TDI Acquisition and Estimating Time (synchronous vs. asynchronous)

[pgopon]

When we do a TDI with 1 sec counting intervals, the analysis time goes from 2 minutes (60 sec peak, 30 each bkg) to over 8 minutes.  When you watch the counting output in the acquire window it is like it is acquiring one second of data on spectrometer 1 then writing it, before continuing onto the next spectrometer counting for another second writing that, etc etc. for each spectrometer in sequence, not simultaneously counting on each spectrometer.  Is this a glitch in the output of the acquire windows that it can't keep up, or is it actually doing each spectrometer individually, in sequence.  Based on the increased time the analysis takes it makes me think that it is not doing simultaneous acquisition, but then I didn't write the code.

[John Donovan]

When we do a TDI with 1 sec counting intervals, the analysis time goes from 2 minutes (60 sec peak, 30 each bkg) to over 8 minutes.  When you watch the counting output in the acquire window it is like it is acquiring one second of data on spectrometer 1 then writing it, before continuing onto the next spectrometer counting for another second writing that, etc etc. for each spectrometer in sequence, not simultaneously counting on each spectrometer.  Is this a glitch in the output of the acquire windows that it can't keep up, or is it actually doing each spectrometer individually, in sequence.  Based on the increased time the analysis takes it makes me think that it is not doing simultaneous acquisition, but then I didn't write the code.

Hi Phil,
That does not sound right at all!

You're using the "self" TDI option correct?  (as opposed to the "assigned" TDI option)  Where only the first element on each spectrometer is acquired using the TDI intervals?

The "assigned" TDI is only for special (statistical) circumstances.  See here:

http://probesoftware.com/smf/index.php?topic=71.msg3384#msg3384

[pgopon]

Confirmed using the self correction TDI, only one element per spectrometer.

phil

[John Donovan]

Confirmed using the self correction TDI, only one element per spectrometer.

phil

Hi Phil,
Well that's a relief, but you're going to have to provide more information because we have a student on the instrument today running TDI and their acquisition time without TDI is 261 seconds and with TDI is 271 seconds (they are using 4 TDI intervals)...

Here is their setup:

Last (Current) On and Off Peak Count Times:
ELEM:    mg ka   mg ka   mg ka   ca ka   na ka   sr la   ba la   mn ka   si ka   fe ka
BGD:       MAN     MAN     MAN     MAN     OFF     OFF     OFF     OFF     OFF     OFF
BGDS:      MAN     MAN     MAN     MAN     LIN     EXP     LIN     LIN     EXP     LIN
SPEC:        1       4       2       3       1       2       3       3       4       5
CRYST:     TAP     TAP    LTAP    LPET     TAP    LTAP    LPET    LPET     TAP     LIF
ORDER:       1       1       1       1       2       2       2       3       2       1
ONTIM:   20.00     .00     .00   20.00   30.00   20.00   20.00   20.00   20.00   20.00
HITIM:    ----    ----    ----    ----   15.00   10.00   10.00   10.00   10.00   10.00
LOTIM:    ----    ----    ----    ----   15.00   10.00   10.00   10.00   10.00   10.00
UNFAC:       1       1       1       1       3       4       2       2       4       4
ONTIME   20.00     .00     .00   20.00   90.00   80.00   40.00   40.00   80.00   80.00
HITIME    ----    ----    ----    ----   45.00   40.00   20.00   20.00   40.00   40.00
LOTIME    ----    ----    ----    ----   45.00   40.00   20.00   20.00   40.00   40.00
AGGR:                                                                                 

Miscellaneous Sample Acquisition/Calculation Parameters:
KILO:    15.00   15.00   15.00   15.00   15.00   15.00   15.00   15.00   15.00   15.00
ENERGY   1.254   1.254   1.254   3.691   1.041   1.807   4.466   5.895   1.740   6.400
EDGE:    1.305   1.305   1.305   4.039   1.073   1.940   5.247   6.539   1.839   7.112
Eo/Ec:   11.49   11.49   11.49    3.71   13.98    7.73    2.86    2.29    8.16    2.11
STDS:      141     141     141     358     301     251     835     143      14     395

Combined Analytical Condition Arrays:
ELEM:    mg ka   mg ka   mg ka   ca ka   na ka   sr la   ba la   mn ka   si ka   fe ka
SPEC:        1       4       2       3       1       2       3       3       4       5
CONDN:       1       1       1       1       2       2       2       2       2       2
CONDO:       1       1       1       1       2       2       2       2       2       2
KILO:     15.0    15.0    15.0    15.0    15.0    15.0    15.0    15.0    15.0    15.0
CURR:     20.0    20.0    20.0    20.0    50.0    50.0    50.0    50.0    50.0    50.0
SIZE:     10.0    10.0    10.0    10.0    10.0    10.0    10.0    10.0    10.0    10.0


Maybe it would be easier if you just emailed me your MDB file because what you are saying makes no sense to me- unless you're using hundreds of TDI intervals?
john

[Ben Buse]

Here are my results (attached below) from performing 10 stage scans each 32 pixels wide (~4000x or ~94 um wide), using carbon Ka (and nitrogen Ka) on pure Fe (freshly polished and coated with 15 nm of Ag), with each scan repeated on the same area.

Hi John,

I like the idea as I understand it for a single point analysis - what your doing is scaning the stage 1 pixel by 32 pixels. Then you repeat it 10 tens to give TDI. Might be good for unstable samples - TDI for maps. I wondered also could you do the same by deflecting the beam?

The carbon the TDI curve looked very good. What do you think is actually happening regarding contamination. My concern is that you do this stage scan 1 second per pixcel so that around each pixel your depositing carbon (ie. low carbon on 1st pixcel, then each subsequent pixel is on the carbon donut wall of the previous pixel). Then when you do the 10 repeat maps for TDI are you stacking carbon donuts. i.e. lying wall after wall on top of each other.

I surpose my point is that Phillipe work on steels suggested that if you do a single point analysis carbon tends to be deposited around the beam - and as 1 sec movement speed are you producing a TDI curve of carbon build up around the beam - in the donut wall - which you then correct back to zero.

I remember seeing somewhere if you do a very fast beam raster you get the carbon deposited around the raster area.

Ben

[Probeman]

Here are my results (attached below) from performing 10 stage scans each 32 pixels wide (~4000x or ~94 um wide), using carbon Ka (and nitrogen Ka) on pure Fe (freshly polished and coated with 15 nm of Ag), with each scan repeated on the same area.

Hi John,

I like the idea as I understand it for a single point analysis - what your doing is scaning the stage 1 pixel by 32 pixels. Then you repeat it 10 tens to give TDI. Might be good for unstable samples - TDI for maps. I wondered also could you do the same by deflecting the beam?

Hi Ben,
Yes, exactly. Stage or beam scan it should work the same!

The carbon the TDI curve looked very good. What do you think is actually happening regarding contamination. My concern is that you do this stage stage 1 second per pixcel so that around each point your depositing carbon (ie. low carbon on 1st pixcel, then each pixcel is the carbon donut wall of the previous point). Then when you do the 10 repeat maps for TDI are you stacking carbon donuts.

Possibly, except... that the stage scanning is continuous so there are no rings forming during the scan!   That can be seen in the data, don't you think?

Now I haven't performed a coarse (32 pixels over 100 um) scan using the beam.  Does a beam scan use "discrete" positions or continuous scanning?

I suppose my point is that Phillipe work on steels suggested that if you do a single point analysis carbon tends to be deposited around the beam - and as 1 sec movement speed are you producing a TDI curve of carbon build up around the beam - in the donut wall - which you then correct back to zero.

I remember seeing somewhere if you do a very fast beam raster you get the carbon deposited around the raster area.

Again, the stage scans smoothly, so no carbon rings.  But the question is: does the beam scan scan smoothly or discretely?   

Please try it on a fluorescent sample and let us know- you have both a JEOL and Cameca, so please try this coarse beam scan test on both instruments and let us know what you see!

[John Donovan]

Confirmed using the self correction TDI, only one element per spectrometer.

phil

I looked at the MDB file you sent, and so yes, you are doing 60 TDI intervals!   Not hundreds maybe, but a lot. The default is 5 TDI intervals you know. And it seems you are also acquiring an absorbed current measurement for every TDI interval as well...  These kinds of TDI parameters would certainly test the TDI "overhead" calculation...

So the issue is that the software predicts ~2 minutes, but you are seeing acquisition times of ~8 minutes?   

If so, because I do not see this on my instrument, I wonder if either the counting overhead or the absorbed current measurement overhead is much longer with the SXFIVE than the SX100?

Can you tell us how long the same acquisition takes without the 60 absorbed current measurements?   Other valuable data would be running in "time stamp mode" and posting that log output so we can see exactly how long each step is taking on your new instrument.

[Ben Buse]

Possibly, except... that the stage scanning is continuous so there are no rings forming during the scan!   That can be seen in the data, don't you think?

That;s very interesting - and way all x-ray mapping of carbon works is that there is an constant or even deposition of carbon across the map - by using tdi mapping you get back to the original value. Perhaps the best way of analysing carbon at high spatial resolution? - in that you don't need to worry at varying carbon deposition - none under the beam - but some adjacent to the beam

Ben

[Probeman]

Possibly, except... that the stage scanning is continuous so there are no rings forming during the scan!   That can be seen in the data, don't you think?

That;s very interesting - and way all x-ray mapping of carbon works is that there is an constant or even deposition of carbon across the map - by using tdi mapping you get back to the original value. Perhaps the best way of analysing carbon at high spatial resolution? - in that you don't need to worry at varying carbon deposition - none under the beam - but some adjacent to the beam

Ben

Exactly.

Question: are beam scan pixels "discrete" or continuous as they are in the stage scan pixels.

[UofO EPMA Lab]

Possibly, except... that the stage scanning is continuous so there are no rings forming during the scan!   That can be seen in the data, don't you think?

That;s very interesting - and way all x-ray mapping of carbon works is that there is an constant or even deposition of carbon across the map - by using tdi mapping you get back to the original value. Perhaps the best way of analysing carbon at high spatial resolution? - in that you don't need to worry at varying carbon deposition - none under the beam - but some adjacent to the beam

Ben

Exactly.

Question: are beam scan pixels "discrete" or continuous as they are in the stage scan pixels.

I ran a quick test on my SX100 today with a fluorescent sample and interestingly enough, unlike the stage scan pixels, the beam scan pixels are discrete!

So, the proposed TDI scanning method for extrapolating to zero time carbon intensity maybe gets more interesting in the case of beam scanning. With regard to the carbon ring formation...  at least for Cameca instruments...  can anyone chime in on JEOL beam scanning?

I know the JEOL stage scan pixels are continuous...
john

[John Donovan]

Confirmed using the self correction TDI, only one element per spectrometer.

phil

Hi Phil,
Ok, I adjusted the time estimation code for TDI acquisitions (both with and without absorbed current and synch mode).

Try out the latest version and let me know if this is a little more accurate. 
john

PS I swear, these acquisition time estimates are harder to get accurate than the matrix correction physics!