Preshower Tile Test at UVa, 2014

Summary and Next step:

1. Test thicker fibers or fiber with higher dye concentration.
   
2. Will setup COMPASS module, vertically but will be tilted (dark box not tall enough), test MIP response;
3. Setup TDC and test timing response;
4. Test R11102 and R9800 PMTs;
5. Current best estimate for Preshower: 50*1.05(glue or grease)*0.8(connector)*(0.63-0.83)=26 for LA, 35 for FA
   

December 8^th - 18^th, 2014

1. Time resolution measurements using 3-bar tests:
   Using the EJ-200 plastic scintillators (5 cm x 5 cm x 30 cm) with each end coupled with optical grease to a XP2262 PMT for readout. The HVs for the six PMTs were approximately gained match so the MIP would be near channel 1100.
   
2. For the first set of measurements, all three bars were resting on the bottom of the dark box, and the trigger was generated by forming an "AND" of the left and right PMT signals. Finally these "AND" signals were "ORed" together, so that the trigger was the "OR" of the "ANDs". For each bar, the right PMT's signal was delayed by 16 ns using a ORTEC delay box (model DB463). The three trigger bars were oriented parallel to each other.
   
3. For the second set of measurements, a new stand was built so that the bars could be stacked vertically with respect to each other. The three trigger bars were oriented parallel to each other. The vertical spacing is about 24.2 cm between the top of the top bar and the top of the bottom bar. The trigger was generated by forming an "AND" of the left and right PMT signals of the top and bottom scintillator bars. The top right bar's signal was delayed by 16 ns with respect to the other signals so that this PMT would always carry the trigger time. The middle bar was not part of the trigger.
   
4. For all measurements, the PMT signals were split from the dark box with one signal passed to a discriminator channel set at either -40 mV or -80 mV. The other signal was sent through 80 feet (123 ns) of delay cable before being sent to the QDC module.
5. Conclusions and Observations:
   
1. With only one bar in the trigger, the MIP peak sits on top of a large backward, which is suppressed in the vertical bar test. After run 1462, the threshold was doubled to -80 mV to suppress the low energy "peak" near caused by the discriminator threshold. In the three bar vertical measurements, the low energy events show up as a plateau below the MIP location.
2. With the different trigger configurations, the pedestal moved around by about 10-15 channels. Runs 1465 and 1466 were taken to find the pedestals for the PMTs of the top and bottom bars, which cannot be seen in the normal running configuration due to these bars being in the trigger.
3. The energy deposited by the cosmics is attenuated by the shelves of wood that the bars rested and the other bars by about 11% on average from the top to the bottom bars. The thickness of the plywood is 0.75 inches (1.9 cm).
4. The timing resolution of the three bar measurments was determined using the procedure discussed in the CLAS12 FTOF report.
1. For each run, the QDC pedestals were deterimined in order to subtract the offset from the QDC response. The pedestal was isolated by placing an anti-cut on the TDC for the corresponding QDC channel.
2. Then the time from the TDC was plotted versus the pedestal subtracted QDC response (1/sqrt(QDC - pedestal)): run 1464 and 1467. The top right PMT is self-timed, and hence there is no time dependence wrt the QDC signal.
3. A profile histogram was created so that the distribution could be fit: run 1464 and 1467. The slope from the fit was used to correct for the time-walk effect.
4. In a second iteration, a cut was also placed on the pedestal subtracted QDC distributions to remove the low energy events:
   - Top right cut = 800 channels
   - Top left cut = 1050 channels
   - Middle right cut = 750 channels
   - Middle left cut = 750 channels
   - Bottom right cut = 780 channels
   - Bottom left cut = 700 channels
   Applying these cuts was seen to greatly improve the timing resolution.
   
5. Using the expression T = 0.5*(t_t + t_b) - t_m, where t_t, t_m, and t_b represent the time the particle passes through the top, middle, and bottom bars, respectively, the time resolution of the bars was determined.
6. The time can be expressed in terms of the times of PMT signals as T = 0.25*(t_tr + t_tl + t_br + t_bl) - 0.5*(t_mr + t_ml), where the subscript 'r' and 'l' denote the right-side and left-side pmts, respectively, for that particular bar.
   
7. The time resolution is then expressed as σ_ref = sqrt(2/3)* σ_T, where σ_T is determined from a fit to the timing distribution.
8. The timing resolution for the two runs (1464 and 1467) are consistent with each other. Without applying QDC cuts to remove the low energy plateau, a long tail is still seen even after the time-walk corrections are applied: 1467, though the timing resolution does improve by about 50% with the time-walk correction.
9. By applying QDC cuts to the tdc signals in the expression for "T", the long tail is greatly reduced in both the distribution before and after time-wallk corrections: 1464 and 1467.
10. As a last iteration, the QDC cuts were also applied when the time-walk correction was determined. With better time-walk coefficients, the timing resolution improved an additional 17%: 1467. The limitation can be traced to two factors: a non-gaussian tail around 2.4 ns and above and substructure that is seen in the time versus QDC plots near the MIP location. This substructure is not removed by the time-walk corrections and seems to be isolated to only the left-hand side pmts for each bar.
11. This text file contains a summary of the time resolution results.




Timing resolution using three bar method with XP2262 PMTs


run#	Discriminator threshold [mV]	Pedestals	measured MIP positions	comments
1462	-40	file1	file2	Trigger: "OR" of three bars
1463	-80	file3	file4	Trigger: "OR" of three bars
1464	-80	file5	file6	Trigger: "AND" of four pmts top and bottom bars
1465	-80	file7	----	Trigger: "AND" of two pmts only top bar
1466	-80	file8	----	Trigger: "AND" of two pmts only bottom bar
1467	-80	file9	file10	Trigger: "AND" of four pmts top and bottom bars

October 16^th - November 13^th, 2014

1. Test PMT-fiber-tile time resolution: Using the hexagon IHEP tile with 2x 1-mm diameter fibers coupled to a XP2262 PMT for readout. The HV for the PMT viewing the WLS fibers was kept at -2100 V (HV3) to compare with earlier tests using the same tile and PMT.
   
2. For these measurements, the trigger was generated by two scintillator bars from A. Camsonne. Three of the PMTs are similar to XP2262, and the other is a XP2972. The top trigger bar consists of one XP2262 and the other is the XP2972. The bottom bar is readout using two XP2262's. The two trigger bars were oriented parallel to each other.
   
3. The HV settings were adjusted to ensure each PMT had average pulse heights that would pass a -30 mV threshold and were approximately the same amplitude to each other. The HV settings are HV1 = -1425 V, HV2 = -925 V, HV4 = -1525 V, and HV5 = -1325 V.
   
4. The discriminator thresholds were changed from -100 mV to -30.5 mV.
   
5. The delay for the PMT pulse to QDC was changed to 20 ns to avoid having the pulse too close to the closing edge of the gate. This change was made using the PS 792 delay module.
   
6. All four trigger PMTs signals are sent to the TDC after discrimination for timing studies.
   
7. The optical grease on the PMT surface was smoothed out before placing the WLS fiber ends against the PMT surface.
   
8. The fibers were also repositioned into the tile groove, since the outer turns had come out of the groove. No additional optical grease was applied.
9. The first test is a reproducibility study to ensure that the measurements are comparable to the MIP response achieved in the earlier study.
   
10. Conclusions and Observations:
    
1. The series of tests described below indicates some reproducibility issues in the system, which probably indicates a potential instability somewhere.
2. The MIP response compared to run 1384, which should be a similar configuration, was found have a 25% increase (2598/2074). At this point, it is unclear what has caused the increase in amplitude, though perhaps the repositioning of the fibers and smoothing of the grease increased the light collection efficiency.
   
3. The SPE was never clearly identified, and the number of p.e.is unknown, since the MIP is quite different than expected.
   
4. To better identify the SPE, the trigger was changed from the "AND" of all four PMTs to be the "AND" of the top-right PMT and the bottom-left PMT (one set of diagonal PMTs were chosen), i.e., a looser trigger was formed, which allows lower energy events to form a trigger. However, the SPE was not seen in this configuration either.
   
5. Run 1444 was taken with the splitter to send the tile timing information to the TDC. Interestingly, the MIP is comparable to runs 1384 and 1385, which should be similar testing conditions.
   
6. With succesive runs, the MIP value continued to increase. Between runs 1445 and 1446, it was realized that tile PMT might be viewing light escaping from the tile through holes in the tyvek wrapping. Before run 1448, the tile was removed and the tyvek paper was resecured. Also the tile PMT was covered with black material to avoid any unwanted light from entering the PMT.
7. For run 1448, the MIP was still about 15% higher than before, but the SPE was identified and the number of p.e. is close the the maximum achieved for run 1384.


Multi-fiber test results (two Y11, 1mm dia, 1.177m length) with optical grease in grooves


run#	# fiber turns	measured single p.e. position	measured MIP position	measured # p.e. using single p.e. position	comments
1442	2.5 each fiber	SPE?	2598	----	Trigger: four pmts
1443	2.5 each fiber	----	2468	----	Trigger: two pmts (diagonal)
1444	2.5 each fiber	----	1042	----	Trigger: four pmts Using Splitter to TDC
1445	2.5 each fiber	----	2894	----	Trigger: four pmts Without Splitter
1446	2.5 each fiber	----	3093	----	Trigger: four pmts Without Splitter
1447	2.5 each fiber	----	937	----	Trigger: four pmts Using Splitter to TDC
1448	2.5 each fiber	35.2 ± 2.3	2384	67.8	Trigger: four pmts Without Splitter

October 3^rd - 7^th, 2014

1. Test SDU Square SPD tiles: Using a 0.5 cm thick, SDU square tile with a XP2262 PMT. The HV for the PMT viewing the tile was kept at -1850 V to compare with the tests on the IHEP hexagon 2 cm thick tile.
   
2. The delay for the PMT pulse to QDC was changed from 44 ns to 32 ns to avoid having the pulse too close to the closing edge of the gate. This change was made using the PS 792 delay module.
   
3. All runs for this set of tests were taken without the 50-50 splitter.
   
4. The tile was wrapped in tyvek paper and then wrapped with black tape, except for a 5 cm opening on one end. Optical grease was placed on the tile surface and then was placed against the PMT cathode.
   
5. Conclusion:
   
1. The SPE for run 1430 is right next to the pedestal and sitting on the pedestal's tail. So even though the fit error is small, the systematic is larger than half a QDC channel.
   
2. The SPE is easily identified in run 1432 and approximately where expected from earlier data with WLS fibers embedded.
   
3. The NPE value of 48 is considerably more accurate than the value of 65.
   
4. The gain factor for HV = -1850 V and HV = -2100 is determined to be (1579/333) = 4.7, which is significantly larger than the estimated factor of 3.4 from the XP2262B specficiations. This changes the NPE estimate for the IHEP 2 cm hexagon tile direct coupling measurement from earlier.
   
5. Run 1433 with the 3 mm tile had the tyvek paper printed side facing the tile surface. Though only half of the paper had print, this caused a substantial drop in signal amplitude seen in the QDC (~265) versus the expected 945.
   
6. Run 1438 has the 3 mm tile rewrapped with tyvek paper with the printed side facing away from the tile surface. At worse, I would expect the same result as run 1433, but instead found an additional drop of about 30% in amplitude, while the SPE remained in the same position. This indicates an additional inefficiency in light collection compared to the earlier measurement.
   


MIP Response with SDU Square Tile, No Fibers


run#	measured single p.e. position	measured MIP position	measured # p.e. using single p.e. position	comments
1430	5.1 ± 0.5	332.8	65.2	XP2262, HV = -1850 V 5 mm tile
1432	32.8 ± 1.4	1579	48.1	XP2262, HV = -2100 V 5 mm tile
1433	32.4 ± 0.5	264.8	8.2	XP2262, HV = -2100 V 3 mm tile, printed side of tyvek is facing the tile
1436	----	2361	----	XP2262, HV = -1800 V 2 cm IHEP hexagon tile
1438	32.0 ± 0.6	203.1	6.3	XP2262, HV = -2100 V 3 mm tile, rewrapped with printed side facing out

September 26^th - October 2^nd, 2014

1. Test PMT-tile time response: Using IHEP hexagon tile with wide grooves and either a XP2262 PMT or R9800 PMT. The HV for the PMT viewing the tile was reduced to -1850 V to avoid saturating the QDC. The estimated reduction in gain is a factor of 3.37 as determined from the gain vs. HV curve.
   
2. The delay for the PMT pulse to QDC was changed from 12 ns to 32 ns to avoid having the pulse too close to the opening edge of the gate. This change was made using the PS 792 delay module.
   
3. The tile was wrapped in tyvek paper and then wrapped with black tape, except for a 5 cm opening on one end. Optical grease was placed on the tile surface and then the PMT was placed against the open end. It should be noted that there is a very small region where the tile and PMT do not overlap, since the tile is rectangular and the PMT circular.
   
4. Conclusion:
   
1. The rise time of the PMT pulse was determined using an oscillscope and taking the time difference between the time at 10% and 90% of the signal pulse height.
   
2. The rise time from the XP2262 PMT specifications is 2-2.3 ns, depending on the voltage divider, and 1.0 ns for the R9800.
   
3. From the measured rise times, it appears the rise time is dominated by the scintillator material rise time of ?? ns.
   
4. The trigger formed from the "AND" of the top and bottom CSI crystals has a time resolution of 19 ps based on the fit. The trigger going into the TDC is self-timed, since the stop for the TDC is also generated by the trigger. This explains the great timing resolution for the trigger. The smallness of the number (19-20 ps) compared to the TDC resolution is an artifact of histogram binning.
   
5. The time resolution of the tile signal is about 630 ps for both types of PMT, though the peak has long tails. For the first part of these tests, the tile was viewed by only one PMT, and the position dependence of the hit on the tile cannot be removed.
   
6. During run 1413, it became clear that the pedestal was still wide compared to few channels previously seen without the splitter. In fact, the pedestal is about 22 channels wide now, which makes it difficult to find the spe peak. After some lengthy debugging, it was found that the wide pedestals were caused by a series of connectors leading to the QDC input. I replaced these BNC-type connectors with a single lemo barrel connector. The pedestal width is now a few channels without going through the splitter.
   
7. Runs 1413 and 1425 are attempts to find the spe to determine the number of photoelectrons, though the spe was not identified.
   
8. Run 1427 used the R9800 to readout the tile with two 1-mm diameter fibers and grease in the grooves. This run can have an approximate direct comparison with run 1411, which involves the direct light collection from the tile without fibers. There is a factor of four difference (951.5/237.6). However, the light collection for run 1413 might be reduced by having the empty circular groove in the tile.
   
9. Due to the low gain for the R9800, the SPE is not visible. With the same tile and a XP2262 PMT, the SPE is at channel 29 with about 70 photoelectrons. The R9800 and XP2262 have about the same QE (25-26%) at 400 nm, taking the ratio of amplitudes and assuming the same npe, then the SPE should be around channel 3.4, which cannot be isolated from the pedestal.
   


5. Timing Resolution and MIP Response with IHEP Tile, No Fibers
   

   run#	Rise Time [ns]	Measured MIP Position	Estimated #pe	Trigger Time Resolution [ps]	Tile Time Resolution [ps]	Time Walk corrected* [ps]	comments
   1407	2.76 zoomed in image	1353	387-436a	19	631	570	HV = -1850 V; SPE not located; XP2262
   1410	N/A	159	N/A	N/A	N/A	N/A	HV = -1100 V; Gain too low; R9800
   1411	2.48 zoomed OUT image	472.3	----	19	625	522	HV = -1350 V; SPE not located R9800
   1412	----	589.6	----	19	616	----	HV = -1420 V; SPE not located; R9800
   1413	----	951.5	----	19	N/A	N/A	HV = -1350 V; Splitter Bypassed; R9800
   1425	----	936.2	----	20	N/A	N/A	HV = -1350 V; Splitter Bypassed; R9800
   1427	----	237.6	----	---	N/A	N/A	HV = -1350 V; Splitter Bypassed; R9800 readout of tile with two fibers.
   1428	----	342.1	----	---	N/A	N/A	HV = -1450 V; Splitter Bypassed; R9800 readout of tile with two fibers.

   
   
^aThe number of photoelectrons was estimated by taking the reduction in gain factors: 2 for the splitter, and 4.7 as determined on October 3 for the change in HV. The total gain reduction for the combined changes is 9.4. The spe was located around channel 29.2 from the tests with two fibers or 32.6 from the 5 mm tile measurements. Dividing the spe channel by the reduction due to the changes gives 3.1-3.5 channels. npe = MIP/SPE = (1353/3.1) = 436 photoelectrons or (1353/3.5) = 387 photoelectrons.


^*The time walk correction was applied by plotting the TDC time in nanoseconds versus 1/sqrt(ADC - pedestal) and fitting the dependency to a line. This correction was then applied to the raw time using the expression:


time_corr = time_raw - time_tw,

where

time_tw = a/sqrt(QDC - pedestal)

with a being the slope of the line, and QDC is the QDC value for each event.

September 25^th, 2014

1. TDC time resolution calibration: Using the only the top CSI crystal as the trigger. HV stayed the same
   
2. The trigger pulse was delayed going to the TDC by selecting different delays from the Phillips Scientific Dual Delay module. Only the bottom channel was used. Unfortunately, I could not easily measure the other half of the TDC range, since I was unable to easily remove delay between the pulse and the TDC stop using the same test setup.
   
3. The time delay was measured on the lab oscilloscope with a estimated uncertainty of 0.2 ns. Then the TDC channel was measured with runs taken for the various delays listed in the table below.
   
4. Conclusion:
   
1. The splitter was verified to divide the signal 50-50. Hence, any data taken going through the splitter has its amplitude reduced by a factor of 2.
   
2. For run 1401, using the maximum possible delay per channel of the PS 792, no signal was seen in the TDC. No attempt was made to understand this lack of signal.
   
3. The correlation between the added delay and the TDC channel shows a reasonable linear relationship.
   More than likely the error estimate of 0.2 ns is over estimated.
4. The TDC resolution was determined by taking TDC channel without delay (run 1391) as the reference channel and subtracting the results from this reference.
   
5. The measured delay time from the oscilloscope in picoseconds was then divided by the channel difference:
   
   
   resolution_i = (Osc. delay_i)/(1962 - TDC_i),.
   
   where i represents the run number.
   
6. The mean of the 9 measurements was determined to provided a TDC resolution = 36.2 ± 0.6 ps.
   
5. 
   
6. Calibration of CAEN V775 TDC Resolution
   

   run#	TDC Channel	Peak Width	Delay [ns]	oscilloscope measured delay [ns]	comments
   1391	1962	0.56	0	N/A	Baseline Delay to TDC
   1394	1887	0.6	0	2.8	Delay through PS 792
   1395	1444	0.6	16	18.6	Delay through PS 792 + 16 ns
   1396	1223	0.59	24	26.6	Delay through PS 792 +24 ns
   1397	999	0.57	32	34.8	Delay through PS 792 + 32 ns
   1398	778	0.57	40	42.8	Delay through PS 792 + 40 ns
   1399	555	0.57	48	50.8	Delay through PS 792 + 48 ns
   1400	335	0.59	56	58.8	Delay through PS 792 + 56 ns
   1401	N/A	N/A	63.5	66.0	Delay through PS 792 + 63.5 ns
   1402	1778	0.68	4	6.7	Delay through PS 792 + 4 ns
   1403	1558	0.79	12	14.6	Delay through PS 792 + 12 ns

September 12^th-September 24^th, 2014

1. Multi-fiber test: Using the IHEP Preshower hexagon tile. Now use two 1.0-mm diameter, 1.5-m long Y11 fiber. The fiber ends were cut off so that only 20 cm of fiber is not embedded in the tile. The fiber ends were then polished with polishing paper from JLab. HV stayed the same
   
2. Optical grease was coated onto the pmt cathode surface, and the four fiber ends were placed up against the coated cathode. The fibers were held rigid using a plastic block from the machine shop.
3. Conclusion:
   
1. The first test produced a lower than expected number of photoelectrons, especially considering the shorter fiber length. One issue is that the quality of the fiber ends was not assessed after placing the fibers through the narrow holes in the plastic block.
   
2. The fiber ends were repolished, and the test was conducted again with the newly repolished ends. However, only a slight improvement is seen, which is not statisically significant.
   
3. After the first run, the position of the s.p.e. appears to have shifted lower, though it is within 2 sigma from the average.
4. After the aluminized mylar test, it was noted that the inside of the top panel for the mylar is discolored, and the mylar was bulging out, which might have caused light loss reducing the yield for this test.
   
4. The table below includes the data with different turns for each of the two fibers.
   


5. Multi-fiber test results (two Y11, 1mm dia, 1.177m length)
   

   run#	# fiber turns	fiber length embeded (cm)	measured single p.e. position	measured MIP position	measured # p.e. using average single p.e. position: 28.9	comments
   1371	2.5 each fiber	155.4	30.2 ± 1.0	1418	49.1	Optical grease on PMT Tyvek wrapping
   1372	2.5 each fiber	155.4	27.5 ± 1.3	1429	49.4	Fiber ends repolished Tyvek wrapping
   1373	2.5 each fiber	155.4	28.3 ± 1.4	1345	46.5	Aluminized mylar wrapping
   1374	2.5 each fiber	155.4	26.8 ± 1.9	1479	51.2	Reproducibility of Run 1372
   1375	1.5 + 2.5 fibers	120.1	27.8 ± 1.0	1324	45.8	Bottom fiber: 2.5 turns; Top fiber: 1.5 turns
   1376	1.5 fibers each	91.8	27.4 ± 1.3	1048	36.3	Bottom fiber: 1.5 turns; Top fiber: 1.5 turns
   1378	0.5 + 1.5 fibers	63.5	29.5 ± 0.5	970.5	33.6	Bottom fiber: 1.5 turns; Top fiber: 0.5 turns
   1379	0.5 fibers each	35.3	29.4 ± 1.2	630.2	21.8	Bottom fiber: 0.5 turns; Top fiber: 0.5 turns

6. Now repeat the steps above but with optical grease in the tile grooves. The grease is Saint-Gobain BC-630, and was applied with a thin plastic cable tie.
   
7. Conclusion:
   
1. The addition of optical grease improves the yield on average by a factor of about 1.4.
   
2. The variation in improvement is likely due to how well the grease is applied. For example, the lower improvement for less fiber turns is probably due to difficulty in placing the grease deep into the narrow grooves. For the 2.5 turns per fiber, the last turn of the top fiber is difficult to keep inside the groove, but this is true with or without grease.
   
3. For run 1385, the PMT pulse was sent through the splitter, where half was taken to QDC and the other half to a discriminator (-40 mV threshold) and to the TDC. For run 1386, the splitter was bypassed and the pmt pulse passed directly to a discriminator and then to the TDC.
   
4. The full scale range (FSR) was set to 850 ns (212.5 ps/ch resolution). However, the timing difference between the trigger and stop and the tile signal and stop were measured on the oscilloscope to be 182 ns and 218 ns, respectively. If a TDC resolution of 212.5 ps/ch is used, then the time measured by the TDC is a factor of 2 too large. The oscilloscope measured values indicated that the resolution is really about 106.25 ps/ch and the FSR = 425 ns. It's not clear why this is. When CODA is downloaded, it claims that the FSR is set to 850 ns. The calculation in the JLab V775 library appears to be correct. An independent TDC resolution measurement needs to be performed to clarify the actual resolution, though no further measurements are planned at FSR = 850 ns.
   
5. The tile timing resolution does appear to be affected by the splitter. Without passing through the splitter, the timing resolution is about 100 ps better.
   
6. The difference in trigger time resolution is probably due to the quality of the fit and is probably closer to 78 ps.
   


Multi-fiber test results (two Y11, 1mm dia, 1.177m length) with optical grease in grooves


run#	# fiber turns	fiber length embeded (cm)	measured single p.e. position	measured MIP position	measured # p.e. using average single p.e. position: 29.2	Ratio of Grease to no Grease Result	comments
1380	0.5 fibers each	35.3	29.5 ± 0.6	843.4	28.9	1.33	Bottom fiber: 0.5 turns; Top fiber: 0.5 turns
1381	0.5 + 1.5 fibers	63.5	29.4 ± 2.4	1337	45.8	1.36	Bottom fiber: 1.5 turns; Top fiber: 0.5 turns
1382	1.5 fibers each	91.8	29.0 ± 0.6	1673	57.3	1.58	Bottom fiber: 1.5 turns; Top fiber: 1.5 turns
1383	1.5 + 2.5 fibers	120.1	27.0 ± 2.2	1951	66.8	1.46	Bottom fiber: 2.5 turns; Top fiber: 1.5 turns
1384	2.5 each fiber	155.4	28.5 ± 2.2	2074	71.0	1.39	Bottom fiber: 2.5 turns; Top fiber: 2.5 turns
1385	2.5 each fiber	155.4	?	1019	69.8a	N/A	Using Splitter to TDC FSR = 850 ns

^aThe average single p.e. position divided by 2 was used.



Timing Resolution with Multi-fiber test results (two Y11, 1mm dia, 1.177m length) with optical grease in grooves


run#	# fiber turns	Measured MIP Position	Trigger Time Resolution [ps]	Tile Time Resolution [ns]	comments
1385	2.5 each fiber	1019	52	1.117	Using Splitter to TDC FSR = 850 ns
1386	2.5 each fiber	N/A	78	1.013	No Splitter to TDC FSR = 850 ns

September 5th-September 9th, 2014

1. 2-mm diameter fiber test: Using a IHEP Preshower hexagon tile with wider grooves wrapped in Tyvek. Now using one 2.0-mm diameter, 3-m long Y11 (200 ppm) fiber. HV stayed the same. The gooves are 7.5mm deep, with diameter 9cm and each side "leg" approx. 3.5cm long towards the end of the tile.
   
2. The yields are much lower than we expected-- if we assume the same bending loss and attenuation length as the 1mm fiber. (For example, the test gave 25 p.e. for 5.5 turns). However, a bending loss in the 25%-per-turn range can explain the yield well. And this is consistent with the 2mm fiber being non-S type, and our grooves have 9cm diameter.
   
3. Additionally, as the fiber was unwound to reduce the number of embedded turns, it was noticed that there are several cracks along the fiber where it was wound into the tile. This probably explains the cause for the lower yield.
   


4. 2-mm diameter fiber test results (one Y11 (200), 2 mm dia, 2.0 m length), no grease in grooves
   

   run#	# fiber turns	fiber length embeded (cm)	measured single p.e. position	measured MIP position	measured # p.e. using average single p.e. position	#p.e. per fiber length (cm-1)	# p.e. expected if using minimum lengthb
   1352	5.5	162.5	34.9	862	25.0	0.154	~
   1353	4.5	134.2	33.5	814.6	23.7	0.176	~
   1354	3.5	106.0	34.8	781.4	22.7	0.213	~
   1355	2.5	77.7	35.0	787.1	22.9	0.295	~
   1356	1.5	49.4	34.4	722.5	21.0	0.425	~
   1357	0.5	21.1	33.6	451.8	13.1	0.621	~

^aThe average single p.e. position is 34.4
^bminimum length is embeded length + 20cm on each side (40cm total), using 3.5m as decay length

September 3rd-September 5th, 2014

1. 1-mm diameter fiber test: Back to IHEP Preshower hexagon tile with 1.0-mm diameter, 3-m long Y11 fiber. The gooves are 7.5mm deep, with diameter 9cm and each side "leg" approx. 3.5cm long towards the end of the tile.
   
2. The results should be compared against those from run 1300, which found 37.7 photoelectrons; this represents a 17% reduction in the number of photoelectrons from earlier studies. The fiber ends had dark spots, and it was decided to polish the ends to check for improvement.
   
3. The fiber ends were polished with 3 micron polishing paper from JLab. After polishing, 35.2 photolelectrons were observed, indicating that the low number of photoelectrons were due to poor light collection from the darkened fiber ends.

August 4th-September 2nd, 2014

1. Shashlyk COMPASS-II Module test: Using the same CSI blocks and pmts for triggers as in the preshower tile tests. HV stayed the same for all runs, except the last, where the gain was decreased by a factor of four on the shashlyk PMT.
   
2. Conditions:
   
1. For most of the test runs, the shashlyk module was oriented vertical, except where noted as horizontal.
2. The top CsI crystal was located on the top shelf of the dark box, and the bottom CsI crystal was located on the lower shelf of the box. The vertical separation between the two crystals was about 30 inches (~76 cm). Most of the time, the crystals were oriented perpendicular to each other to narrow their area to better match the area of the shashlyk module.
3. The overlap area of the crystals was 5 x 5 cm², and the lateral area of the shashlyk is 3.8 x 3.8 cm².
4. The timing of the shashlyk signal wrt the gate to the QDC was adjusted by adding 20 ns of delay using a 4-ns lemo cable and 16 ns from a PS 792 delay module.
5. Run 1331 and runs from 1345 have the splitter removed from the shashlyk cabling. For other runs, the shashlyk signal is split and sent to the QDC after delay and also to a discriminator and channel 0 of the v775 TDC.
6. From run 1348, the CsI trigger thresholds were set to -100 mV, previously set to -40 mV to remove low energy triggers.
3. Conclusions:
   
1. With the signal passing through the splitter, it was found that the shashlyk signal did not produce saturation at high QDC channels, run 1323.
2. The splitter at UVA was found to both shift the pedestal and broaden the pedestal by about a factor of 10, from 1-2 channels up to 10 channels wide.
3. The saturation effect of the QDC was tested using the signal from top CsI crystal into the QDC, run 1334. The shelf above channel 3400 are events that saturated the QDC.
4. Since, the MIP was hard to identify with the shashlyk vertical, some data was taken with the shashlyk oriented horizontal and lying on top of the bottom CsI crystal. For most of these tests, only the bottom CsI crystal formed the trigger. From the horizontal tests, the MIP was easily identifiable, run 1337.
5. In the table below, the threshold refers to the trigger threshold either coincidence, bottom CsI, or shashlyk. The shashlyk discriminator threshold is typically -40 mV, unless otherwise noted. The efficiency is defined as the number of events with a TDC hit compared to all triggered events.
6. This figure shows a comparison of a few of the different configurations: shashlyk_horizontal_tests.jpg. The red and green curves are from run 1337 with and without a TDC cut, respectively. The black curve curve is from run 1344, and the magenta curve is from run 1345, where the splitter has been removed from the circuit. Removing the splitter increases the amplitude by about a factor of two, while narrowing and shifting the pedestal to lower channel.
7. Triggering directly on the shashlyk, run 1344, indicates that the module sees numerous events with low energy deposition. These are eliminated by triggering on the CsI crystal below.
Horizontal Shashlyk Tests


8. run#	Trigger	Rate [Hz]	Trigger Threshold [mV]	Efficiency [%]
   1337	CsI Coincidence	0.04 Hz	-40	N/A
   1338	Bottom CsI	330	-40	N/A
   1339	Bottom CsI	11.5	-250	3.25
   1341	Bottom CsI	2.8	-500	16.4
   1343	Bottom CsI	2.1	-500 shashlyk threshold = -30 mV	26.6
   1344	Shashlyk	22.7	-40	N/A
   1345	Bottom CsI	2	-500	N/A
   1346	CsI Coincidence	0.01	-250	75.

   
   
4. The vertical shashlyk tests were resumed with runs 1348 and 1349. The CsI discriminator threshold was set at -100 mV, and the HV was decreased from -2105 V to -1805 V between runs 1348 and 1349 for the shashlyk PMT. For run 1348, saturation was seen above channel 3500. Reducing the gain removed the saturated events as seen in run 1349. A comparison of the two spectra with different gains is shown here. Besides a cusp around channels 500-600, nothing else is prominent.

May 22nd-May 27th, 2014

1. Multi-fiber test: Using same IHEP Preshower hexagon tile as the Tyvek test of last week. Now use two 1.0-mm diameter, 1.5-m long Y11 fiber. HV stayed the same
   
2. Conclusion:
   
1. Two-fiber test results are higher than expected, indicating an attenuation length of the fiber shorter than 350cm, or a bending loss. A rough "fit" to all tests starting 4/21 indicates that either the attenuation length is 200cm, or the attenuation length is 350cm but there is a 6%/turn bending loss.
2. The "# p.e. expected if using 2mm-dia fiber and mim length" (right-most colume) of all tables starting 4/21 have been updated using 350cm attenuation length and a 6%/turn bending loss.
3. Both the expected yield using two fibers at min length and the expected yield using 2mm fiber and min length are higher than before, with the double-1mm-dia fiber configuration "winning" slightly. Also, the 2mm-dia calculation assumed the same 350cm attenuation length which is probably too optimistic. Kuraray informed me they do not make S-type fibers (better performance in bending than non-S type) for 2mm diameters and above. All 1mm-dia Y11 fibers we used in the test are S-type.  Which one is better awaits for further testing.
3. 
   
4. Multi-fiber test results (two Y11, 1mm dia, 1.5m length), no grease in grooves
   

   run#	# fiber turns	measured single p.e. position	measured MIP position	measured # p.e. using average single p.e. position	# p.e expected if using minimum length (two fibers)	# p.e expected if using 2mm dia fiber and minimum lengthc
   	3.5 each fiber	N/A	N/A	N/A	~57	~56
   1307	2.5 each fiber	31.1	1644.3	51.1	~54	~52
   1308	1.5 each fiber	33.3	1212.5	37.7	~40	~37

May 7nd-May 21st, 2014

1. Tyvek test: Back to IHEP Preshower hexagon tile with 1.0-mm diameter, 3-m long Y11 fiber. The gooves are 7.5mm deep, with diameter 9cm and each side "leg" approx. 3.5cm long towards the end of the tile.
2. Result: yield in average 10% above printer paper wrapping.
   

May 2nd-May 6th, 2014

1. Start setting up the TDC. Hardware seems to be working. Need to adjust the delay and add TDC to the analyzer.
   
2. Aluminized mylar test: Back to IHEP Preshower hexagon tile with 1.0-mm diameter, 3-m long Y11 fiber. The gooves are 7.5mm deep, with diameter 9cm and each side "leg" approx. 3.5cm long towards the end of the tile.
1. Existing data on mylar: see (result of google search): PHENIX test on Tyvek 1055B (1997); Another BNL test (1996)-flawed in Al. Mylar;
2. Summary of reflectivity info: Tyvek 1055B: 95%; Al Mylar: 90% (from more google search); copier paper: 75-80%.
3. Result: yield is in average 17% above printer paper wrapping, systematic uncertainty at the 5-10% level.

1. run#	# fiber turns	fiber length embeded (cm)	measured single p.e. position	measured MIP position	measured # p.e. using average single p.e. position a	#p.e. per fiber length (cm-1)	# p.e. expected if using minimum lengthb	# p.e. expected if using 2mm dia fiber and minimum lengthc
   1280	6.5	190.8	25.0	1114	41.5	0.218	45.8	62.9
   1281	5.5	162.5	26.2	1066	39.7	0.244	45.7	59.6
   1282	4.5	134.2	27.4	920	34.3	0.255	41.0	50.9
   1283	3.5	106.0	27.5	786	29.3	0.276	36.5	43.0
   1284	2.5	77.7	27.2	668	24.9	0.320	32.3	36.1
   1285	1.5	49.4	25.0	490	18.3	0.370	24.7	26.2
   1286	0.5	21.14	28.8	265	9.9	0.467	13.9	14.0

   ^aThe average single p.e. position is 26.81
   ^bminimum length is embeded length + 20cm on each side (40cm total), using 3.5m as decay length
   ^cassuming same depth of fiber embedding (thus half embedded length) + 20cm on each side, using 3.5m as decay length
   ^dall expected value assumed 6%/turn of light loss due to bending

April 29-May 2nd, 2014

1. Test IHEP Preshower hexagon tile with 1.0-mm diameter, 3-m long BCF fibers. The gooves are 5.5mm deep, with diameter 9cm and each side "leg" approx. 3.5cm long towards the end of the tile.
1. Embed 4.5 turns, took a movie with the scope readout to compare: BCF92 vs. Y11. Also see this directory for snapshots of MIP.
2. Existing data: A quick google search found a recent super-B test showing ~60% for this comparison for straight fibers. ATLAS test (Fig.5-24 on EC TDR) indicated that for a bending radius of 10cm BCF91A loses >20% of light (while Y11 shows minimal loss). Our observed loss could be a combination of the yield (absorption/emission efficiency) and the bending loss.

3. run#	fiber type	# fiber turns	# p.e.	# p.e., relative to Y11 with the same # turns
   1267	BCF92	3.5	10.5	37% (run 1262)
   1268	BCF91A	3.5	13.7	53% (run 1263)
   1269	BCF91A	2.5	10.7	52% (run 1264)
   1270	BCF91A	1.5		57% (run 1265)

   
   
2. 
   

April 21-28, 2014

Test IHEP Preshower hexagon tile with 1.0-mm diameter, 3-m long Y11 fiber. The gooves are 7.5mm deep, with diameter 9cm and each side "leg" approx. 3.5cm long towards the end of the tile.

March 31-Apr 9, 2014

1. single WLS fiber output: nphe=469.3/24.8=20 (single p.e.; mip) <-- this is comparable to the previous result for 2 WLS fibers;
2. with Leoni fiber connector, optical grease, pushed both fibers in contact, then placed gently: spectrum (no mip is observed). Played around with the connector and saw no difference;
3. single WLS -> Delrin scrap piece with a 1mm through hole -> clear fiber, pushed both fibers in contact with a dab of optical grease, then placed gently down. No loose movement of fiber because the hole diameter matches fiber well: (mip). Loss of 2-m clear and Delrin connector combined: 76%, nphe ~ 14.3. Not bad!  Picture of setup.
   

March 13-22, 2014

1. nearly 8 turns (run 1242) nphe=917/26.8=34.2 (single p.e.; mip), nphe per turn = 4.3;
   
2. 7 turns (run 1244) nphe=912.1/24.9=36.6 (single p.e.; mip), nphe per turn = 5.2;
3. 6 turns (run 1245,1247) nphe=788.2/24.1=32.7 (single p.e.; mip), nphe per turn = 5.5; <- note this run could have had only 5.5 turns
   
4. 5 turns (run 1248) nphe=788.3/24.8=31.8 (single p.e.; mip), nphe per turn = 6.4;
5. 4 turns (run 1249) nphe=633.7/24.4=26.0 (single p.e.; mip), nphe per turn = 6.5;
6. 3 turns (run 1250) nphe=500.9/25.1=20.0 (single p.e.; mip), nphe per turn = 6.7;
7. 2 turns (run 1253) nphe=378.0/23.0=16.4 (single p.e.; mip), nphe per turn = 8.2;
8. 1 turn (run 1254) nphe=203.8/21.4=9.5 (single p.e.; mip), nphe per turn = 9.5;

March 5-12, 2014

1. nearly 5 turns (run 1232) nphe=223.3/28.3=7.89 (single p.e.; mip), nphe per turn = 1.58;
   
2. 4 turns (run 1233) nphe=191.8/26.0=7.38 (single p.e.; mip), nphe per turn = 1.85;
3. 3 turns (run 1236) nphe=150.4/28.0=5.38 (single p.e.; mip), nphe per turn = 1.80;
4. 2 turns (run 1238) nphe=110.8/(~29)=3.7 (single p.e.; mip), nphe per turn = 1.85;
5. 1 turn (run 1239) nphe = 56.3/36.5=2.1 (single p.e.; mip), nphe per turn = 2.1;
6. Data showed that nphe yield proportional to fiber volume. With the same volume, thicker fibers (1mm, see Feb28 run below) seems to give 15% more light (compare to 0.5mm, 4 turns above). Based on this, using 1mm-fiber and 12 turns will give about 100 ph.e.
   

February 28 - March 1, 2014

1. reduced fiber to single turns. See pictures: pic1 pic2 pic3;
   
2. MIP response: chin_tile_20140228_singleturn.pdf or PNG; Single p.e. at about ch25. Number of p.e. about 8.5 which is what we expected if it is linear with fiber length embedded.

1. 2.2 turns (run 1229) nphe=409.5/26.5 = 15.5(mip), nphe per turn = 7.0;
2. 1 turn (run 1230) nphe = 213/25 = 8.5(mip), nphe per turn = 8.5;
   

February 27, 2014

• LHCb(Russian) tile:

1. 1.5cm thick or 18/32", 12x12cm^2 square;
2. fiber (Y11) embedded 3.5 turns (nearly 4 turns if including the straight portion);
3. fiber length about 30cm on either side (see December 2013 record and picture)
   
4. single p.e. peak around 26; MIP peak around 541; see russ_tile_20140227_3.5turns.pdf or PNG
   

• Chinese tile:
1. 2cm thick or 24/32", 6.25-cm side hexagon;
2. We inserted a 3-m long Y11 fiber, embedded 2.3 turns (the 3rd turn keeps popping out, only partially in the tile). The fiber sticked out pretty long before the PMT end (at least 1m more decay length compare to LHCb tile).
   
3. single p.e. peak around 27;
4. Based on LHCb tile response we expected the Chinese tile MIP peak to be at: 24/18*2.3/3.5*27/26*541=492 +/- 20 given the error on determining the s.p.e. peak. 
   
5. We measure 497 for the four hour run 1228, see Chinese-tile-20140227-4hour_2turns.pdf or PNG
   
6. An overnight run gave slightly lower MIP response, could be because part of the fiber popped out, see pictures: pic1 pic2 pic3; MIP response:  chinese-tile-20140227overnight_2turns.pdf or PNG
   

February 26, 2014

1. ;