[Halld-cal] Fwd: Re: [GlueX] FCAL beam test NIM article

[Elton Smith]

Hi Kei,

Thanks for incorporating many of the suggestions on the first draft.
This version is much improved. However, I still have many comments,
which I hope can be incorporated into the next iteration. To provide
feedback, it is also useful to send out the draft with line
numbers, as it makes it easier to reference changes.

In some cases it was easier to copy the text and edit instead of sending specific changes.

------------------------

  page 1, abstract, line 2, suggestion
...at relatively low energies, from 260 down to 110 MeV. [This matches
the order of 14 and 20%]

page 1, abstract, next to last line
...timing resolution was 0.38 ns for a SINGLE 100 mV pulse, ....

page 1, abstract, keywords: add "lead glass, Flash ADCs"

page 1, left column, last paragraph, remove "the"
...equivalent to type F8

page 1, left column last line
...view of one of the 2800 FCAL modules.

page 1, right column first paragraph, suggestion for changes following
the first sentence.

...JLab[8.9]. In the GlueX experiment, proper accommodations are made to
shield the PMTs from the stray fields of the solenoidal magnet of up to
200 G. The configuration consists  of placing the PMT well inside in a
double layer of soft iron and mu-metal shield. The light collection from
the lead glass block to the PMT is facilitated using an acrylic light
guide glued to the PMT and a silicone optical "cookie" to  connect the
guide to the lead glass block. Simulations show that this arrangement
improves the light collection compared to  previous experiments that had
an air gap between the lead glass and the PMT. The primary goal for this
beam test is to measure the improvements of the energy resolution due to
these new additions and validate the final module assembly of the
modules including readout electronics.

page 2, left column second line, remove "mostly" [it is polarized, even if not 100%]

page 2, left column, 4th line
The photon rates are expected of up to 10^8 s^-1 in the coherent peak between 8.4 and 9.0 GeV, leading to a high...

page 2, left column first paragraph next-to-last and last line
... and also help in reducing accidental backgrounds.

page 2, right column, Fig 3
Note: Thanks for adding the lines for the trigger scintillators. Question: Is the lead glass not aligned with the center three paddles? If so, does
introduce a trigger bias?

page 2, right column second sentence add parenthesis
...hitting the detector (see Fig. 5)

page 2, 2 lines down
...connected to a Photonis XP2020 PMT.


page 3, left, above figure
...and angle AT THE EXIT OF the magnet

page 3, left middle second paragraph
...The T-counters FORM A SECOND plane of....

page 3, left, next-to-last line from bottom
...electron energy fractions (\approx 23%).   [Note: why is this plural; is there more than one fraction?]

page 3, right, top sentence,
...hence did not use THOSE runs....

page 3, 2 lines down
...(Sec. 4) ARE NOT VERY SENSITIVE TO the detector alignment

page 3, right column, 3rd and 4th paragraph middle, suggest replacement. [It was easier to rearrange the entire paragraph
than send comments to each line]. Note a couple of sentences from the next page are moved forward, where I think they are more appropriate.

  Each of the channels of the fADCs samples the PMT output voltage every 4 ns, and 50 samples
were stored for each event. Figure 6 shows the read
out samples from a single channel. However, zero-suppression was implemented
by setting a pre-programmed threshold and
only channels that had at least one signal sample larger than
this threshold were written to the output stream. The
threshold applied to each channel was based on the pedestal,
or baseline, which was measured without any input signals.
The threshold was nominally set to 10 ADC counts above the
pedestal. Off-line studies showed that the resolution worsens
with larger thresholds. After data taking, it was discovered that
the thresholds for the modules in the top two rows were inadvertently
set much higher, leading to a significantly worse
energy resolution. Therefore, the sub-sample of events with showers centered on these modules were
not used in the determination of the energy resolution (see Sec. 3.5).

  The module response to energy deposited in the lead glass is obtained
by summing the pedestal-subtracted samples over a range of
time. This range was optimized to be over 20 samples starting
at sample 9. This sum was performed off-line for the beam test, although
for the GlueX experiment the determination of the signal size and time
will be computed in FPGAs to manage the size of the output data stream.

Question: Were 20 samples used, or more? You might want to be precise for the paper.


page 3, right column, last paragraph, middle
...inside trigger SCINTILLATOR, were determined....


page 4, left column, Sec 3.2. par 5, Eliminate first two sentences (moved to previous page)
Begin with "The total signal size for the event is given by..." [eliminate "then"}

page 4, same paragraph last sentence
...that was USED to obtain....

page 4, par 6,
...measure the initial ENERGY resolution.

page 4, par 6 2 lines down
...distribution by SCALING the signal size.... [I assume it is a scaling; more precise than just correction]

page 4, same par, after ref [8]
...constraints and MEASURED THE DIFFERENCES FROM THE MESON MASS were minimized.

page 4, same par, last sentence
...and the CORRECTION FACTORS determined...

page 4, last par, first sentence, I assume you mean
...the distribution of the total signal from all modules for two given runs...

page 4, right column, second paragraph.
I suggest you eliminate this paragraph, as I don't think it contributes much to the paper.

page 4, right column, Figure 7.
You should probably use one significant figure in your resolution numbers (as you do in the text)

page 4 right column, last paragraph. suggestion for rewording, but check that I understood your meaning.

For all runs, the statistical errors from the fits to determine
the widths were negligibly small compared to the systematic uncertaintes due to data selection.
Therefore, various selections of the data were tried in an effort to find a
consistent prescription to obtain the minimal resolution. Although
the correction factors were determined using a subset of the data to
minimize the influence of events in the tails of the distribution, the
resolution for the run was determined by applying the correction factors to all events.
The selection range of the summed signals for....

p 5, left column second paragraph, bottom, suggestion

The gain factors were calculated FOR THE NINE trigger
combinations, then applied to each subset
to calculate the resolution using that subset
before and after the software gain balancing.


p 5, left column next-to-last paragraph, simplify: add one sentence to this paragraph and omit the last paragraph

...Table 1 summarizes the results of our simulation for all energies.
With the assumption that the intrinsic detector resolutions and the spread
of energies add in quadrature, we can subtract this contribution to obtain the
intrinsic detector resolution.

p 5, right column top paragraphs: Omit. The contribution of this correction is small and subtracted,
so no need for further discussion

p 5, right, Subtitle: Just call this 'Energy Resolution'

p 5, right, first par in Sec 3.5: I suggest to simplify and rearrange the ext under 3.5 as follows:

Figure 8 shows the energy resolution for five different
trigger combinations. Three trigger combinations were excluded due to the
incorrect hardware threshold settings and a fourth combination (h3 and v1) was also
excluded because it consistently gave poor resolution, although a clear reason was not identified.
Each marker represents a different trigger
combination, and have been shifted horizontally slightly for
visual clarity.
The measured energy resolution for the three electron energies
of 114.2+/-4.5 MeV, 209.3+/-5.2 MeV, and 263.7+/-5.7 MeV,
as given by the average and standard deviation of the five trigger
combinations, are 19.8+/-0.5%, 14.3+/-0.4%, respectively.
The determination of the contributions to the energy resolution
over this narrow energy range is difficult, so we did not parameterize our final
results.

Our expectation was that the resolution in our
setup would better than that of RadPhi due to the addition of
the optical guides and silicone cookies that couple the lead glass
blocks to the PMTs. This is especially important at low energies,
where photostatistics are the dominant factor in determining
the energy resolution. RadPhi measured the energy resolution
with incoming photon energies between 0:6 and 3:0 GeV
was given in Ref. [8] as

E(%) = 7:3  0:06 pE(GeV) + (3:5  0:8) ; (1)

and is extrapolated to lower energies as the solid blue curve in Fig. 8.
Our resolution results were consistently below the curve given
by the RadPhi, as expected owing to the improved hardware conditions.

p 6 Section 4, first sentence

THE SECOND important performance metric....

second sentence

...Since the fADCs for the FCAl sample the PMT pulses only every 4 ns....

same par, use 507,828 instead of "507 828"

same sentence. Need a period at the end.

same par last line

....range of sample PULSE heights.

p 6 left column Sec 4.1, suggestion to reword first par

An example of digitized PMT signal by the fADC is shown in Fig. 6.
We use the time at which the pulse reaches half the recorded maximum
to specify the signal arrival time. This time, which we call t0, is
chosen because it is in the region of maximum slope, and therefore is
easy to determine accurately.

p 6, right column second par, first sentence, reword

To increase the probability of sampling a single shower, we restrict
the differences to be between adjacent modules,....

same par further down. I don't think you have defined "S0"

p. 7, left column top line
enough -> sufficient

p. 7, Fig 11. The cluster points and the weighted average present a cluttered picture to the reader.
I suggest you either show the cluster points, or the average, but not both. The reason to show the
cluster points is to emphasize consistency between them. Since you already do something similar in Fig. 10,
I suggest to simply show the average with the standard deviation, which summarizes the differences
between measurements. If you choose to show the cluster points, you should add a legend.

p. 7, left column, second and third (last) paragraph. I suggest you merge the paragraphs and lose the first
sentence of the second paragraph (beginning with "To quantify..."). The text will also have to be
modified to correspond to the figure if some of the points are omitted.

p. 7, right column, second paragraph.
The entire description of the comparison of parameter "a" between the current document and ref13 can be
simplified (almost eliminated) if you move it to after par 3, where the value is quoted in terms of ns*mV.
If you quote both results in these units, then the values can be compared directly and the full scale
discussion can be eliminated. You can also quote the resolution at specific pulse heights (such as 100 and
500 mV, which was done in Ref13). For Ref13, I get a=40.4+/-16.3, compared to 8.9+/-0.65 ns*mV. The
values in ref13 are consistent within 2sigma of the present document, but with much larger uncertainties.
[I note, however, that the uncertainties in the Gaussian fits are much smaller for some reason].

general: use "fADC full scale" instead of just "fADC scale"
also note typo for error in the value of a=8.9+/-065 ns*mV

p. 8 left column, conclusions

General: In the summary/conclusions you probably want to give a couple of quantitative results, similar to
what is in the abstract for energy and timing resolution.

specific suggestions for wording
...RadPhi calorimeter, which can be attributed to improvements in the light collection efficiency. Precision
timing measurements of the FCAL signals is a new feature...

last line:
...illuminated by a single shower will allow a determination of the beam bunch and rejection of out-of-time
background.



  Elton Smith
Jefferson Lab MS 12H5
12000 Jefferson Ave STE 16
Newport News, VA 23606
(757) 269-7625
(757) 269-6331 fax

On 3/29/13 11:48 AM, Kei Moriya wrote:
> To those who were involved in the FCAL beam test,
>
> The draft of the NIM article is now available on the DocDB
> at
> http://argus.phys.uregina.ca/cgi-bin/private/DocDB/ShowDocument?docid=2183
>
> The draft has been circulated to the Calorimetry Working Group,
> and revised based on comments I received.
>
> Please let me know if there are any additional comments or requests.
> If you think your name should be on the author list, please let me know.
> For those who are already on the list, please check your name
> spelling.
>
> Thanks,
> 	Kei
> _______________________________________________
> GlueX mailing list
> GlueX at dustbunny.physics.indiana.edu
> http://dustbunny.physics.indiana.edu/mailman/listinfo/gluex