Geo,

Why is there ?4.8mm of Al attenuator in the beam path? ?What is the no collimated diameter of the micro-focus beam on the target? Is it well aligned with the 1mm collimator, if it’s not you’ll have ?a lot of attenuation.? Collimation usually uses 2 ?separated collimators and sometimes one on the detector. ?Start simple, no collimator, no attenuators. Look at the beam diameter at the target plane. Set up the collimation to get the beam diameter you want at the target plane. ?Then move the Detector close to the target say about 1 – 2cm. ?Take a shot using the lowest beam current and look at Dead Time. You should be able to count at 40Kcps and have good resolution and count time.? Play with the beam current. Play with Detector distance to target.

Your comment “…allows the sensor to be considerably close to the center of the chamber. Not that? it maters very much” ?What is considerably closer? It matters very much. The beam will expose a small circle on the target that will fluoresce over a solid angle. The flux in is proportional to the flux out. The percentage of that signal your detector will get is directly dependent on the distance from the target and that solid angle. Close up its 1/r, at a distance it falls as 1/r2.? Remember the time distance shielding thing

?Do a sensitivity analysis. What gets the best signal? What changes things the most? – where do I optimize my effort? Is it Distance, geometric angle between beam and detector, collimation, need for attenuator, what size beam spot do I want on the target, what s that get me in count time etc.

The problem here is not the difference between a SDD and a SI-PIN it’s the set up geometry and the shield system limiting you to very poor performance.

Dud

KK7IF

?

?

Geo,

Getting only 1800 counts/sec ?in 600 sec is saying something is not quite right with the setup. Not separating the Fe and Ni Kb and Ka is also not right but may? be related to the low count rate. Before you go any further find out why you’re getting the low count rate. This spectra should come in in ?30 sec count. ?I suspect the beam flux is too low (what’s the collimator diameter and where is the micro-focus point?) and if you haven’t moved the detector closer to the target ?that may be the problem as well.? Get rid of the lead shield setup and set up the geometry by hand without it (but keeping ?shielding between you and it). Play around getting an optimal geometry and beam current. Start by getting the detector closer to the target.? I’d think running ?a test with and without the plastic cap on the low energy end would be interesting test to see as well. Use a Si target.

Dud

?

Baby steps- a 304 stainless bolt from the shop.

Geo

I’m not sure what the Amptek look up library can handle. It seems it only is set up for a Ka Kb or La Lb can ?we mix K’s and L’s ?in there as well and have it report correctly?? Inquiring minds want to know.

Dud

?

Geo,

The Rb you’re looking at , sorry, the computer is looking at is U Ka at 13.6, ?Rb has a Ka of 13.4 they can get confused if the ROI is not selected correctly. However in this case the Amptek Ka library doesn’t have a listing for U Ka so it’s never reported.? We need to check and fix these libraries.

Here’s some sage advice; The computer is not your friend, it hates you and everything you stand for because you’re smarter than it is. Its crafty – It has a Critical Need Detector and it’ll wait for the best possible moment to get its revenge. To avoid this DO EVERTHING YOURSELF.

?

To interpret spectra? first look at it. Good well defined peaks or Noisy? Rerun it for a longer count

Then start at the low energy side and manually set ROI’s for each peak. For each peak find its peak energy and the look up in the x-ray table to determine what are the possible elements. Then look the for the Kb or Lb and confirm what you are looking at. Ka to Kb ratio is abt 5:1 La to ?Lb ?is 1:1 or so. If it doesn’t have a brother then it’s not a positive ID just unconfirmed.? Low Z elements won’t have any L peaks only K. Middle Z will have both and High Z will have L and possibly K depending on excitation energy. ?Some peaks are going to be good peaks others will be unknowns. The unknowns are artifacts and will include the sum, escape, ?backscatter, Compton, Rayleigh and merged peaks that are there just to harass and humble you.

In each case it’s up to you to interpret the data, the computer picks will 100% get you into trouble without verification .

Do not preconceive anything. For the Theromino and PXRF ?crowd - Don’t set any element lines on the screen. YOU set them as you ID an element. Don’t bias yourself. Believe nothing verify everything.? Just because you always had a Pb line set up doesn’t mean this time it’s not an As.? If you can’t figure it out without help lines you’re using a faulty interpretation model that will surely get you in trouble.? BTW ?the Garbage-In-Garbage- Out phrase was actually based on computer XFR interpretations ?not on the computer models it is usually attributed to.

XRF interpretation is not quick or easy but It’s very satisfying…when it works

Dud

?

?

Geo,

Attached is a manually selected ROI showing the Cr, Fe, Mo, and the missing Ni. ?Using auto generated ROI computer picks usually are not very helpful and in the end its probably best to set your own using a quasi-heuristic approach (by eye-ball) for the ROI’s.

Note that these peaks are noisy and under sampled with the largest peak, Fe, having only 212 counts in it for a 20 sec count.? The count rate is 2306 cps which is slow for this detector which can handle a much higher count rate. ?For your 4.8 usec peaking time you should be able get up to 50,000 cps.? To do that try a couple of things, ?get the detector closer to the target, increase the beam current and optimize the collimator’s beam diameter. ?I suspect? the large standoff distance of the detector may be the most sensitive to the problem.?

The Cal Check.bmp ?attachment is a 30 sec 50 kV shot of a 316 SS calibration disk resulting in a count rate of 31,861 cps.? Note the Fe peak is well defined and has a 1909 count peak.

Dud

KK7IF

?

?

Geo,

My system cal uses a 316 SS disk. Take the wire with you and see if they have a xrf metal analyzer and get an ID that way.

Dud

?