Testing new multi-channel analyser PHA capability for Cameca SX

[Philipp Poeml]

Karsten,

I think I have the same issue here. It just seems I can't set the values right in the probewin.ini. I will try and check with the verbose output as well and plot the curves in Excel.

Cheers
Ph

[Karsten Goemann]

Hi Phillip,

I don't know if you've read your way through the whole thread, but based on the testing that Gareth and I have done on our instruments there might be as few as two different settings for these limits which depend on the instrument generation. But it would be good to confirm this on more instruments.

Range of 0-5V for newer SX100 and SXFive, which appear to have this "dynamic baseline" behaviour, where the minimum baseline value you can set in PeakSight depends on the gain setting, i.e. a linear correlation from baseline 0.369V at gain zero to 0.994V at gain 4096.

Range of 0.56 - 5.54V for older SX100s, where the minimum baseline you can set in PeakSight is always 0.56V. This is basically a 5V range as well - I think the missing 0.02V to 5.56V might just be the width of the final channel (5V / 256 channels = 0.0195V, rounded to 0.02V).

One way to find out your PHA range might be by saving a PHA scan to ASCII in PeakSight (click the Save button in the top right corner of the SX Control | Display | WDS | PHA window, only available in newer PeakSight versions). PeakSight seems to display the last PHA ROM/MCA scan even if it was acquired with PFE, so you can export the data for the same PHA scan in both softwares (in PFE doing both the verbose log window and normal export) and compare directly in Excel, which is how I created the graph I attached to the last post.

One thing we're also seeing on our "old" SX100 is a slight shift in the PHA distribution depending on if it is acquired directly after the Faraday cup is switched out or if the beam has already been on the sample/standard for some time. The counting system seems to have to settle for some time once hit by x-rays. I don't know exactly for how long, might be less than a second. But this is why I when doing these tests always switch the cup out, wait a couple of seconds and then do the PHA scan. After that initial "settling" period the PHA peak doesn't appear to shift anymore.

Cheers,

Karsten

[Karsten Goemann]

PS: I see in Gareth's PeakSight data for the "new" SX100 that it is the same there too, i.e. the voltage range is 0V to 4.98V, not 0 - 5 V, so the missing 0.02V is probably again just the width of the last channel of 0.02V.

[Probeman]

I've tested the new PHA MCA keywords in Probewin.ini a bit.

We really appreciate it!

I can get a perfect match between PFE and PeakSight if I use PFE's verbose output of the 256 channels to the log window, see Excel plot screenshot attached. This is using 0.56V for the range minimum and 5.54V for the range maximum. The blue curve is not visible because it is completely covered by the green curve.

I'm using the same minimum and maximum values in Probewin.ini, but PFE's output of the same PHA scan (using the Export Data function in PHA | Display and Export Scans) seems scaled differently (red curve). This is using 0.1 for count time and 40 intervals.

How are you processing the data into 40 channels from 256? Would it be possible to just use the 256 raw channels and recalculate the x value based on the min-max settings? At the moment I can't set the intervals value for PHA acquisition to 256 without getting the error I've reported before.

Good question. Here is the latest code for the conversion from 256 to n points (it is not uploaded yet, probably later tonight).

SX100GetPHADistributionMCASum = 0#

' Calculate range (total range is 256 values or 0 to 255)
m% = BIT8& / CSng(npoints%)
i% = m% * (n% - 1) + 1
j% = (i% + m%) - 1

If i% < 1 Then i% = 1
If i% > BIT8& + 1 Then i% = BIT8& + 1
If j% < 1 Then j% = 1
If j% > BIT8& + 1 Then j% = BIT8& + 1

' Sum range
temp! = 0#
For k% = i% To j%
temp! = temp! + CSng(bArray(k% - 1))
Next k%

' Average range
temp! = temp! / ((j% - i%) + 1)

SX100GetPHADistributionMCASum = temp!


The new code (will upload tonight) also now handles 256 point acquisitions just fine as seen here:



But I'm still a little dismayed that the MCA method seems to have lower energy resolution than the traditional PHA method as seen here:





Note in the two PHA scans above that both scans peak around 2.5 volts, but the traditional PHA scan shows the escape peak more clearly.

[Philipp Poeml]

Guys,

I tested the .exe John shared on dropbox. This is what I get now:



This is PeakSight PHA scan over the PfE MCA scan. My parameters are:

PHAMultiChannelMin=0   ; Cameca MCA PHA minimum x-axis voltage
PHAMultiChannelMax=5   ; Cameca MCA PHA maximum x-axis voltage

This is for U Ma on a high pressure spectro and Qtz crystal. .1 counting; 40 channels.

This looks pretty good, doesn't it? As expected. I think the new code is good.

I did not try to look at any escape peaks now. But I am happy that the PHA peak is at the right voltage. Basically for non-escape peak studies this is usable now.

My SXR should be a "new" type.

How are your scans going?

Cheers
Philipp

[Philipp Poeml]

Btw, trying with .5 seconds counting and 256 channels I got:



Ph

[John Donovan]

Btw, trying with .5 seconds counting and 256 channels I got:

You must be a very patient man!  I've written the code so the routine continues to average the MCA PHA values for a time period equal to the count time times the number of points, so 0.5 times 256 equals 128 seconds which is a long.... time!

I can only assume that maybe you accidentally tried to move a stage or spectro axis during the PCA acquisition?

I'm more interested in why the Cameca MCA intensities seem to form line segments, e.g.:

[Philipp Poeml]

You must be a very patient man!  I've written the code so the routine continues to average the MCA PHA values for a time period equal to the count time times the number of points, so 0.5 times 256 equals 128 seconds which is a long.... time!

I can only assume that maybe you accidentally tried to move a stage or spectro axis during the PCA acquisition?

No, in fact there was no time difference between 0.1/40 and 0.5/256. It did not take two minutes at all. The acquisitoin time was exactly the same, about 10-15 seconds or so.

I will try again, but I did not move anything; there is also no reason to move something during these 10 seconds.  I can also count the time, if that helps.

[John Donovan]

You must be a very patient man!  I've written the code so the routine continues to average the MCA PHA values for a time period equal to the count time times the number of points, so 0.5 times 256 equals 128 seconds which is a long.... time!

I can only assume that maybe you accidentally tried to move a stage or spectro axis during the PCA acquisition?

No, in fact there was no time difference between 0.1/40 and 0.5/256. It did not take two minutes at all. The acquisition time was exactly the same, about 10-15 seconds or so.

I will try again, but I did not move anything; there is also no reason to move something during these 10 seconds.  I can also count the time, if that helps.

If you track down the cause, please let me know.  I will also try this long count time on some MCA PHA scans of my own.

[Karsten Goemann]

I've tested again on our "old" SX100 #846 with the latest PFE version 10.3.4, downloaded this morning, using settings of 0.56 for minimum and 5.54 for maximum, 256 "intervals".

PeakSight and PFE PHA scans match perfectly for all spectrometers. There are some small differences but that is I think due to PeakSight only exporting the last scan and PFE averaging multiple scans.

It doesn't seem to matter what the actual baseline/window settings are or if it is set to integral/differential, the range returned by the SX always appears to be the same, 0.56 to 5.54V, i.e. a 5V range minus the width of channel 256. I'm confident that this is the right setting and will use it from now.

PFE doesn't let me set the counting time per interval lower than 0.1s which as far as understand means the actual scan runs 25.6 seconds, which seems about right. Within that time it's probably averaged at least 10 PHA scans which seems a little excessive.

Edit by John: OK, I've set the minimum PHA interval time to 0.05 sec, so if the default is say 40 points that's only 2 seconds of total acquisition.

[jared.wesley.singer]

Hi,

Please, can anyone summarize the differences between PHA scans in peaksight and PFE?  On this "old-SX100" using PFE software I obtain the asymmetric distribution of spikelets or extreme narrow distributions, and errors if I choose more than 100 points.  I think I am using always "traditional" PHA, because I have no MCA options in the PHA Properties window.

Thanks in advance,

Jared




 

[Probeman]

Please, can anyone summarize the differences between PHA scans in peaksight and PFE?  On this "old-SX100" using PFE software I obtain the asymmetric distribution of spikelets or extreme narrow distributions, and errors if I choose more than 100 points.  I think I am using always "traditional" PHA, because I have no MCA options in the PHA Properties window.

Note sure what you mean by "spikelets", can you post a pic? The main difference between PeakSight and Probe for EPMA is that PeakSight only has the MCA PHA type while PFE has both the traditional and MCA type.
john

The option to change the PHA acquisition type is found in the Acquisition Options dialog.

Note also, the default PHA acquisition type can be specified in the INI file.

[jared.wesley.singer]

I can not post images. 

I have no MCA in Acquisition Options.   

Can I suggest a greater selection of frustrated and sad faces?

It's quitting time. 

[Probeman]

I can not post images. 

Oh yes you can!

http://probesoftware.com/smf/index.php?topic=6.msg14#msg14

I have no MCA in Acquisition Options.   

Oh yes you do!

http://probesoftware.com/smf/index.php?topic=42.msg883#msg883

Can I suggest a greater selection of frustrated and sad faces? \

Probably!

It's quitting time. 

You got it now!

[jared.wesley.singer]

Many thanks, John  ().  Updated software and a shiny-new tinypics account!  I'll try out the MCA next week ASAP.