News:

:) If you can't see some "in-line" images in a post, try "refreshing" your browser!

Main Menu

PHA settings

Started by Rom, February 25, 2022, 08:51:10 PM

Previous topic - Next topic

Rom

Greetings,
Why PHA settings for the same channel/crystal depends of measured substance?
Optimum PHA will less or more change for the same channel/crystal if we use during PHA calibration different substances.

Could you give recommendations about right steps of PHA calibration for WS collecting when spectrometer works in a wide range.
Thank you.

sem-geologist

#1
A Short answer is: It doesn't!
It depends from counting rate, and if substances have different count rates for given line, then PHA can differ.
I can't talk about Jeol, I have absolutely 0 idea how it works there, but I have 100% knowledge how it works and fails on Cameca instruments. You can find lots of cargo cult in books and here in forums on this subjects by people who clearly never hanged oscilloscope to the raw signals or ever tried to look what is under the lid. Generally it is hard to discuss these issues also without pointing to circuit schematics which unfortunately at least for Cameca is confidential. PHA problems are the consequence of inadequate electric design, which was the state-of-art in late 80ies, but unfortunately was not updated to bring the electronic innovations of last few decades. It is often spot misconception that this is problem of Gas proportional counter - which is not! The old design particularly started show its ugly face with introduction of large crystals making PHA circuits to cope with much higher count rates than it was designed for.

Why do we see PHA shifts in the first place with high count rate? it is because very often there is custom amplification circuit which is fed with bipolar output from pre-amplifier. Bipolar output has shorter pulse than monopolar, but it overshoots to negative V after the pulse. (Look here https://www.amptek.com/internal-products/a203-charge-sensitive-preamplifier-shaping-amplifier and go at bottom to typical waveforms to understand what that means). Single pulse has not so bad overshot, but with higher counting rates there is huge increase in pile up peaks (so the overshot is also proportionally longer and deeper) and the probability for consequent pulse to start in that depression zone increases too - which leads all PHA shift toward lower voltages (see my attached concept figure below), as PHA circuit measures the amplitude from absolute zero, not from the base (in that case in negative V) of the pulse. The solution would be to digitize the output from preamplifier and feed that to FPGA with beefy DSP, which would be able to measure relative pulse heights and do pulse-pile up corrections. A203 is also IMHO inadequate to large crystals and high counting rates. Amptek had already went with 2 new generations in design. That means that pre-amplifiers still stuffed to new generation instruments is already 2 generations behind from what it would be possible to do currently. A203 outputs very long signals (in sense of current electronic designs and capabilities and technology), and that was OK in late 80'ies as there was no fast enough ADC. Actually the pulse is even more lengthened/(holded) later in the pipeline as there were no performant ADC then, and for some strange reasons single ADC is shared between few spectrometers...
Currently it is easy to find the ADC which would go with 12bit 50MHz signals (or even 100MHz), and if budget is not limit there are even much more performant ADC commercially available.

Knowing all the peculiarities I personally use PHA differential mode only in very very very (very, ...) rare cases when other means (interference correction) is not capable to take care of higher order peaks, and I am aware that even in that cases it only minimize, but not eliminate higher orders. PHA in many cases are not able to take directly care of 2nd and 3rd orders... but biggest miss in community is not to realize that it is impossible to completely eliminate even higher order (>=4th order) peaks due to Argon escape pulses which often are directly hidden under/inside our main PHA peak within the selected PHA energy window. Yes You can minimize, but won't eliminate it as the filtering is happening at digital domain (again without circuit diagram it is hard to discuss this).

For wide range WS use the integral mode there PHA is completely ignored, all incoming pulses (from analog to digital domain) are registered. On Cameca side that is easy as background noise is cut out with hardware (simple Zener diodes, or special IC with capability to set the threshold of pulses, still at analog domain) and it never reaches PHA digital domain or influence the dead-time, so all passed pulses are significant pulses. On Jeol it can be a problem (at least some older probe models which are in some books and articles, I have no idea how newest Jeol does) and You have to use PHA as background noise is exposed in PHA (around 0V). Use then as wide as possible PHA window, reducing the gain so that peak from one side at spectra and other side of spectra would fit inside PHA window.

Probeman

#2
Ben Buse had a short topic on this question a while back, specific to JEOL instruments. Unfortunately he linked to some images which are no longer accessible on the web (which is why one should always use the forum image gallery when embedding in-line images). But he does mention the use of "empirical" PHAs which is described in the Probe for EPMA Reference Manual, so there is additional information there.

Anyway the empirical PHA topic is found here:

https://smf.probesoftware.com/index.php?topic=420.0

In any event, it is my observation (in spite of some proclamations to the contrary) that PHA settings for a given spectrometer depend on several variables, including P-10 gas purity and mixture accuracy, barometric pressure, line energy, and count rate.

It is important to note that on JEOL instruments the gain can only be adjusted in "steps", but on both instruments both bias and gain will shift the PHA "peak".

My personal philosophy is to leave the baseline and window settings as "loose" as possible and deal with interferences using the quantitative interference correction. That means on the Cameca instrument to set the baseline to 0.3 or 0.4 volts and the window to 4 volts or so. On the JEOL (which I have rarely used), that corresponds to 0.5 to 9.3 volts or so. This prevents issues from count rate shifting when switching from major (standard) to trace (unknown) concentrations.
The only stupid question is the one not asked!

Rom

Thank you for explanation.
If the general recommendation is to use a wide window, am I right I should change our default settings (all range 0.5-9.3V, base line 4V - supplier recommendation, window 0.1V) to new settings for instance base line 2v, window 5v. In this case I can forget about continuous PHA calibration, just sometimes check it.
What bad consequences should I expect at these settings - low spectral resolution?

Probeman

#4
Here is what I would do if this is not already done:

1. Edit the default baseline and window PHA settings in your SCALERS.DAT file on lines on lines 42 to 54 and set all the baseline values to 0.5 and all window values to 9.3. You only need to have one line for each crystal, so 2 lines for 2 crystals and 4 lines for 4 crystals. The other values can be left zero (PFE supports up to 6 Bragg crystals per spectrometer!).

2. Then edit lines 66 to 71 for the integral/differential mode and set the values for -1 so differential mode is turned on. That is all you have to do (this should already have been done by the person who set up the PFE configuration for your instrument initially).

When you go to set up your run for different elements, just leave the default baseline/window and mode values where they are.  As mentioned in posts above, it's better to leave the baseline/window wide open and then simply adjust your gain and bias to place the PHA peak at about 1/3 of the X axis (roughly 3 to 4 volts on a JEOL instrument and 1 to 2 volts on a Cameca instrument), when checking the PHA settings on a major concentration (high intensity) of the element.

When one goes to a trace amount (low intensity) of the element, the PHA peak will shift somewhat to a higher voltage (~5 volts or so on a JEOL instrument and ~ 2.5 volts on a Cameca instrument) but you won't lose any of the photons due to a too "tight" baseline/window settings!

Too much gain and you only broaden the PHA peak which is of no benefit.

The only reason the baseline/window values were set tight "in the old days" was to try and eliminate spectral interferences from higher order lines (PHA settings are of no help at all for 1st order spectral interferences!).   But too small a window could cause non-linear response from the detector, and give inaccurate results as the PHA peak moves around due to intensity shifting. So if there are spectral interferences present (whatever the Bragg order!) it is better to let the photons "all in", and deal with them using the quantitative spectral interference corrections in PFE.  You will obtain much better results this way.

Please note that the PHA settings have *nothing* to do with spectrometer spectral resolution. Spectral resolution is determined by the Bragg crystal and the sin theta of the spectrometer (and the detector slit width if those can be adjusted).

The PHA settings only affect the detector energy resolution, and as already stated, are only of use if high order spectral interferences are present and even then are only a partial solution (at best). Better to "let everything in" and use the interference corrections sort things out.

As already mentioned you'll still have to adjust the gain and bias to place the PHA peak properly on the x axis and this is done differently for JEOL and Cameca instruments. 

On the Cameca one adjusts the bias to obtain enough excitation in the detector to achieve a proper proportional response, usually 1300 volts for 1 atm detectors and 1850 volts for 2 atm detectors. Then the gain is adjusted to place the PHA peak at roughly 1 to 2 volts when observing a high intensity emission line.

On the JEOL instrument the gain settings are only in 2x steps, so the procedure is slightly different though the goal remains the same: to get the PHA peak between 3 and 5 volts with enough bias voltage to be in a stable proportional response regime. That usually means a bias voltages of 1580 to 1880 volts. I'll let someone with a JEOL instrument offer their preferred method for setting the gain and bias PHA values.
The only stupid question is the one not asked!

sem-geologist

#5
Rom,
As from given PHA voltage ranges I understand that your instrument is JEOL - Probman's recomendations applies in your case for qti as also WS methods.

I however completely disagree with recommendation of wide window on Cameca instruments. On Jeol looks You have better range - 10V, And it would be possible to fit least and largest energies without clipping from low and high side of the PHA window. Where in Cameca there is only 5V (or up to 5.5V in older electronics, which could actually also be up to 10V, but most significant bit of ADC is ignored, thus only half of the range), and It is impossible to set the bias and gain universally enough so nothing would  be clipped with widest PHA window possible either from low V or high V side. I just repeat again, with integral mode You get only hardware clipping (fixed value to filter out 0V noise), and chance to get into trouble is only if bias is too low and low side of PHA peak gets hardware filtered-out. as example lets consider TAP: with gain and bias for F to be not clipped by window baseline the Si PHA peak will be clipped from high energy side inevitably in diff mode. With integral mode anything what You see in the PHA scan is passed + anything You dont see at higher than scanned voltages.
Additionally to that, If You have new generation Cameca probe, in diff mode it will not allow You to set the window baseline so low as 0.2V, it will be following gain, and You will be allowed in best case to set it to something like 0.45V, and with high gain it will move to something like 1V. There is no way to get no clipping of peaks at WS in diff mode on new generation Cameca instruments.

So for full-range WS on Cameca should be used only integral mode with high enough bias so the lowest energies diffracted by set XTAL would be not clipped from lowest energy in PHA. Also for WS you want to see all possible spectral artefacts and features, in windowed mode (PHA diff) You can be falsely led to be unaware of some higher order interference, which would not go over noise in WS, but at quantitative step for trace elements it would bite you back.

Probeman

#6
Quote from: sem-geologist on February 27, 2022, 01:20:04 AM
and It is impossible to set the bias and gain universally enough so nothing would  be clipped with widest PHA window possible either from low V or high V side.

If by "set the bias and gain universally" you mean use the same gain and bias for all elements, I never suggested that, but I  now see that Rom is asking about a WS which I am guessing means a full range spectrometer wavescan.  That is a different animal of course!

If a full range spectrometer scan is what Rom is trying to do then I basically agree with your suggestions because my comments apply only to quantitative measurement of multiple elements on a single spectrometer.

By the way, what makes you think that the 0 to 5 volts range of the Cameca instruments does not correspond to the 0 to 10 volt range of the JEOL?  When I look at Cameca and JEOL PHA outputs, they look very similar unless I look at the scale of the X-axis.

Quote from: sem-geologist on February 27, 2022, 01:20:04 AM
Additionally to that, If You have new generation Cameca probe, in diff mode it will not allow You to set the window baseline so low as 0.2V, it will be following gain, and You will be allowed in best case to set it to something like 0.45V, and with high gain it will move to something like 1V. There is no way to get no clipping of peaks at WS in diff mode on new generation Cameca instruments.

I never said it should be set to 0.2 volts!  I said ~0.3 to 0.4 volts for a Cameca. That works just fine on my SX100.
The only stupid question is the one not asked!

Probeman

#7
Quote from: Rom on February 25, 2022, 08:51:10 PM
Greetings,
Why PHA settings for the same channel/crystal depends of measured substance?
Optimum PHA will less or more change for the same channel/crystal if we use during PHA calibration different substances.

Could you give recommendations about right steps of PHA calibration for WS collecting when spectrometer works in a wide range.
Thank you.

Hi Rom,
And all this time I thought by "WS" you meant "wavelength spectrometer"! 

Well at least now you have a good summary from me of dealing with quantification of elements with different emission energies on the same Bragg crystal!

That is why my first response was to point you to Ben Buse's topic on this.  You should also read Paul Carpenter's Powerpoint presentation on PHA calibration which Ben linked to. It is very much worth the time.
The only stupid question is the one not asked!

sem-geologist

#8
Probeman,
I am glad we came to some agreement (as for wavescan).
Yes older SX100 allows to set the baseline quite low, but new electronics (new gen WDS board with FPGA) ties that to gain and is not allowing to set the baseline low, in many cases when dealing with lowest energies at given XTAL/spectrometer the only way to loose that adjusting baseline minimum is increase the bias, which is not influencing the firmware enforced limit of baseline. It is generally good idea to always increase the bias a bit more from what automatic procedure (Auto PHA) sets and decrease the gain. That is way to increase the resolution of PHA. (because OPAMP "gain" noise is larger than gas amplification noise). While PHA peak stays inside PHA window there is absolutely no influence to the count rate with such adjustment.

I don't know how it is on Jeol, but from circuit schematics of Cameca on old electronics it is clear that ADC sends to microprocessor only 7 bits! 8th, the most significant bit, is ignored. It is quite strange as signal is checked before ADC to fit into 0-10V range (from -15/+15V; and ADC is rated to 10V), but only a half of that is measurement goes further. That means that You need to squeeze the PHA into 7 bit resolution instead of 8. With same bias and gain You would get the additional 5V space at high V side, which would allow to use wider window. Are Jeol using all 8 bits or similarly using only 7? I have no idea, I have access only to Cameca probes, and looked under lid only of those. Maybe all those limitations are from µC on old WDS boards which had pin limitations, and thus only 7bit and mux'ed ADC are placed there..? I also could be wrong about MSB, maybe it is LSB ignored, I should check the schematics again.

On new electronics (and firmware) it is a different design, and raw signal is downsized to 5V (from max 15V), and FPGA reads in all 8 bits from 5V ADC (at least traces of 8 bits are connected so, 8th bit could be ignored inside FPGA, we will never know). There is also shady part in that design there it is using a simple voltage divider (which AFAIK and had experience with other electronics) for signal scaling down before feeding it to ADC. Voltage dividers are quite susceptible to temperature - that could additionally shift the PHA peak i .e. due to clogged/cleaned up filter of VME cabin. Also voltage dividers have some drawbacks with impedance and changing frequencies, and that again could influence peak shift of PHA.

Probeman

#9
Quote from: sem-geologist on February 27, 2022, 10:30:46 AM
I don't know how it is on Jeol, but from circuit schematics of Cameca on old electronics it is clear that ADC sends to microprocessor only 7 bits! 8th, the most significant bit, is ignored. It is quite strange as signal is checked before ADC to fit into 0-10V range (from -15/+15V; and ADC is rated to 10V), but only a half of that is measurement goes further. That means that You need to squeeze the PHA into 7 bit resolution instead of 8. With same bias and gain You would get the additional 5V space at high V side, which would allow to use wider window. Are Jeol using all 8 bits or similarly using only 7? I have no idea, I have access only to Cameca probes, and looked under lid only of those. Maybe all those limitations are from µC on old WDS boards which had pin limitations, and thus only 7bit and mux'ed ADC are placed there..? I also could be wrong about MSB, maybe it is LSB ignored, I should check the schematics again.

I'm guessing that they squeeze the PHA signal into 0 to 5 volts and 8 bits because they are only using an 8 bit MCA (multi channel analyzer) to perform the PHA integration? That is, I observe that the returned PHA intensity data is always 0 to 255 integers.

I really like this MCA PHA feature on the Cameca as it is much faster than the SCA (single channel analyzer) serial method used by JEOL. I just wish they used a better MCA, say 16 bit...

Here are some of the raw PHA data returned by the SX100:

SX100GetPHADistributionMCA2: Index= 256
SX100GetPHADistributionMCA2 (Byte): Min= 0, Max= 254, CountTime= 1.836
SX100GetPHADistributionMCA2: 0, 0
SX100GetPHADistributionMCA2: 1, 0
SX100GetPHADistributionMCA2: 2, 0
SX100GetPHADistributionMCA2: 3, 0
SX100GetPHADistributionMCA2: 4, 0
SX100GetPHADistributionMCA2: 5, 0
SX100GetPHADistributionMCA2: 6, 0
SX100GetPHADistributionMCA2: 7, 0
SX100GetPHADistributionMCA2: 8, 0
SX100GetPHADistributionMCA2: 9, 1
SX100GetPHADistributionMCA2: 10, 2
SX100GetPHADistributionMCA2: 11, 2
SX100GetPHADistributionMCA2: 12, 2
SX100GetPHADistributionMCA2: 13, 3
SX100GetPHADistributionMCA2: 14, 3
SX100GetPHADistributionMCA2: 15, 4
SX100GetPHADistributionMCA2: 16, 4
SX100GetPHADistributionMCA2: 17, 5
SX100GetPHADistributionMCA2: 18, 6
SX100GetPHADistributionMCA2: 19, 8
SX100GetPHADistributionMCA2: 20, 9
SX100GetPHADistributionMCA2: 21, 10
SX100GetPHADistributionMCA2: 22, 12
SX100GetPHADistributionMCA2: 23, 13
SX100GetPHADistributionMCA2: 24, 13
SX100GetPHADistributionMCA2: 25, 15
SX100GetPHADistributionMCA2: 26, 18
SX100GetPHADistributionMCA2: 27, 20
SX100GetPHADistributionMCA2: 28, 22
SX100GetPHADistributionMCA2: 29, 25
SX100GetPHADistributionMCA2: 30, 27
SX100GetPHADistributionMCA2: 31, 30
SX100GetPHADistributionMCA2: 32, 30
SX100GetPHADistributionMCA2: 33, 33
SX100GetPHADistributionMCA2: 34, 36
SX100GetPHADistributionMCA2: 35, 40
SX100GetPHADistributionMCA2: 36, 43
SX100GetPHADistributionMCA2: 37, 47
SX100GetPHADistributionMCA2: 38, 50
SX100GetPHADistributionMCA2: 39, 53
SX100GetPHADistributionMCA2: 40, 53
SX100GetPHADistributionMCA2: 41, 57
SX100GetPHADistributionMCA2: 42, 61
SX100GetPHADistributionMCA2: 43, 64
SX100GetPHADistributionMCA2: 44, 68
SX100GetPHADistributionMCA2: 45, 72
SX100GetPHADistributionMCA2: 46, 75
SX100GetPHADistributionMCA2: 47, 79
SX100GetPHADistributionMCA2: 48, 79
SX100GetPHADistributionMCA2: 49, 83
SX100GetPHADistributionMCA2: 50, 86
SX100GetPHADistributionMCA2: 51, 90
SX100GetPHADistributionMCA2: 52, 94
SX100GetPHADistributionMCA2: 53, 97
SX100GetPHADistributionMCA2: 54, 101
SX100GetPHADistributionMCA2: 55, 105
SX100GetPHADistributionMCA2: 56, 105
SX100GetPHADistributionMCA2: 57, 109
SX100GetPHADistributionMCA2: 58, 113
SX100GetPHADistributionMCA2: 59, 117
SX100GetPHADistributionMCA2: 60, 121
SX100GetPHADistributionMCA2: 61, 125
SX100GetPHADistributionMCA2: 62, 130
SX100GetPHADistributionMCA2: 63, 134
SX100GetPHADistributionMCA2: 64, 134
SX100GetPHADistributionMCA2: 65, 139
SX100GetPHADistributionMCA2: 66, 144
SX100GetPHADistributionMCA2: 67, 150
SX100GetPHADistributionMCA2: 68, 155
SX100GetPHADistributionMCA2: 69, 160
SX100GetPHADistributionMCA2: 70, 166
SX100GetPHADistributionMCA2: 71, 171
SX100GetPHADistributionMCA2: 72, 171
SX100GetPHADistributionMCA2: 73, 176
SX100GetPHADistributionMCA2: 74, 182
SX100GetPHADistributionMCA2: 75, 188
SX100GetPHADistributionMCA2: 76, 194
SX100GetPHADistributionMCA2: 77, 199
SX100GetPHADistributionMCA2: 78, 205
SX100GetPHADistributionMCA2: 79, 211
SX100GetPHADistributionMCA2: 80, 211
SX100GetPHADistributionMCA2: 81, 215
SX100GetPHADistributionMCA2: 82, 220
SX100GetPHADistributionMCA2: 83, 224
SX100GetPHADistributionMCA2: 84, 229
SX100GetPHADistributionMCA2: 85, 233
SX100GetPHADistributionMCA2: 86, 238
SX100GetPHADistributionMCA2: 87, 242
SX100GetPHADistributionMCA2: 88, 242
SX100GetPHADistributionMCA2: 89, 244
SX100GetPHADistributionMCA2: 90, 246
SX100GetPHADistributionMCA2: 91, 247
SX100GetPHADistributionMCA2: 92, 249
SX100GetPHADistributionMCA2: 93, 251
SX100GetPHADistributionMCA2: 94, 252
SX100GetPHADistributionMCA2: 95, 254
SX100GetPHADistributionMCA2: 96, 254
SX100GetPHADistributionMCA2: 97, 252
SX100GetPHADistributionMCA2: 98, 250
SX100GetPHADistributionMCA2: 99, 247
SX100GetPHADistributionMCA2: 100, 245
SX100GetPHADistributionMCA2: 101, 243
SX100GetPHADistributionMCA2: 102, 241
SX100GetPHADistributionMCA2: 103, 238
SX100GetPHADistributionMCA2: 104, 238
SX100GetPHADistributionMCA2: 105, 232
SX100GetPHADistributionMCA2: 106, 226
SX100GetPHADistributionMCA2: 107, 221
SX100GetPHADistributionMCA2: 108, 215
SX100GetPHADistributionMCA2: 109, 209
SX100GetPHADistributionMCA2: 110, 203
SX100GetPHADistributionMCA2: 111, 197
SX100GetPHADistributionMCA2: 112, 197
SX100GetPHADistributionMCA2: 113, 189
SX100GetPHADistributionMCA2: 114, 181
SX100GetPHADistributionMCA2: 115, 173
SX100GetPHADistributionMCA2: 116, 165
SX100GetPHADistributionMCA2: 117, 157
SX100GetPHADistributionMCA2: 118, 149
SX100GetPHADistributionMCA2: 119, 141
SX100GetPHADistributionMCA2: 120, 141
SX100GetPHADistributionMCA2: 121, 133
SX100GetPHADistributionMCA2: 122, 125
SX100GetPHADistributionMCA2: 123, 117
SX100GetPHADistributionMCA2: 124, 109
SX100GetPHADistributionMCA2: 125, 101
SX100GetPHADistributionMCA2: 126, 94
SX100GetPHADistributionMCA2: 127, 86
SX100GetPHADistributionMCA2: 128, 86
SX100GetPHADistributionMCA2: 129, 80
SX100GetPHADistributionMCA2: 130, 74
SX100GetPHADistributionMCA2: 131, 68
SX100GetPHADistributionMCA2: 132, 62
SX100GetPHADistributionMCA2: 133, 57
SX100GetPHADistributionMCA2: 134, 51
SX100GetPHADistributionMCA2: 135, 45
SX100GetPHADistributionMCA2: 136, 45
SX100GetPHADistributionMCA2: 137, 41
SX100GetPHADistributionMCA2: 138, 38
SX100GetPHADistributionMCA2: 139, 34
SX100GetPHADistributionMCA2: 140, 31
SX100GetPHADistributionMCA2: 141, 27
SX100GetPHADistributionMCA2: 142, 24
SX100GetPHADistributionMCA2: 143, 20
SX100GetPHADistributionMCA2: 144, 20
SX100GetPHADistributionMCA2: 145, 19
SX100GetPHADistributionMCA2: 146, 17
SX100GetPHADistributionMCA2: 147, 15
SX100GetPHADistributionMCA2: 148, 13
SX100GetPHADistributionMCA2: 149, 12
SX100GetPHADistributionMCA2: 150, 10
SX100GetPHADistributionMCA2: 151, 8
SX100GetPHADistributionMCA2: 152, 8
SX100GetPHADistributionMCA2: 153, 8
SX100GetPHADistributionMCA2: 154, 7
SX100GetPHADistributionMCA2: 155, 6
SX100GetPHADistributionMCA2: 156, 6
SX100GetPHADistributionMCA2: 157, 5
SX100GetPHADistributionMCA2: 158, 4
SX100GetPHADistributionMCA2: 159, 4
SX100GetPHADistributionMCA2: 160, 4
SX100GetPHADistributionMCA2: 161, 3
SX100GetPHADistributionMCA2: 162, 3
SX100GetPHADistributionMCA2: 163, 3
SX100GetPHADistributionMCA2: 164, 3
SX100GetPHADistributionMCA2: 165, 3
SX100GetPHADistributionMCA2: 166, 2
SX100GetPHADistributionMCA2: 167, 2
SX100GetPHADistributionMCA2: 168, 2
SX100GetPHADistributionMCA2: 169, 2
SX100GetPHADistributionMCA2: 170, 2
SX100GetPHADistributionMCA2: 171, 2
SX100GetPHADistributionMCA2: 172, 2
SX100GetPHADistributionMCA2: 173, 2
SX100GetPHADistributionMCA2: 174, 2
SX100GetPHADistributionMCA2: 175, 2
SX100GetPHADistributionMCA2: 176, 2
SX100GetPHADistributionMCA2: 177, 2
SX100GetPHADistributionMCA2: 178, 2
SX100GetPHADistributionMCA2: 179, 2
SX100GetPHADistributionMCA2: 180, 2
SX100GetPHADistributionMCA2: 181, 2
SX100GetPHADistributionMCA2: 182, 2
SX100GetPHADistributionMCA2: 183, 2
SX100GetPHADistributionMCA2: 184, 2
SX100GetPHADistributionMCA2: 185, 2
SX100GetPHADistributionMCA2: 186, 2
SX100GetPHADistributionMCA2: 187, 2
SX100GetPHADistributionMCA2: 188, 2
SX100GetPHADistributionMCA2: 189, 2
SX100GetPHADistributionMCA2: 190, 2
SX100GetPHADistributionMCA2: 191, 2
SX100GetPHADistributionMCA2: 192, 2
SX100GetPHADistributionMCA2: 193, 2
SX100GetPHADistributionMCA2: 194, 2
SX100GetPHADistributionMCA2: 195, 1
SX100GetPHADistributionMCA2: 196, 1
SX100GetPHADistributionMCA2: 197, 1
SX100GetPHADistributionMCA2: 198, 1
SX100GetPHADistributionMCA2: 199, 1
SX100GetPHADistributionMCA2: 200, 1
SX100GetPHADistributionMCA2: 201, 1
SX100GetPHADistributionMCA2: 202, 1
SX100GetPHADistributionMCA2: 203, 1
SX100GetPHADistributionMCA2: 204, 1
SX100GetPHADistributionMCA2: 205, 1
SX100GetPHADistributionMCA2: 206, 1
SX100GetPHADistributionMCA2: 207, 1
SX100GetPHADistributionMCA2: 208, 1
SX100GetPHADistributionMCA2: 209, 1
SX100GetPHADistributionMCA2: 210, 1
SX100GetPHADistributionMCA2: 211, 1
SX100GetPHADistributionMCA2: 212, 1
SX100GetPHADistributionMCA2: 213, 1
SX100GetPHADistributionMCA2: 214, 1
SX100GetPHADistributionMCA2: 215, 1
SX100GetPHADistributionMCA2: 216, 1
SX100GetPHADistributionMCA2: 217, 1
SX100GetPHADistributionMCA2: 218, 1
SX100GetPHADistributionMCA2: 219, 1
SX100GetPHADistributionMCA2: 220, 1
SX100GetPHADistributionMCA2: 221, 1
SX100GetPHADistributionMCA2: 222, 1
SX100GetPHADistributionMCA2: 223, 0
SX100GetPHADistributionMCA2: 224, 0
SX100GetPHADistributionMCA2: 225, 0
SX100GetPHADistributionMCA2: 226, 0
SX100GetPHADistributionMCA2: 227, 0
SX100GetPHADistributionMCA2: 228, 0
SX100GetPHADistributionMCA2: 229, 0
SX100GetPHADistributionMCA2: 230, 0
SX100GetPHADistributionMCA2: 231, 0
SX100GetPHADistributionMCA2: 232, 0
SX100GetPHADistributionMCA2: 233, 0
SX100GetPHADistributionMCA2: 234, 0
SX100GetPHADistributionMCA2: 235, 0
SX100GetPHADistributionMCA2: 236, 0
SX100GetPHADistributionMCA2: 237, 0
SX100GetPHADistributionMCA2: 238, 0
SX100GetPHADistributionMCA2: 239, 0
SX100GetPHADistributionMCA2: 240, 0
SX100GetPHADistributionMCA2: 241, 0
SX100GetPHADistributionMCA2: 242, 0
SX100GetPHADistributionMCA2: 243, 0
SX100GetPHADistributionMCA2: 244, 0
SX100GetPHADistributionMCA2: 245, 0
SX100GetPHADistributionMCA2: 246, 0
SX100GetPHADistributionMCA2: 247, 0
SX100GetPHADistributionMCA2: 248, 0
SX100GetPHADistributionMCA2: 249, 0
SX100GetPHADistributionMCA2: 250, 0
SX100GetPHADistributionMCA2: 251, 0
SX100GetPHADistributionMCA2: 252, 0
SX100GetPHADistributionMCA2: 253, 0
SX100GetPHADistributionMCA2: 254, 0
SX100GetPHADistributionMCA2: 255, 0

Basically when one starts a PHA MCA acquisition on the Cameca it acquires photons until one of the MCA 256 "bins" hits 255 photons counted. It then returns the array of PHA photon count data plus the acquisition time so it can be normalized to cps or whatever.  If it doesn't reach at least 255 photons in any bin within 20 sec it returns with an error.
The only stupid question is the one not asked!

sem-geologist

Quote from: Probeman on February 27, 2022, 11:24:58 AM

I'm guessing that they squeeze the PHA signal into 0 to 5 volts and 8 bits because they are only using an 8 bit MCA (multi channel analyzer) to perform the PHA integration? That is the returned PHA intensity data is always 0 to 255 integers.

I really like this MCA PHA feature on the Cameca as it is much faster than the SCA (single channel analyzer) serial method used by JEOL. I just wish they used a better MCA, say 16 bit...

MCA is just a buzz word. Seriously. It some kind of bookish abstraction, as raw pulses are still analyzed sequentially (serial). (Fun fact: Even EDS is serial/sequentially analysed, it just happens so fast that it looks it updates many channels at once). The question would be why that SCA on Jeol works slower (maybe it is more precise? probably it lets the signal in the pipeline to reset to zero before processing next pulse, where on Cameca it is rushed). What is max throughtput with SCA on JEOL? Cameca WDS throughtput ceiling is about 300kcps (at PHA integral mode with 1µs dead time set). So there actually is nothing like 8 bit MCA on Cameca, as MCA is abstraction of bunch of components designed to work like one, and not some off-shelf ready-to-use component (albeit You can buy such devices, but Cameca "MCA" is custom designed and programmed). The main parts of MCA on Cameca is pulse hold chip (as amplified pulse needs to wait in line for multi spectrometer shared ADC (analog-digital conventer)  to be freed), then muxer (many-into-one), then ADC, and then micro-controller/processor (older use two philips 680xx, new use single FPGA which includes all MCA logic and VME communication protocoles in single package). So "MCA" is just a bunch of components coupled with custom written logic. Now when it comes to Phillips 68070 - it is 16bit/32bit microprocessor. The only bottleneck is 8bit ADC. FPGA are actually bit-agnostic (as logic is customely written, the implementation decides about bus width it can be even bizarre bit number like 22bit). I guess that ADC it is limited to 7 from 8bit in old design as (1) there were not enough input pins to Phillips 68070, probably that is the same reason why a single ADC serves 3 spectrometers. Three independently connected 8bit ADC would need 8*3 pins + clock/read_enable * 3 which would be 27 i/o pins (+ 3 for pulse hold), where implemented muxed design needs only 7  pin + 3 spectrometer selection pins (in sum 10 pins + 3 pulse hold). There are 2 68070 on old board. If took a close look to the board it is obvious that it is already fully crowded and there is no place for 3rd 68070. Also there comes ADC delay, generally the more wider the bus the more clock cycles needs to pass for outcome to be output after input. It normally cause no trouble in sequentially digitized single pulse source, but because this is muxed (full word for that is "multiplexed" btw), it would over-complicate design tremendously. And so limitation to 8bit ADC in old design has its reasons.

But when I look to new design (with FPGA) I see completely no possible limitations which would still force using 8bit ADC and muxing the pulses to single ADC. The board was downsized, that large free space is effect of merging MCU and VME logic into a single FPGA. There is no space constraints. I guess that used FPGA probably has enough pins (and even if it is not, just use a bit more beefy FPGA, some commercially available FPGAs expose thousands of i/o pins), I can't understand the reason that muxing of pulse signals is still here, probably it is just saving a some $, can't explain it otherwise. The 12bit ADC got pretty fast and cheap in last decades, there are plenty good candidates. (most of EDS MCA use 12bit, as that is still very fast, and gives 4096 channels which is plenty enought).
But before moving to even up to 12bits, the amplification part needs to be revised, as it clearly adds lots of statistic noise (broadens tremendously the PHA peak) and that looks just ctrl-c, ctrl-v from previous design.

I already wrote that some time ago, but will repeat myself: How is that possible that on oscilloscope (cheapo, slow Chinese no name brand worth 250$) hanged to signal coming from pre-amplifier circuit clearly exposes amplitudes of single and double pile ups with piled up Ar escape pulses, but at PHA graph it is just single muddled peak? The only explanation I can come up is that gain amplification introduce lots of statistic noise and pulse hold chip is not registering maximum amplitude. Signal is worsened a lot somewhere between charge pre-amplifier and ADC.

So In my opinion it would be best to move away from centralized processing, and distribute small FPGAs per spectrometer with 12 bit ADC directly digitising the signal outputted from charge pre-amplifier circuit and feeding to small FPGA. No more muxing, no peak holding, no cascade OPAMP amplification, impedence mismatching due to different frequency (pulse density) response, sending of analog signals for kilometers (exagerated, but even those few meters has its impact)... I even am not sure that this would be expensive.

John Donovan

#11
Quote from: sem-geologist on February 27, 2022, 02:12:14 PM
MCA is just a buzz word. Seriously. It some kind of bookish abstraction

Whatever. It's just a word we use to distinguish between the two modes of software operation for the benefit of the user. Mostly because on the Cameca instrument Probe for EPMA can utilize either the Cameca (MCA) method or the traditional JEOL (SCA) method where the baseline is stepped serially through the voltage range.

Indeed the SCA method does produce a sharper peak as you mention, but I find the MCA acquisition so much faster, it's well worth the loss of energy resolution. But the user gets to choose which PHA acquisition method to use.

Thanks for the very useful information on the internal electronics.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Probeman

#12
All this discussion on PHA settings caused me to go to the lab and do some quick measurements in an effort to tease out this relationship between energy and count rate in terms of PHA peak energy shifting.

That is, how much gain shift do we see from the incident x-ray emission line energy change vs. change in count rate?  We know the PHA peak shifts to a higher energy as the concentration (count rate) drops, because the detector is under less "load" at low count rates, and hence the bias voltage can maintain a more consistent high voltage.  But how much is PHA shifting due to the difference in the emission line energy?

So starting with a TAP crystal I selected two emission lines that were at the extreme ends on the spectrometer range, F Ka and P Ka. That's an energy difference of 0.667 keV to 2.013 keV. First on the low energy side I peaked up F Ka on BaF2, noted the x-ray intensity on the ratemeter,  and collected a PHA scan:



Note that for all PHA scans on this TAP crystal, the baseline, window, gain and bias settings were exactly the same. Next, moving to the higher energy (lower sin theta) P Ka position, I again peaked the spectrometer and collected a PHA scan on the apatite standard:



As you can see the PHA peak has shifted to the right, indicating an energy gain. But, because the P Ka emission line is much stronger than the F Ka line due to the increased geometric efficiency at low sin theta, we would expect that the PHA peak shift to the left because of the high P Ka intensity (the fluorine/phosphorus concentrations are roughly the same in each standard).

So to separate these two effects, I de-tuned the P Ka peak position until the ratemeter read the same as for the F Ka peak and performed another PHA scan:



Ok, so that shifted the PHA peak even a bit further to the right (higher energy) as expected (the Y-axis scales on the F Ka scan and the rate adjusted P Ka scan are not the same because the PHA scans reflect the integrated area of the incident x-rays).

So we can see that the PHA peak is significantly shifted to a higher energy (as expected) by an increase in emission line energy, when the x-ray count rate has been adjusted to be the same.  At least for TAP. Next let's look at the PET crystal...
The only stupid question is the one not asked!

Probeman

Now we look at a similar difference in energy for the PET crystal, but using the Si Ka line (for the low energy) vs. the Mn Ka line (for the high energy line).

First peaking up on Si ka on PET (at high sin theta) on Si metal:



Again we note the x-ray ratemeter reading on Si Ka and move to Mn metal and peak Mn Ka and perform a PHA scan:



As was the case for the TAP crystal we see an energy shift to higher energies. But how much is lost due to the increase in the Mn Ka count rate at low sin theta?  After de-tuning the Mn Ka peak to obtain the same count rate as we saw with the Si Ka peak, we obtain this PHA scan:



Again (as expected) the gain shift is even higher simply due to the change in the emission line energy.  I didn't have time to check the LiF crystal, but assume we would observe similar results there.

I would also expect the same result from JEOL detectors and PHA electronics, but would anyone be willing do run a quick experiment for us to confirm?
The only stupid question is the one not asked!

Probeman

So in summary, we can conclude that when keeping the baseline/window wide open and utilizing constant gain/bias settings, one should be able to acquire full spectrometer range wavescans.

That is, so long as a low energy (high sin theta) emission line (at the end of the spectrometer range) is initially tuned to place the PHA peak near the low energy side of the PHA voltage range (1 to 2 volts for Cameca or 3 to 4 volts for JEOL).

Furthermore, while the photons on the high energy (low sin theta) end of the spectrometer will cause the PHA peak to shift towards the right of the x-axis, it should still be within the PHA scan range (so long as the baseline and window are wide open).

Finally, the increase in count intensity on the high energy (low sin theta) end of the spectrometer range (due to increasing geometric efficiency) will actually help to keep the PHA peaks from shifting as much towards the right.
The only stupid question is the one not asked!