project explorer dialog. This function will block until the dialog has been closed.

Gareth

[John Donovan]

John, I wonder (in a world of endless time and resources) whether the:

TEVB ProjectExplorerDisplay

function could be sent the first time EDS data is calculated in a .mdb, to force the user to make a selection of an appropriate project??

From TEPortal doc:
TEVB ProjectExplorerDisplay
This function is used to instruct NSS to display its project explorer dialog. This function will block until the dialog has been closed.

I don't see why I couldn't do this at the EDS interface initialization... especially if it's a solution for these auto ID problems you are seeing.

I will look at this when I get back from GSA.

[Gseward]

Update: John has now implemented the TEVB ProjectExplorerDisplay function.

This will prompt the user to select a suitable NNS folder before sending spectra to NSS for GetNetIntensity processing. If nothing else, this serves as a reminder to the user to 'think before acting' . Thanks for this John, it is very helpful.

Gareth