How to improve WDS pulse processing

[Probeman]

You miss the point, it is very similar problem as the EDS, also it is similarly divided between hardware and software.

No, I do not miss the point at all.  I've already agreed with you on this issue as I stated here previously:

OK, with that out of the way, let's proceed with the promised discussion regarding SG's comments on the factors contributing towards dead time effects in WDS spectrometers, because there is no doubt that several factors are involved in these dead time effects, both in the detector itself and the electronics.

I've actually mentioned several times previously that there are several separate effects are involved in the WDS dead time correction, so please stop with the "straw man" arguments. 

Here's another example that just popped into my head: have we thought about satellite line production at high beam currents that could rob us of counts on the main emission line due to double and triple atomic ionizations in our sample?  This effect (if significant) is of course only a problem in WDS because of its high energy resolution, and can *only* be corrected in software because the photons don't even ever reach the WDS detector!   

But the more important point is that from what we see in the constant k-ratio data, we can make software corrections for photon coincidence quite easily as you yourself have already acknowledged:

However, I think probeman's et al model is too little physically realistic as it accounts for pulse pile up's too weakly (not to strongly as Brian's argumentation suggests) and that gets obvious at higher currents/higher count rates (I don't see that high count rates as anomaly, but as one from pivotal points in testing the correctness of the model). But classical "linear" model does not do that at all, so in that sense this new log function is much better as it do it at least partially, and while it is still not perfect, it is a movement in the right direction.

You're welcome!   

I can live with "not perfect"!       So, with this new log expression we have enabled WDS quantitative analysis at count rates up to 10 times greater than what was previously possible. But above that we run into the more electronics dependent limitations as I stated here previously:

But however we measure these dead time effects by counting photons, they are all combined in our measurements, so the difficulty is in separating out these effects.  But the good news is that these various effects may not all occur in the same count rate regimes.

For example, we now know from Monte Carlo modeling that at even relatively low count rates, that multiple photon coincidence events are already starting to occur. As seen in the above plot starting around 30 to 40 nA (>50K to 100K cps), on some large area Bragg crystals.

As the data reveals, the traditional dead time expression does not properly deal with these events, so that is the rationale for the multiple term expressions and finally the new logarithmic expression. So by using this new log expression we are able to achieve normal quantitative accuracy up to count rates of 300K to 400K cps (up to 140 nA in the first plot). That's approximately 10 times the count rates that we would normally limit ourselves to for quantitative work!

So, yes, this log expression produces *only* a 10 fold improvement in quantitative analysis at high beam currents!   

But if you ever do come up with an paramatizable expression that can correct for these WDS electronics limitations at count rates over 400K cps (for JEOL) or 300K cps (for Cameca) please let us know and maybe we can implement them as a separate correction, in addition to the normal dead time (photon coincidence) correction.  You can name it whatever you want!   

The problem I see is that there may be significant differences is how these electronic issues should be treated  on JEOL vs. Cameca hardware.  But maybe there is enough similarity, though the Cameca "enforced" dead time would seem to be one difference between them.

JEOL here has the advantage because JEOL instruments have intrinsic dead times about half of Cameca's, so clearly this is more important for Cameca instruments, but certainly could help on all instruments.

The SDDs are fast because they have a much better signal P/B ratio, which made it the much smaller shaping times (like 25ns and even less  ) practical, while manufactorers of WDS had not updated its counting design practically from 80es (there is only replacement of some counting component chips with equivalent modern counter parts). That is how EDS can cope with 8Mcps (at least on the paper: newest AMPTEK SDD systems) input count rates with "extendable" deadtime and practically with no PHA (energy) shifting, while our WDS counting can't cope with not-extendable deadtime and experience severe PHA shifting already at 150kcps. Software can do some corrections, but it is a bit uphill battle.

I could not agree more with you on this point. It's simply criminal that WDS is still relying on 1980s electronic technology! I'm just doing what can be done in software because that is the only area I have any influence on. 

But again, if you come up with a software correction for these WDS electronics limitations I would be very happy to test them out.

[sem-geologist]

JEOL here has the advantage because JEOL instruments have intrinsic dead times about half of Cameca's, so clearly this is more important for Cameca instruments, but certainly could help on all instruments.

I already wrote in the other thread, but... stop making Cameca's dead times look bad compared to Jeol! JEOL "knows" better than the user and so Jeol probes has  fixed dead time blanking (You call it "intrinsic", I need to look up thesaurus every time to understand it  , and I still don't get its meaning in this context), where Cameca has it by default 3µs, but on Cameca SX it is user changeable. If You say A say B; Yes Jeol has shorter fixed dead time, which introduce severe PHA broadening and shift at higher counting rates. Cameca SX has default (not fixed) dead time of 3µs to postpone the PHA shifts and broadening to higher count rates. However, you can set Cameca SX dead time on 1µs and get even lower dead time than Jeols, and blow Jeol's probes throughput-socks-off with that (in integral mode only as You will get even funnier PHA shifts than those on Jeol at such silly set dead time). Your "dead time constant" calibration at such settings will probably open portal to the other dimension and produce the negative dead time , but joking aside, I would not be surprised if You could get "dead time constant" < 1.0µs seeing the nature of how this log equation works. Anyway Cameca gives much more user freedom than Jeol (unfortunately they never pushed that at marketing - that could had been one of strong selling points). Even with FEG, we have completely open system on Cameca (protected by some passwords) and I can abuse the emitter to my hearts content, where on Jeol I could not do any of that as settings would be hidden and restricted from user (reserved only to service).

The problem I see is that there may be significant differences is how these electronic issues should be treated  on JEOL vs. Cameca hardware.  But maybe there is enough similarity, though the Cameca "enforced" dead time would seem to be one difference between them.

After last seeing PHA taken at different count rates on JEOL (other thread) and the rate how it shifts I doubt there is significant difference how it is working between those two vendors. Most likely Jeol probes has same shaping time (could have different vendor for shapping amplifier but most probably it is also 250ns, that is more decided by P10 gas probbaly) and hardware physical dead times fixed to 2µs (I guess). Severe PHA shifting reveals the same counting architecture (comparator - sample/hold chip tandem) and lower dead time. Would there be less PHA shifting, then only we could argue that it uses rather shorter shaping amplifier, but as it is so much shifting the guess about shaping time of 250ns rather should be correct.

But again, if you come up with a software correction for these WDS electronics limitations I would be very happy to test them out.

Why only software, I am ambitious to come up with hardware solution. Actually the more I look into it, the more hardware solution looks for me more feasible. Because with software we are able to correct to some degree the count rates, but not PHA shifts and broadening. With hardware mod it would be possible to address and push forward both limitations. Also I am suspecting more and more that half of normal PHA broadening is caused by comparator-sample/hold chip tandem which does not hold the peak, but near-peak values of the pulse. Again I find Cameca system more open for such modifications and I have looked through tap and injection points for this to work.

For example, we now know from Monte Carlo modeling that at even relatively low count rates, that multiple photon coincidence events are already starting to occur. As seen in the above plot starting around 30 to 40 nA (>50K to 100K cps), on some large area Bragg crystals.

Actually the first time I came across the pulse-pile ups was on oscilloscope, it is not only modeled with MC simulation(which followed the observations), but can be observed physically with even cheap oscilloscope. Furthermore - it is not only large crystals causing the trouble, actually the first time I saw it with oscilloscope it was on signal of spectrometer with normal sized TAP. 

[Probeman]

JEOL here has the advantage because JEOL instruments have intrinsic dead times about half of Cameca's, so clearly this is more important for Cameca instruments, but certainly could help on all instruments.

I already wrote in the other thread, but... stop making Cameca's dead times look bad compared to Jeol! JEOL "knows" better than the user and so Jeol probes has  fixed dead time blanking (You call it "intrinsic", I need to look up thesaurus every time to understand it  , and I still don't get its meaning in this context), where Cameca has it by default 3µs, but on Cameca SX it is user changeable. If You say A say B; Yes Jeol has shorter fixed dead time, which introduce severe PHA broadening and shift at higher counting rates. Cameca SX has default (not fixed) dead time of 3µs to postpone the PHA shifts and broadening to higher count rates. However, you can set Cameca SX dead time on 1µs and get even lower dead time than Jeols, and blow Jeol's probes throughput-socks-off with that (in integral mode only as You will get even funnier PHA shifts than those on Jeol at such silly set dead time).

Don't be so sensitive!      I myself am mostly brand agnostic (both instruments have their advantages and disadvantages) though I confess I have bought two Cameca instruments in my career.

So you say that one can set the Cameca enforced dead time to 1 usec, but then you say it will produce "even funnier PHA shifts than those on Jeol at such silly set dead time".  So what is the lowest enforced dead time on a Cameca that is not "silly"?

Do we suppose one cannot run the Cameca at much lower than ~3 usec dead times?  In that case the JEOL ~1.5 usec dead times are roughly twice as fast as Cameca ~3 usec deadtimes, correct? Do you agree that shorter dead times are better than longer dead times?

But it's interesting because back in the day when I was using the Carpenter Excel spreadsheet for dead time calibrations, I would always first set the enforced (integer) dead times on my Cameca instrument to 1 usec, just to see what the "intrinsic" (sorry) dead times of the spectrometers were, without any "enforced" dead time (on the SX50/51 instruments one could set the integer dead times to 0 usec, but on the SX100 the lowest accepted value is 1 usec) .

And what I saw each time was that the *measured* (intrinsic) dead times were actually around 2.2 to 2.9 usec, as plotted using Carpenter's Excel spreadsheet (using a 1 usec integer/enforced dead time).  So I'm not sure why you think the Cameca dead times are so much faster than JEOL's...  maybe your instrument is different?

I would then set the integer (enforced) dead times to 3 usec (to cover the measured intrinsic dead times), and then I would re-run the dead time calibrations and that is where I would get measured dead times varying from 2.9 to 3.2 usec on the various spectrometers.

But again, if you come up with a software correction for these WDS electronics limitations I would be very happy to test them out.

Why only software, I am ambitious to come up with hardware solution. Actually the more I look into it, the more hardware solution looks for me more feasible. Because with software we are able to correct to some degree the count rates, but not PHA shifts and broadening. With hardware mod it would be possible to address and push forward both limitations. Also I am suspecting more and more that half of normal PHA broadening is caused by comparator-sample/hold chip tandem which does not hold the peak, but near-peak values of the pulse. Again I find Cameca system more open for such modifications and I have looked through tap and injection points for this to work.

Why only software?  Because software is what I do!   

But I applaud your ambition in designing new hardware, so go man, go!   Please keep us up to date in this new topic on your endeavors as I'm sure one could design much better WDS pulse processing electronics with modern technology.

If you did come up with some amazing new electronics please (please!) license it to JEOL and Cameca as both these instruments need better WDS electronics.