<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=Windows-1252">
<style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style>
</head>
<body dir="ltr">
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
Hi Andrew,</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
It will be nice to get specific information how your script calculating efficiency. Sure, the low current data will help. My suggestion is to take on Dummy target at 2, 4, 8, and 15 uA, as we should use this target soon and because the cut on the Y-target could
help selection of the real track.</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
Is 500k triggers per run enough?</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
Bogdan</div>
<div id="appendonsend"></div>
<hr style="display:inline-block;width:98%" tabindex="-1">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Andrew Puckett <puckett@jlab.org><br>
<b>Sent:</b> Thursday, December 16, 2021 11:48 PM<br>
<b>To:</b> Kondo Gnanvo <kagnanvo@jlab.org>; Bogdan Wojtsekhowski <bogdanw@jlab.org>; Bai, Xinzhan (xb4zp) <xb4zp@virginia.edu>; Liyanage, Nilanga K (nl8n) <nl8n@virginia.edu><br>
<b>Cc:</b> sbs_gems@jlab.org <Sbs_gems@jlab.org><br>
<b>Subject:</b> Re: [EXTERNAL] GEM operating voltage</font>
<div> </div>
</div>
<style>
<!--
@font-face
{font-family:Wingdings}
@font-face
{font-family:"Cambria Math"}
@font-face
{font-family:Calibri}
p.x_MsoNormal, li.x_MsoNormal, div.x_MsoNormal
{margin:0in;
font-size:11.0pt;
font-family:"Calibri",sans-serif}
a:link, span.x_MsoHyperlink
{color:blue;
text-decoration:underline}
p.x_xmsonormal, li.x_xmsonormal, div.x_xmsonormal
{margin:0in;
font-size:11.0pt;
font-family:"Calibri",sans-serif}
span.x_EmailStyle20
{font-family:"Calibri",sans-serif;
color:windowtext}
.x_MsoChpDefault
{font-size:10.0pt}
@page WordSection1
{margin:1.0in 1.0in 1.0in 1.0in}
div.x_WordSection1
{}
-->
</style>
<div lang="EN-US" link="blue" vlink="purple" style="word-wrap:break-word">
<div class="x_WordSection1">
<p class="x_MsoNormal">PS—as an addendum to the fact that the plots you circulated strongly suggest an overall ~10-12% reduction of the tracking efficiency at 25 V below nominal, it seems to me that the effect of lowering (or increasing) the applied HV at any
given background rate/luminosity condition is probably dwarfed by the effect of the background rate and the loading of the HV divider in lowering the overall gain of the GEM, therefore lowering the signal/noise ratio and the efficiency. The issue is that we
cannot just lower the online and/or offline zero suppression thresholds much below where they are now because then we are accepting too much noise. So what is really limiting the efficiency is the low overall gain and the low signal/noise ratio. But we clearly
see that when we lower the beam current the gain (and the efficiency and even the yield of tracks/triggers) goes up. That is why I requested the low-current runs, so we could look at the effect of lowering the HV in the absence of the very severe beam loading
of the HV divider and associated gain drop. </p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">If using individual HV supplies on each of the three GEM foils allows to operate at slightly lower (but stable) gain up to much higher background rates, then maybe the higher-rate experiments aren’t hopeless after all. But that is a question
for the hardware experts, not the software expert.</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">Andrew</p>
<p class="x_MsoNormal"> </p>
<div style="border:none; border-top:solid #B5C4DF 1.0pt; padding:3.0pt 0in 0in 0in">
<p class="x_MsoNormal" style="margin-bottom:12.0pt"><b><span style="font-size:12.0pt; color:black">From:
</span></b><span style="font-size:12.0pt; color:black">Sbs_gems <sbs_gems-bounces@jlab.org> on behalf of Andrew Puckett <puckett@jlab.org><br>
<b>Date: </b>Thursday, December 16, 2021 at 10:46 PM<br>
<b>To: </b>Kondo Gnanvo <kagnanvo@jlab.org>, Bogdan Wojtsekhowski <bogdanw@jlab.org>, Bai, Xinzhan (xb4zp) <xb4zp@virginia.edu>, Liyanage, Nilanga K (nl8n) <nl8n@virginia.edu><br>
<b>Cc: </b>sbs_gems@jlab.org <Sbs_gems@jlab.org><br>
<b>Subject: </b>Re: [Sbs_gems] [EXTERNAL] GEM operating voltage</span></p>
</div>
<p class="x_xmsonormal">Hi Kondo, </p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal">I think it is probably okay to lower most or all of the HVs by -25 V or so (-50 V maybe with more careful study) and I have said that for at least a week now. I also hope both of these runs were “apples to apples” in terms of beam current,
target, trigger threshold, etc. </p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal">However, with all due respect I must point out the obvious self-contradiction when you say
<b><i><span style="background:yellow">“Using the track based efficiency as only way to decide everything on how to operate the GEMs is a huge mistake”</span></i></b> and then you turn around and base your conclusion that we are not losing anything in lowering
the HV on… the track-based efficiency, while <b><i><span style="background:yellow">ignoring the across-the-board ~10-20% drop in the number of tracks found per trigger when operating at lower HV</span>!</i></b> (depending on whether you look at numerator
or denominator the answer is roughly the same). You are making exactly the same mistake that you accuse others of making because in this particular case it supports your preferred conclusion. I have only ever advocated that decisions be made based on sound
evidence and solid analysis about what the data do (and do not) allow us to conclude, and as the author of the analysis software I understand that better than anyone else.</p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal">That very real and significant drop in the yield, which you have now confirmed with 10X higher statistics, was why we initially hesitated to permanently reduce the HV for production running (I had no part in that decision process). In
fact, if you look at the numerator histogram in the plots that you sent, you see that the number of events in the numerator histogram is exactly the same for all four layers in both cases. This is NOT an accident, it is a byproduct of the fact that we are
running with only 4 tracking layers now. The track-based efficiency calculation always requires tracks to have a minimum of four hits to be counted as “efficient”; i.e., to be included in the “numerator” histogram. This is to reduce the bias of the efficiency
calculation by requiring hits in at least three layers OTHER than the layer whose efficiency is being measured. However, because we are running with the bare minimum of redundancy in our tracking setup, in the present context, it allows us a fairly robust
estimate of the reduction in the yield of 4-hit tracks when lowering the HV. </p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal"><b><span style="background:yellow">We know that the fraction of 4-hit tracks that are fake is rather small. The numerators of your efficiency plots for a constant number of BigBite triggers, assuming all other experimental conditions
were the same, clearly show that the yield of 4-hit tracks is reduced by about 11% at -25 V. Now this does not translate directly to an 11% reduction of the reconstruction efficiency for real tracks because some of the four-hit tracks at the higher HV would
still be reconstructed as 3-hit tracks at the lower HV. But the reduction in the overall yield of 3-hit tracks is similar! I had previously estimated that about 50% of 3-hit tracks were fake. The fake-track fraction could be larger or smaller at reduced HV,
but assuming the fraction of 3-hit tracks that are fake stays the same, your efficiency plots imply pretty robustly that you are losing about 10-12% in overall tracking efficiency at 25 V below nominal. But you still need to do an estimate of the fake track
fraction (I can help with this) in both cases to get a more robust estimate of the real change in the yield and/or track reconstruction efficiency.</span>
<span style="background:yellow"></span></b></p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal">If we have compelling theoretical justification (and some circumstantial empirical evidence from our experience of this run) that the “nominal” HVs we are using are dangerous for the long-term stability and reliability of the GEMs, and
that running at -25 V or -50 V meaningfully improves their long-term stability and reliability (I am not yet totally convinced of that), then by all means, we should lower the HVs a bit.
<b><span style="background:yellow">But let’s not pretend that you have conclusively demonstrated that we aren’t losing anything, because the plots you circulated very strongly suggest otherwise, and all you’ve done so far is reproduce the previous result with
higher statistics.</span></b> You have your theoretical reasons for believing we aren’t losing anything, but I go by the data, and the data suggest otherwise, in the absence of major efforts in fine-tuning of the analysis to recover some of the lost efficiency.
It could be that a 10-12% reduction of the overall tracking efficiency is acceptable for the present circumstances. But in GEP these detectors are supposed to operate at about 30X higher background rate than what we have now. If we can’t safely operate them
at full efficiency now, what hope is there for GEP, let alone SOLID or other ambitious high-luminosity experiments? Basically none. If you can only operate your detector safely at ~30-70% efficiency OR at high efficiency but 10-30X lower background rate/luminosity
than advertised, then you need to go back and redesign the whole experiment and explain to the PAC why the capabilities of the technology were so dramatically oversold…
</p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal">Best regards,</p>
<p class="x_xmsonormal">Andrew </p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal"> </p>
<p class="x_MsoNormal"> </p>
<div style="border:none; border-top:solid #B5C4DF 1.0pt; padding:3.0pt 0in 0in 0in">
<p class="x_MsoNormal" style="margin-bottom:12.0pt"><b><span style="font-size:12.0pt; color:black">From:
</span></b><span style="font-size:12.0pt; color:black">Sbs_gems <sbs_gems-bounces@jlab.org> on behalf of Kondo Gnanvo <kagnanvo@jlab.org><br>
<b>Date: </b>Thursday, December 16, 2021 at 8:36 PM<br>
<b>To: </b>Bogdan Wojtsekhowski <bogdanw@jlab.org>, Bai, Xinzhan (xb4zp) <xb4zp@virginia.edu>, Liyanage, Nilanga K (nl8n) <nl8n@virginia.edu><br>
<b>Cc: </b>sbs_gems@jlab.org <Sbs_gems@jlab.org><br>
<b>Subject: </b>Re: [Sbs_gems] [EXTERNAL] GEM operating voltage</span></p>
</div>
<p class="x_MsoNormal">Dear all, </p>
<p class="x_MsoNormal">Please find attached the preliminary analysis on 500k replay (Thanks Sean) of two run where we were comparing the efficiency with nominal GEM HV (red) and with the HV lowered 25 V (blue) on all the chambers.
</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">For all layers, the difference in efficiency is within 1% <span style="font-family:Wingdings">
è</span> it is even higher for the UV GEM Layer 2 at the lower HV setting (which might be due to the opposing effect of the less current drop in the divider)</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">Conclusion is that it does not seems to be any significant difference in the performances.
</p>
<p class="x_MsoNormal">Xinzhan post the same results with 50k replay (less statistic) and I am going to post a log entry for this slides as well</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">Our plan now is to have a run at -50 V when we move back to LD2 production and to compare once again and after we can make a final decision
</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal">Best regards </p>
<p class="x_MsoNormal">Kondo</p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal"> </p>
<p class="x_MsoNormal"> </p>
<div>
<div style="border:none; border-top:solid #E1E1E1 1.0pt; padding:3.0pt 0in 0in 0in">
<p class="x_MsoNormal"><b>From:</b> Bogdan Wojtsekhowski <bogdanw@jlab.org> <br>
<b>Sent:</b> Thursday, December 16, 2021 11:09 AM<br>
<b>To:</b> Kondo Gnanvo <kagnanvo@jlab.org>; Xinzhan Bai <xb4zp@virginia.edu>; Liyanage, Nilanga K (nl8n) <nl8n@virginia.edu><br>
<b>Subject:</b> Re: [EXTERNAL] GEM operating voltage</p>
</div>
</div>
<p class="x_MsoNormal"> </p>
<div>
<p class="x_MsoNormal"><span style="font-size:12.0pt; color:black">I came to the same idea. We must reduce HV today!</span></p>
</div>
<div class="x_MsoNormal" align="center" style="text-align:center">
<hr size="0" width="100%" align="center">
</div>
<div id="x_divRplyFwdMsg">
<p class="x_MsoNormal"><b><span style="color:black">From:</span></b><span style="color:black"> Kondo Gnanvo <<a href="mailto:kagnanvo@jlab.org">kagnanvo@jlab.org</a>><br>
<b>Sent:</b> Thursday, December 16, 2021 11:07 AM<br>
<b>To:</b> Xinzhan Bai <<a href="mailto:xb4zp@virginia.edu">xb4zp@virginia.edu</a>>; Liyanage, Nilanga K (nl8n) <<a href="mailto:nl8n@virginia.edu">nl8n@virginia.edu</a>>; Bogdan Wojtsekhowski <<a href="mailto:bogdanw@jlab.org">bogdanw@jlab.org</a>><br>
<b>Subject:</b> RE: [EXTERNAL] GEM operating voltage</span> </p>
<div>
<p class="x_MsoNormal"> </p>
</div>
</div>
<div>
<div>
<p class="x_xmsonormal">Hi Xinzhan, </p>
<p class="x_xmsonormal">I agree with you and I am making this case for quite a while now and I send a slide about it a few days ago.
</p>
<p class="x_xmsonormal">Using the track based efficiency as only way to decide everything on how to operate the GEMs is a huge mistake
</p>
<p class="x_xmsonormal">That is why the data taken at -25 and -50 V need to be analyze to see what we loose in efficiency. I am sure we don’t loose anything at -25 V and very small at -50 V.
</p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal">Best regards </p>
<p class="x_xmsonormal">Kondo</p>
<p class="x_xmsonormal"> </p>
<p class="x_xmsonormal"><b>From:</b> Xinzhan Bai <<a href="mailto:xb4zp@virginia.edu">xb4zp@virginia.edu</a>>
<br>
<b>Sent:</b> Thursday, December 16, 2021 10:47 AM<br>
<b>To:</b> Liyanage, Nilanga K (nl8n) <<a href="mailto:nl8n@virginia.edu">nl8n@virginia.edu</a>>; Kondo Gnanvo <<a href="mailto:kagnanvo@jlab.org">kagnanvo@jlab.org</a>>; Bogdan Wojtsekhowski <<a href="mailto:bogdanw@jlab.org">bogdanw@jlab.org</a>><br>
<b>Subject:</b> [EXTERNAL] GEM operating voltage</p>
<p class="x_xmsonormal"> </p>
<div>
<p class="x_xmsonormal">Dear Nilanga and Kondo,</p>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">Regarding the newly found dead sector, see HALOG entry:</p>
<div>
<p class="x_xmsonormal"><a href="https://logbooks.jlab.org/entry/3964165">https://logbooks.jlab.org/entry/3964165</a></p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">I checked around, I think that it is possible that this is due to discharge from a large ionization.</p>
</div>
</div>
<div>
<p class="x_xmsonormal">in the attached two papers (I think similar to our application):</p>
</div>
<div>
<p class="x_xmsonormal">Page 4 of "J_D_Swift_1969....pdf", </p>
</div>
<div>
<p class="x_xmsonormal">and Page 6 of "HV_discharge_acceleration_by_....pdf"</p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">I can see a clear increase of the discharge probability when applying a higher</p>
</div>
<div>
<p class="x_xmsonormal">voltage upon a fixed distance. </p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">Maybe we don't need to lower 50 V for all chambers, but lowering 20 ~ 30 V can also make a big difference in reducing the discharge probability. I think reducing voltage can improve the lift-span of our GEM chambers under high beam current
operation. </p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">To me, I am not 100% sure we are losing real tracks considering we have fake tracks. The efficiency from tracking is only around 50%~70% suggests that the fake tracks are more than 20%.</p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">Best,</p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal">Xinzhan</p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
<div>
<p class="x_xmsonormal"> </p>
</div>
</div>
</div>
</div>
</div>
</div>
</body>
</html>