Testing new multi-channel analyser PHA capability for Cameca SX

[Karsten Goemann]

I've done some testing on our SX100 #846, vintage 2003, using both PeakSight and MCA and "traditional" PHA with Probe For EPMA. I've done FeKa on LLiF with the gain about twice the normal value to shift the PHA peak to higher V and better resolve the escape peak.

Exporting the PeakSight PHA data to ASCII can be done by clicking the "Save" button in the Display WDS window (at least using PeakSight 4.2).

I've attached an Excel plot.

PeakSight exports 256 points starting at 0.56V and ending at 5.54 V. This matches the lowest possible baseline on the SX100 of 0.56V and maximum window of 5V (total of 5.56V). Interestingly, the data has a repeated Y value every 0.156V in X, and all values in between are perfectly on the plot line, so these in between values might only be interpolated and not real MCA channels.

PFE "traditional" scans are similar in peak position as the PeakSight PHA scan, better in resolution but the peak shapes are bad.

The PFE MCA scan with range 0-5V is shifted compared to the PeakSight scan, but not by 0.56V as expected, more like 0.8V. So PeakSight might be cutting something off at the high end. This would be supported by the curves in the attached graph where the PFE MCA curve has more "background" on the right of the PHA peak than PeakSight.

I will try to read out the data in PFE's verbose mode that's actually returned by the SX.

[Karsten Goemann]

So below is what I get in debug mode, same settings as before (FeKa on LLiF...). Completely different baseline and window values returned, so maybe they're just random? Is it somehow possible to view the actual 256 data points that are returned?

Acquiring PHA acquisition...
Setting spectro  5 PHA (BWGBMD):  .56 4.99 700 1824 -1 .000003

SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 0
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 1
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 2
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 3
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 4
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 5
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 6
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 7
SX100GetPHADistributionMCA2: Motor= 5, Status= 0, MaxTries= 8
SX100GetPHADistributionMCA2: Motor= 5, Acquisition number=1
SX100GetPHADistributionMCA2: Baseline= 1517
SX100GetPHADistributionMCA2: Window= 1125
SX100GetPHADistributionMCA2: Alignment= 0
SX100GetPHADistributionMCA2: Index= 256
SX100GetPHADistributionMCA2 (Byte): Min= 0, Max= 254, CountTime= 1.555
SX100GetPHADistributionMCA2 (PHA Sum): Min= 0, Max= 249.8571
SX100GetPHADistributionMCA2 (Averaged): Min= 0, Max= 249.8571
SX100GetPHADistributionMCA2 (Normalized): Min= 0, Max= 160.6798

SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 9
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 10
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 11
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 12
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 13
SX100GetPHADistributionMCA2: Motor= 5, Status= 1, MaxTries= 14
SX100GetPHADistributionMCA2: Motor= 5, Status= 0, MaxTries= 15
SX100GetPHADistributionMCA2: Motor= 5, Acquisition number=2
SX100GetPHADistributionMCA2: Baseline= 1517
SX100GetPHADistributionMCA2: Window= 1125
SX100GetPHADistributionMCA2: Alignment= 0
SX100GetPHADistributionMCA2: Index= 256
SX100GetPHADistributionMCA2 (Byte): Min= 0, Max= 254, CountTime= 1.308
SX100GetPHADistributionMCA2 (PHA Sum): Min= 0, Max= 496.1429
SX100GetPHADistributionMCA2 (Averaged): Min= 0, Max= 248.0714
SX100GetPHADistributionMCA2 (Normalized): Min= 0, Max= 189.6571
PHA acquisition complete

[John Donovan]

So below is what I get in debug mode, same settings as before (FeKa on LLiF...). Completely different baseline and window values returned, so maybe they're just random? Is it somehow possible to view the actual 256 data points that are returned?

Just turn on "VerboseMode".

Yeah, clearly your data actually is not ranging from 1.1 to 2.6 volts!

[Karsten Goemann]

OK, I've done another test in verbose mode now and am getting the 256 individual values listed in the PFE log window. It shows the same behaviour that Y values are repeated every 8 channels.

If I use the following formula to convert channel number to voltage PFE and PeakSight MCA match (see attached screenshot of plots):

X axis value (V) = 0.56 V + channel number * (5V / 256)

Channel number is zero to 255. So this confirms 0.56 V baseline and 5 V window as MCA range.

This works for our SX100, which is SX#846, installed 2003, 68030 mainboard, Stage, WDS, and Scanning all "Old" as listed in Cameca config.

However, it might be different for newer SX100s and the SXFive, where the minimum baseline appears to be dynamically adjusted by the SX depending on x-ray energy (or spectro position).

[John Donovan]

Wow!  Well I'm *almost* convinced that it's 0.56 to 5.56 volts!

Gareth do you see something similar on your instrument?

[Gseward]

Karsten, John

You show me yours and I'll show you mine!
Looks like I have newer hardware on SX921 since my baseline for MCA acquisition is 'displayed' as 0V (should there be a 'noise' signal at low mV???)

Here is the ascii output from peaksight (full content attached). I deliberately set the gain low so that there are pulses at low mV (i.e. below the ~0.5mV baseline that you are seeing).

Interesting points to note:

1)there are pluses down to the first channel (0-19.5mV)

2)X = : mV (0 to 5000

3)despite the 8 bit cts (0 to 255), no single channel has 255cts - I thought this was the criteria for termination of the testPHA command?
Edit by John: I too have noted this...

4) MCA runs min mV to max mV regardless of base line and window that might be set in PFE or Peaksite. This
seems to be just the way it is.  Presumably this is because the 8-bit MCA channel width is fixed in mV range to ~19.5mV per bit, and hence there would be no advantage to collecting a smaller V range than the max i.e. you can't improve on the ~19.5mV resolution.
Edit by John: That would make sense...

5)peaksight restricts my setting of the baseline , as you describe e.g. 597mV for LTAP on Sp1, 674mV for TAP on Sp2 etc. (this value must be stored somewhere? I would like to edit it - the current value is displayed in Configuration/Crystal but the limit is not set there, as far as I can tell). My window is restricted, as you mention, but in my case to BL + W = 5V.  The dynamic baseline that Karsten mentions is called diff. Auto mode in Peaksight.
Edit by Gareth: Dynamic Baseline is a function of detector Gain. See discussion below.

On the older SX100 hardware are Count Diff. and Count integral both returned, regardless of which mode is being used?

Nothing in life is straight forward!

Gareth

[Gseward]

ascii data attached.

[Karsten Goemann]

I'm not sure this "dynamic baseline behaviour" is the same as Peak Sight's diff. auto. At least on our older SX100 the latter sets much narrower windows.

What I mean is that the minimum baseline setting you can set in both PHA integral and diff is not constant 560 mV as on older SX100s, but a value that depends on the spectrometer position or x-ray energy.

That's at least what I saw on the SXFive.

I.e. you wouldn't have a constant minimum baseline value per crystal, but the value should change depending on where you are in the spectrometer range.

I don't think these values are user adjustable.

In any case it looks like for the newer SXs it the MCA range might be 0-5 V which makes sense because then this "dynamic baseline" behaviour can be ignored for PHA scans for the sake of simplicity.

It's also interesting that in your ASCII data the count value every 8th channel is still a duplicate of the previous value and the "spacing" between those values is 0.156V. So that's the same as with my instrument.

Gareth, could you do a PFE MCA scan in verbose setting and save the corresponding PeakSight PHA scan (they might be identical as PeakSight seems to update the PHA scans display even when collected by PFE)? Then we could really compare directly.

Cheers, Karsten

[Gseward]

On the topic of energy resolution (perhaps a question for someone from Cameca?) :

do we see differences between a 'traditional' PHA scan (with a fixed window and moving baseline) and the MCA scan because the two different methods are using different pulse processing hardware/signal paths ?
e.g. a physically separate single-channel analyser is being used for the 'traditional' scan?

or is the same MCA used for both, with the reduced resolution being the result of different pulse processing between the two methods (e.g. shaping time etc.)?

I guess I could look at the schematic and perhaps find the answer!

Gareth


 

[Gseward]

I'm not sure this "dynamic baseline behaviour" is the same as Peak Sight's diff. auto. At least on our older SX100 the latter sets much narrower windows.

What I mean is that the minimum baseline setting you can set in both PHA integral and diff is not constant 560 mV as on older SX100s, but a value that depends on the spectrometer position or x-ray energy.

That's at least what I saw on the SXFive.

I.e. you wouldn't have a constant minimum baseline value per crystal, but the value should change depending on where you are in the spectrometer range.

I don't think these values are user adjustable.

In any case it looks like for the newer SXs it the MCA range might be 0-5 V which makes sense because then this "dynamic baseline" behaviour can be ignored for PHA scans for the sake of simplicity.

It's also interesting that in your ASCII data the count value every 8th channel is still a duplicate of the previous value and the "spacing" between those values is 0.156V. So that's the same as with my instrument.

Gareth, could you do a PFE MCA scan in verbose setting and save the corresponding PeakSight PHA scan (they might be identical as PeakSight seems to update the PHA scans display even when collected by PFE)? Then we could really compare directly.

Cheers, Karsten

OK, so you have Diff Auto too.  I only just realised that the 'dynamic' baseline set-point is a function of detector Gain (not spectrometer position 'directly').  Thanks for bringing it to my attention. I've been fighting with this for a while. I guess moving the baseline 'up' as the gain increase is designed to prevent electronic noise being included. However, I find the escape peak is sometimes truncated by a cameca imposed (aka dynamic) baseline; here is a good example where increasing the Bias and reducing the Gain, would be a better option.

[Karsten Goemann]

You're right, I dug out my notes on the SXFive again and I've actually collected lower baseline limit vs. gain curves, which is a linear correlation. I didn't have much time for that when working on the instrument.

I think you're right the reason would be cutting off the low energy noise.

[Gseward]

For what its worth:

Gain max = 4095 (presumably 12 bit DAC), BL = 994
Gain min = 0,  BL= 369mV

Dynamic BL = 0.152625*Gain + 369

[Karsten Goemann]

Nice, I got

baseline (V) = 0.1524* gain + 368.73

for the SXFive.

Could that even be put into emppha.dat if we need to reproduce it in PFE?

[John Donovan]

Nice, I got

baseline (V) = 0.1524* gain + 368.73

for the SXFive.

Could that even be put into emppha.dat if we need to reproduce it in PFE?

You guys are good!

Yes. These coefficients could be added to the emppha.dat file...

[Probeman]

Michel Outrequin offers these words of wisdom:

"The 256 intensities returned by the PHA routine are always from the minimum value (805 mV) to the maximum value (5637 mV)."

So I will change the MCA PHA limits in my code tonight...