Labbook¶
What we've learnt:
ALWAYS USE REFERENCE MPPC 11!! This is now automatically set.
Use LED 31 for MPPC background tests as some light does seep into light tight clips.
7.1V gives best signal to noise ratio
LEDs 8 and 10 have the worst signal to noise ratio
Red switch on voltage box must go on an hour before white button to the right and the white button an hour before data taking.
White button must be switched off when light tight sheet is open.
To calibrate LEDs remeasure distance from back of gold bar to centre of bar ie LED point.¶
RUN PLAN: Look at broken fibres and damage before putting on testbed. Run 3 waveguides between background runs for the first 24 waveguides then every 4 waveguides thereafter.
Cover ends of light delivering fibres that are entering risen bars and the ends of the bars themselves to prevent stray light seeping into MPPCs.
Problems:
- LED calibration doesnt work. A new method must be found.
- DO NOT USE MPPCConcurrent.Vi its outdated and wont work
- LED 24 broke on 12 Novemeber at layer 5 so all data after and including layer 5 that was on LED 24 is now on LED 22.
Final Calibrations and Backgrounds.¶
- LED Calibration **
- Has been taken from the average from 200 runs on LEDCalibrationWithAttenuation.vi taken on 14/10/14. The files is on the pc in the lab at MICEScan/LED_FINAL_Calibration.lvm. a copy is here:
- LED Error **
- Max 1%.
- Taken from the LED calibration run we did with three sets of 100 and taking the 5 difference from that run as the error. data in LEDError.ods.
- Background Error **
- From 100 background runs Average error =0.93% and max error =1.25%.
- From 361 background runs Average error =0.89% and max error =1.25%.
----------------------------------------------------------------------------------------------------------------------------------------------------
22/09/14
- Waveguide connector is separated in layers of 10 (or 11) with an L shape numbered 1-128.
and read left to right.
- Calibration piece has 4 layers of 10 and 8 layers of 11 fibres separated by pieces of card and numbered with tape.
-----------------------------------------------------------------------------------------------------------------------------------
23/9/14
- The calibration piece sits at LED 5.
- Initially warm up electronics for 1 hour then practice mounting and consider light tightness of calibration piece.
IMPORTANT NOTE:
When inserting the fibres into the clips:
Open the clip
Insert the clip
BE CAREFUL TO LET THE FIBRES RELAX BEFORE CLOSING THE CLIP AS THY MAY MOVE.
- For today
- Loose connector
- Tight connector
- Diffuser between connector
and compare
+Calibration piece test (Tight screws)
For calibration enter a waveguide number of 0, bar number 2. #Readings 50. LED #5
PROBLEM: Low signal from fibre 8. ANSWER: Tested on highest signal MPPC (uncalibrated) and good signal
- Getting readings of ~5.5x10-7 background and 6.7x10_7 data which is much less than before where background ~1.6x10-6 and data 2-6. Checking for loose cables.
PROBLEM: Software plotting time from 0-10 on x axis rather than MPPC number. Try to fix.
- Due to low signal to noise ratio. we are using fibre 1 of the calibration piece to test MPPCs 16 to 6. **
PROBLEM: MPPC 16 has v high background
PROBLEM: MPPC 8 and 10 have very low signal
NB: If an MPPC is tilted up it seems there may be an increase in background but we are not certain.
With tight screws I tested the SelectiveRun.vi and MPPCCalib.vi with Fibre 5 and got the following results
With tight screws and using SelectiveRun.vi and Fibre 1 and got the following results
----------------------------------------------------------------------------------------------------------------------------------------------------------
24/9/14
- Today we are trying to work out why the background is so much lower using MPPCCalib.vi then the selective run. We will then consider the lose screws and with optical grease scenarios to see the main risks. Then LED calib and background MPPC calib will be run.
Actually background on MPPC calib is standard deviation so just the error on the measurement.
- For the rest of the runs we will use RunSelective.vi just to estimate the backgrounds across the other MPPCS. and then record the signal error and background from MPPCCalibration.vi.
PROBLEM MPPC 16 still running with very high backgrounds. So Have removed it from plots.
- Loose and Optical grease tables and plots
- Using MPPC Calibration file MPPC_17;24;35.lvn Background file MPPC_background_17;39;11_2.lvn
-----------------------------------------------------------------------------------------------------------------------------------------
29/9/14
- Ran NormalisationConstants.vi and inputting MPPC Calibration file MPPC_17;24;35.lvn Background file MPPC_background_17;39;11_2.lvn files from the 24th. It was run twice and the files agreed.
We do not understand why the plots above look so different and when/how these calibration constants are applied.
- Running MPPC Calibration to understand how the vi works and to see if the choice of MPPC makes a difference, since LED 5 has a slight gap before the start of scintillator bar. They do not when using LED 5 and 14 both sets of background data with no fibres attached give readings of around 6E-7 in some cases ie those that are nearer to the light seeping from LED 5 where the fibres don't fit neatly in the scintillator 6.1E-7 with LED 5.
*OPTICAL CLIPS ARE NOT QUITE LIGHT TIGHT AND AN INCREASE OF ONE ORDER OF MAG IS OBSERVED IN SOME MPPSs USING LED 5 TO 14.
*WHY ARE BACKGROUND READINGS AS HIGH AS DATA?
Try ramping current to MPPCs¶
Steve tried and found increased signal to noise ration with an additional 2Volts.
- Tried running MPPCs 5 to 15 with 70, 70.3, 70.5, 70.6, 70.7 and 70.8 Volts. Each sees an increase in signal (background not checked will compare afterwards at best voltage).
- Noise increases with signal.
- Look for the optimal signal to noise ratio and determine which current for which MPPC and swap out any dodgy MPPCs.
*Data at low MPPC numbers <8 is not consistent with high MPPC Values so retest low MPPC numbers. Original data saved as MPPCVoltageCalibrationBeforeRetest. New data as MPPCVoltageCalibration.
Voltage tests before data retest
Voltage tests after data retest
- 70.1V gives best signal to noise ratio.
Run The Background tests on MPPC calibration with the first layer of 10 fibres used for the fibre calib and the background as always
This should tell us when we run RunSelective.vi with the background fibre tests and background no fibre tests files the data should now look like a straight line. And that would mean that the wobbly calibrated plot above is the effect of the calibration piece variation.¶
- At 5.19 and a temp of 25.5 degrees we ran a background test for all MPPCs then a fibre background dun but with the first layer of fibres from the calibration piece in MPPCS:
Fibre - MPPC
10 - 5
9 - 6
8 - 7
7 - 9
6 - 10
5 - 11
4 - 12
3 - 13
2 - 14
1 - 15
Files were then used as calibration input to Selective.vi and output raw and calibrated as follows:
A straight line is seen when calibrated showing that the normalisation works and that the wiggly graph above is due to differences in the calibration piece.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
30/9/14
At 12am temp 23.5 degrees.
Run the scans for signal/noise ratio again at 70.1V including as many MPPCs as is possible to reach by moving the scintillator light guide, to determine which to switch in and out to get the best working central 11 MPPCs.¶
- Again using fibre 5.
- MPPCs 6, 8 and 10 (and 16) are worst just ignore these but dont swap out due to central position.
Switch MPPCs and rerun Normalisation¶
- Using layer 5 of 11 fibres, At 1.39 and a temp of 24.2 degrees we ran a background test for all MPPCs then a fibre background run but with the first layer of fibres from the calibration piece in MPPCS:
Fibre - MPPC
51 - 3
50 - 4
49 - 5
48 - 7
47 - 9
46 - 11
45 - 12
44 - 13
43 - 14
42 - 15
41 - 17
Perform Bar test on calibration piece comparing bars 1 and 2 and waveguide with bars one and two.¶
the differences between the milling of the bars is so great that effects are observed in the light delivery at the level of 10%. This systematic is expected to be low but would be so high as to cripple the experiment therefore the bars are being buffed out by the workshop by lunchtime tomorrow.
Software¶
Daniel has made some changes to the software as requested.
-----------------------------------------------------------------------------------------------------------------------------------
1/10/14
- DO NOT USE TODAYS DATA AS THE WRONG REFERENCE MPPC WAS CHOSEN AND NO PRE AND POST BACKGROUND RUNS WERE TAKEN. INSTEAD USE DATA FROM 6/10/14. **
- temp 24.5 degrees.
- covered ends of fibres and bars where light can seep into MPPCs from LED's
- The workshop were not able to complete the bar improvements and will need to do so tomorrow morning now and so we will perform the the fibre insertion tests using one bar (bar 2) that has not been improved yet.
Perform MPPC Insertion Calibrations (100 times)¶
- Since we are using a bar that is not evenly milled there will be some difference in the absolute currents observed between different fibres so we will only use the standard deviations as errors.
- We do not need to include any background calculations since only an average background per row will be taken and therefore in the standard deviation the effect of the background is cancelled out. See the following maths.:
- Taking 50 (from 150) readings every time.
- For each layer, with MPPCS used as follows
Fibre - MPPC
10 - 4
9 - 5
8 - 7
7 - 9
6 - 11
5 - 12
4 - 13
3 - 14
2 - 15
1 - 17
and vi set as:
- We changed the number of readings from 150 to 50 to save time as otherwise a run would be 9 hours long and the human error would have increased so greatly that the test would be invalid.
*finished at 10.30pm with layer 1
--------------------------------------------------------------------------------------------------------------------------------
2/10/14
- temp 24.5 degrees.
- The workshop completed the bar improvements on the calibration piece bars but will need to do the waveguide bars tomorrow. They are greatly improved!
Perform MPPC Insertion Calibrations (100 times)¶
- Taking 50 readings every time.
- For layer 6, with MPPCS used as follows
Fibre - MPPC
62 - 3
61 - 4
60 - 5
59 - 7
58 - 9
57 - 11
56 - 12
55 - 13
54 - 14
53 - 15
52 - 17
and vi set as:
- Dont understand variation in numbers...
- Up to about 12 percent
--------------------------------------------------------------------------------------------------------------------------------
3/10/14
- temp 23.5 degrees
RunPlan¶
- The variation in last nights data and the fact that the current seems to decrease with time. The run plan for today will begin and end with a background run so that any variation in background that occurs over time can be assessed. Also immediately following the run we will run 50 tests without removing the fibre to compare the difference with the actual reinsertion.
- Took background run at 11.30 23.5 degrees.
Perform MPPC Insertion Calibrations (100 times)¶
- Taking 50 readings every time. For layer 12, with MPPCS used as follows
Fibre - MPPC
128 - 3
127 - 4
126 - 5
125 - 7
124 - 9
123 - 11
122 - 12
121 - 13
120 - 14
119 - 15
118 - 17
and vi set as:
- Next background reading at 15.00 pm and degrees.
- the data again had a 50% decrease in background and the same MPPCs as for layer 6 are high error. This may be due to the background, the change of bar or variation in the MPPCs themselves. Analysis must be done.
- Next background test was made.
- running for 50 runs without removing the fibre only took 45 mins as apposed to 3 hours but showed no standard deviation basically.
- All the data can be seen here:
- Next step is to take all of the raw data and remove background one from the first half of the data and the 2nd background run from the second half and then to calculate the average error and standard deviation on the signal (rather than signal + background) and see if they differ.
THIS DOESNT WORK AS...
--------------------------------------------------------------------------------------------------------------------------------
6/10/14
- temp 23.5 degrees time 3.30pm start.
Fibre Insertion tests¶
- Retaking layer one at the same time of day as layer 6 was taken and after the bars have been changed to see if there is any difference to run one the first time around. We think the reason for the low error for layer one on day one is the change in background for the latter two runs being taken earlier in the day. we are taking a background run before and afterwards again so that background subtraction can be taken as before.
Layer 1 data redone:
Background¶
- The background can be seen to vary downwards in a roughly linear fashion for the first three hours (half the following data set and level out after about 3 hours) Therefore we need to take a background run about every hour and for the purposes of insertion tests make a straight line plot for background between its start and end values and use the formula or the line to subtract the expected data at that point from background. The data is in Melissas laptop MICE/Tracker/FibreScans/LongBackgroundTests/ and below
- the backgrounds from the beginning and end of the runs (layer 12 used below) have been turned into a straight line plot which maps quite well the observed change in background. The equation of the background graph is then subtracted from the data.
Background straight line approximations:
Data with background Plots Subtracted:
- Since the standard deviation decreases by a third while the average signal by a half the overall percentage error increases. Nonetheless this is clearly the correct way to treat the data.
- Analysis **
The insertion of fibres can result in an irreducable 10% background that we cannot handle and therefore all layer one tests for all waveguides must be taken together so that the fibres are not reinserted. Backgrounds will be taken with the fibres still in. See analysis below.
--------------------------------------------------------------------------------------------------------------------------------
7/10/14
- temp 23.5 degrees
Bar calibration Tests: Perform Bar test on calibration piece comparing bars 1 and 2 and waveguide with bars one and two. Using new buffed bars.¶
- Calibration piece bar tests:
1. Run MPPCBackground with no fibres in.
2. Insert layer one of fibres using bar 2 and run as though a MPPCFibreBackground test (the same method we used to check the software normalisation).
3. Change bar 2 for bar 1 without taking out fibres.
4. Run MPPCSelective .vi for new bar and use previous two files as data background normalisation.
5. If there's no difference in the bars a straight line should be seen.
LAYER 1:
Ran from 15.27 to 15.57
v1 Bar changed and fibres not moved, v2 fibres reinserted to make sure tight, v3 with all MPPCS switched on, v4 uses the same bar as calibration ans so should produce a straight plot.
The vi of MainSelective was broken after Daniels latest updates. It now automatically assigns fibres to MPPCs 1 to 10 thus messing up the normalisation. The system needs to be put back to how it was before with the MPPCS as the recorded values not the fibres.
STOPPED UNTIL VI FIXED.
Perform LED calibration using LED with attenuation for green fibre.¶
- Using LEDCalibrationWithAttenuation.vi.
- For MPPC 11 only as the difference between MPPC's is irrelevant to this value. This file should only need to be done once!
- Check using MPPC 5 to see if the calibration is the same.
- Measure the distanced from the back of the gold bar holding the MPPCs and measured the lengths to the centre of the scintillator bars and update the file distances.lvm.
- Cannot use clear fibre cut from the T2K roll at the back of the lab. As there are at least 5 lengths of light guide so we need to calibrate too many lengths.
- data:
- The files we took today were wrong as LEDs 16 and 23 are missing and the software did not take care of this properly it just skipped them meaning that the wrong distnaces were used for the attenuation correction from our distances.lvm file. Also when other vi's tread in the data they will assume that the LEDS are concurrent and not that 16 and 23 are skipped so the knock on result would be huge. Daniel is fixing it tomorrow and then we will rerun for MPPC 5. **
------------------------------------------------------------------------------------------------------------------------------------------------
8/10/14
Software
- * LED Calibration was done incorrectly in the software, as LEDs 16 and 23, which dont exist were skipped but that was not accounted for when it evaluated distance to the MPPC in distances.lvm nor when the file was printed nor when the files were considered in mainSelective.vi. The LED calibrations must now be rerun using the same files. *
- reference LED and MPPC must always be set to the same number throughout usable calibration and data. These have been changed to be permanently set in the software. * Ref LED is 8 and Ref MPPC is 11 *
- Mainselective.vi no longer normalises correctly and must be put back to how it was and retested with Daniel here.
LED calibration Round 2¶
- We completed the test using 3 rounds of 100 tests of a green fibre and LEDCalibrationWithAttenuation.vi in MPPC 5. The three sets of data agree well but the LEDS vary a lot as seen below.
--------------------------------------------------------------------------------------------------------------------------------------------------------------
9/10/14
- Today we tested again with a new green fibre to see if the differences are due to the fibre or a real variation in the LEDS. The shape of the data is identical and therefore, we are convince that it is just the LEDs. As we can see below:
- LED 8 is used as the ref LED it has a high reading but this will not effect the error as it will scale.
background calibrations¶
- Can we measure background while fibres are attached by using LED 14 and covering the light that says the LED is on and the fibres that come from the LED so that we can measure the background without having to take out the fibres and thereby without introducing a 10% error. **
- they are the same so we can do it with fibres in and blacked out LED 31.
Bar calibration Tests: Perform Bar test on calibration piece comparing bars 1 and 2 and waveguide with bars one and two. Using new buffed bars.¶
- Calibration piece bar tests:
1. Run MPPCBackground with no fibres in using LED31.
2. Insert layer one of fibres using bar 2 and run as though a MPPCFibreBackground test (the same method we used to check the software normalisation) and LED_dummy_Calib.lvm for LED calib as only one.
3. Change bar 2 for bar 1 without taking out fibres.
4. Run MPPCSelective .vi for new bar and use previous two files as data background normalisation.
5. If there's no difference in the bars a straight line should be seen with only around a 2% error.
LAYER 1 to 12 and Layer 1 retaken:
(the third reading set on the first Layer 1 data involved reinserting the fibres)
- Waveguide bar tests:
Would have been done as above but for the waveguides...
- The data varies by around 20% when the bar is changed but worryingly it changes by 25% between the first and last layer 1 test where it is normalised to itself. This should be the same exactly as before the only difference is that the fibres have been reinserted (which should nly lead to a 12% error). This suggests that the signal to noise ratio changes with time which we have suggestions of from the fibre insertion tests.
- long background fibre tests should be run to consider the effect.
- A new method for examining light yield between bars must be determined.
- finding a way to fix the connector piece in place so that in minimises movement when we change waveguide would be good.
-----------------------------------------------------------------------------------------------------------------------------
10/10/14
New bar tests¶
- We will do this by comparing the first circular and square bars directly .
1. Attach bar 1 to calibration piece
2. Take a background reading using LED 31
3. Take a Fibre background reading using bar 1 and LED 5.
4. Use the output files from steps 2 and 3 as normalisation input for mainSelective using dummy LED calibration as only 1 LED is used. Run for bar 1 to get error in normalisation.
5. Run mainSelective as above but for bar 3 (the first of the square bars) to get Total error in bar from> bar change errornormalisation error.
6. runs step 5 twice and tabulate results.
7. Do the same for layers 6 and 12
8. Look at data to assess if each layer should be checked.
- As there is around a 1mm air gap between the square scintillator and the calib piece whereas the round scintillator piece sits snuggly **
- Depending on the data analysis from the three layers either try to further improve the bar surface, run for all layers, or simply add the average error to all data measurements.
- Layer 1 shows very good agreement around 2%, however layer 6 is very bad showing around 30% variation. This is likely due to the nature of the milling layer one will have a long even horizontal striation while layer 6 will have several vertical ones. We will try to further buff the bars ourselves to improve this effect.
- The bars were polished using water wet and dry grade 1200 and brasso and the visible milling marks were successful removed.
- After polishing the difference in layer 6 had gone down to 9%.
- Do we just use this 9% error or find a better way to fix it? **
Long fibre scan tests¶
- Take long fibre scan tests plotting data over a period of four hours in 4 separate hour runs with background runs between each without removing the fibres to see how the signal and signal - background varies with time for our calibration piece.
- Long fibre scans were taken using StabilityWithoutWaveguide.vi there were four one hour runs taken with backgrounds between each from 4pm till 8.20pm.
- Over the four hours the signal improved by an order of magnitude more than the background went up. therefore, THE SIGNAL TO NOISE RATIO CHANGES WITH TIME.
-----------------------------------------------------------------------------------------------------------------------------
11/10/14
Support to hold calibration piece¶
To reduce the error introduced by moving the calibration piece when changing the bar/waveguides being used, it is proposed that we construct a support that will hold the calibration piece in place. A possible design for it is shown below. The calibration piece could sit within a wooden bar, fixed in place in the securing pieces that already exist in the metal runners to hold scintillator bars.
-----------------------------------------------------------------------------------------------------------------------------
12/10/14
Bar calibrations¶
- We need to make sure that there is no intrinsic difference in the scintillator by comparing a polished bar one and two.
- diffences in light yield across the system must be minimised and recorded at exact fibre positions by an external system to be obtained.
- The run plan for the rest of the time is called RunPlan:
-----------------------------------------------------------------------------------------------------------------------------
13/10/14
- Bar 1 Polished
- connector block made to secure calibration piece
- LED Normalisation happens by LED current/ LED 8 current = Ln
- Ln * ni = normalisation total
- fibres prepared in bundles by length
LED 5 IS BROKEN one of the LED wls fibres has broken!!! We will now use LED 4 instead and move the positions of all other LEDs down to meet the position. Therefore, we must REDO LED Calib!!!!¶
Compare polished calib bars 1 and 2¶
- There is a difference of u to 8% per fibres on layer 1 and layer 2.
- We performed a subsequent test on removing and reinserting the same bar and the results were down to a difference of ~0.5% Meaning that the wiggle error is not a problem. However the variation between different scintillators is a problem so we must definitely measure the exact light out put in each fibre position from bars 2 and 4 (if 4s what we use).
- This was done by comparing signal-Background for two runs with the bar reinserted inbetween with mainSelective.vi.
Waveguide sorting¶
Maths - Minimum errors possible on each systematic to achieve a 10% error total¶
Ri - Bi
Ln * ni
where:
Fi - Bi = ni
Fref - Bref
Fi = fibre calib reading
Bi = MPPC background calib reading
Fref = fibre calib reading for ref MPPC 11
Bref = MPPC background calib reading for ref MPPC 11
------------------------------------------------------------------------------------------------------------------------------
14/10/14
Compare light across bar¶
*Over the space of just ~40 mins -TOOK 1HR 20M and just using signal plus backgd (could use stability.vi for this but MPPCCalib or MainSelective with dummy calib and taking an average would work):*Test layer one on bar x 10 times at 0 degrees,
- 10 times at 90deg and
- 10x at 270deg
- DONE using stability.vi and the data all shows an increase at MPPC 10 ie the middle of the bar... This cold be due to the extra scintillation delivering more light to the centre of the optical window. But at least this can be measured by MPPC. **
Plot of averages from 10 reinsertons vs MPPC number:
- As above but only inserting the fibre up to the beginning of the optical window to see if the light distribution across the bar improves.
- The fibres the fact that most light comes out of the end of the WLS fibres. In this case it will be too prone an error based system and then we will not be able to do a good enough light comparison of light into calib piece and light into waveguide and so we should avoid it. Also when we do the 90 degree test and so the fibres are closest to the WLS the MPPCs overloaded and so this is not a safe way to run the system and we will avoid it.
Plot of averages from 10 reinsertions vs MPPC number:
*Data is found in LightAcrossBarCalib.xls.
- Test signal/noise ration with fibre part in if it was seen to improve light distribution.
- wont do this as we wont use the fibres part in.
LED Calibration¶
- New LED calibration taken. Data is:
*LED calibration is done internally in MainSelective.vi must make the normalisation file ourselves in order to do the fibre by fibre of different length waveguide comparisons.
Pre-Data taking¶
- We have realised that we need to number the system at 90 degrees rotation to how we had it to make the waveguide match up. To do that we are renumbering the calibration piece. We are also doing runs by 11 fibres rather than per layer. So there will be 11 runs the last of which will be 7 fibres.
- Also there are dowels that tightly fix the calibration piece to the waveguide and changing the waveguide may increase the wiggle error therefore we must remeasure it.
- Do this using Stability with waveguide.vi and taking the waveguide on and off and comparing the average percentage errors for each MPPC.
*A 10% error is seen in some cases but this may well be due to the connection that we are making
------------------------------------------------------------------------------------------------------------------------------
15/10/14
Wiggle Test two¶
- It is thought that last night it was too dark and late and that the large errors are a result of that. Therefore today we will, change waveguide with the lights on and MPPC off and having removed it a few times the dowels are a little looser.
- It worked error now 0.3% over 10 runs!!:
Data Taking!!!!!¶
Layer 1¶
- run a normalisation first and run one waveguide
- run a new normalisation and two waveguides
Do this by running an extra fibre run with the first waveguide in place and then using it as the background file for an extra run with waveguide - determine whether this introduces too high an error due to changes in signal/background ratio
- run the other 2 waveguides of blue group
- if having 1 normalisation per 2 waveguides will work, then try having one normalisation for the first 3 waveguides of yellow group, and check again
(Above logic is based upon output of long fibre tests from 10/10/14)
- We can take them every 3 waverguides as the signal/noise ratio doesn't change very much. **
- starting with groups of 5 waveguides 2.3m (Blue) and 3m (Yellow).
---------------------------------------------------------------------------------------------------
16/10/14
- Run layer 2
- After 24 waveguides did a test of signal to noise ratio for 4 waveguides and it was basically the same as expected from the long background and fibre tests where things settle out after 2 hours. Did another check after another 4 (where there was also an added delay due to SW issues) and the runs were
Table of background and fibre and background signal/noise ration calcs.
- As the four waveguide runs are now down to a 20 minute turn around time which is the same as a three waveguide run a background we can now run every 4 waveguides.
Run Order Waveguide Number
14
15
11
34
51
33
35
53
54
55
52
32
12
60
59
29
27
30
-----------------------------------------------------------------------------------------------------------------------------------
Last week of Fibre QA before cm40 from labbook
-----------------------------------------------------------------------------------------------------------------------------------
12/11/14
We arrived back following CM40 to find that one of the bars, Number 3 the one that is used to attach the D0 end.
We will have to continue as the second bar (4) is even more
damaged than 3.
BUT We cannot get them remade in the time available and so we will photograph the damage every day and check that it doesn't get worse if it doesn't we will measure the light output at each fibre position across the bar on the last day and will only be able to compare by fibre in the last 7 layers that we are testing now.
Layer 4 done today.
Pic of damaged bar at beginning of day:
At end of day:
ALSO we don't know exactly which way up the waveguide goes as we are assigning it depending on which fits best at the round end. Therefore we will from tomorrow put a mark on one side of the D0 end (the long side with fibre 1) and record whether the WG is 180deg or 0 degrees. The next day we will then see if the same way up is observed again in all cases when using the same method. If so great we know how to pull it out. If not we will either have to only pull it out for the middle 5 layers or ensure it is the same way up for the same WG on the last few layers and not compare position prior to that.
-----------------------------------------------------------------------------------------------------------------------------------
13/11/14
A black dot has been placed on the waveguide on the long D0 side with fibre one (indents on the left dot at the top) and we are recording the orientation of the bar dot:
Bar damage at start of day:
- LED 24 has broken!!! ** It started giving really low readouts. We are retaking layer 5 data on this run and all previous runs on LED 22.
Retaken the wg's that were on LED 24 and continued without too great a delay. LEDs position was also moved.
While running on LED 18 we started getting values as low as background. The LED cable was found o be very loose and was resoldered. However, the problem was not fixed as we then realised that none of the LEDs were properly working. When the power was switched on and off the LED would light for a flash then switch off both upstream and downstream of the problem LED. It looks as though some sort of short or too low draw on the power was causing the problem. After extensive investigation no short could be found and while trying to think through the problem we switched off the electronics and let them cool for about 20 minutes. After this the problem was fixed. This suggests there is still an underlying problem but as the system is working and overheating could well be the problem.
Happened again an hour later and required half an hours cooling to fix.
Last 8 WGs taken all together with another background reading at the end so that there was no great increase in error as the last calibration was usually not taken. No no happened again but whole run took 20mins so it was very fast and we didnt increase errors much at all anyway.
Turning computer and full system off overnight.
Bar damage at end of day:
14/11/14
At start of day, the overheating problem was discussed with Steve, and he suggested that the power supply could perhaps be on it's way out.
Bar damage at start of day:
Plan for day had been to scan layer 6 (fibres 56-66) however a number of problems were encountered.
Having taken a background and fibre background reading at LED 4, and seeing numbers of orders of magnitude that were consistent with past days (implying that the set up of the calibration piece and the insertion of the fibres into the MPPCs was fine), we proceeded to start scanning with waveguide 60, at LED 15, also in the same orientation (0 degrees) as the previous day. However the values we were seeing were of background levels. A series of checks were performed, with readings taken at each stage to see if the problem had been resolved. These checks were:
- Check connection between waveguide and bar
- Check connection between waveguide and calibration piece
- Visual inspection of waveguide to check for broken fibres
- Check the LED is turning on
- Check fibres coming out of LED had light coming through
- Switch to a new waveguide connected to LED 15
- Switch to a new waveguide connected to LED 17
Still all of the readings obtained were at background levels, however at this point it was realised that as layer 6 is one of the central layers, it coincides with the damage in the bar, which could account for some loss of light. However it should be noted that the damage in the bar only appears to extend ~half way across, so one would expect only half of the readings to be affected. This issue should be revisited when the other issues discussed below have been addressed.
We then moved to LED 21 with waveguide 42, and decided to redo layer 5, to then compare the readings to those taken previously. Again, the numbers obtained were not convincingly above background levels, typically around 8 or 9E-7, which is higher than the average background readings of ~7E-7, but still within the range of values background readings have been known to take.
Using the numbering convention in the D0 diagram above (NB however there is a mislabelling of the fibre numbers in the diagram on one row), it is most likely that if the crack is going to interfere with the light transmission, the fibres most likely to be affected will be 57-72, and if the damage extends further along the width of the bar than it currently appears, fibres 49-80 could also be affected.
Considering that layer 5 is also close to the damage on the bar, with consideration to the possibility that the damage in the bar had increased overnight in a way that isn't visually obvious, the decision was made to redo layer 1 of waveguide 42, as this is furthest from the damage, and compare this to the previous run from 15/10/14. A comparison of the fibre background, the background, the value written to the raw files (which is actually the raw value - background) written under the column "Fake raw", and (Fake raw + background = raw) written under column "Real raw" for the 1st run from 15/10, and the 2nd run from today, is shown in the spreadsheet below.
Observations from spreadsheet:
1) The background values are consistent, which implies that the MPPCs are functioning correctly
2) For the fibre background, for all MPPCs except for MPPC 7, the new readings are lower than the old ones, in many cases they are more than 30% lower than the old numbers
3) For the "fake raw", the same behaviour is seen, except now MPPC 3 and 7 are higher, but otherwise the same decrease in readings is seen, again in some cases > 30%
4) For the "real raw" this is also the case (which makes sense since fake raw just removed the background readings, which were consistent)
This all lead us to the conclusion that the problem must be with the light output from the LEDs. There are two possible sources for this problem; the power source, or the "black box" that regulates the LEDs. A voltmeter was used to check the voltage being supplied at the power supply end, which was confirmed to be 9V. The same check was done at the LEDs, and they are all receiving ~2.9V. However, it is not known if this is what they were getting before the power supply issues arose.
A new power supply has been ordered and is due to arrive on 17/11. If this does not rectify the problem, it would imply that the voltage being delivered from the LED box has for some reason become too low. This can actually be changed manually using a dial on the box, so that is another thing to check once we have a dependable power supply.
Another point is that the introduction of a new supply could mean that the VIs need rewriting, although how difficult or (hopefully) trivial this will be can be assessed once we've connected it up.
Important things to keep in mind/do on Monday:
1) Need to establish whether once the new power supply is in place, we need to change the VIs
2) See if the new power supply fixes the LED light output. If it doesn't, and the voltage supplied to LEDs from their box is still 2.9V, then we will need to adjust the box manually, after first determining what that voltage should be, no doubt by comparing to past data.
3) Because we were getting a lower signal, due to the lower LED light output, obviously the signal/background ratio for the new layer 1 data is lower than that of the old data. However it should be noted that today's measurements were taken earlier in the day than the old measurements, and the signal to background ratio is known to change over time, although this change has not always been consistent, so if it is a factor it cannot be known.
4) If the problems observed today are due to the power supply, given that it started malfunctioning yesterday, this calls into question the validity of the data taken then, and so layer 5 (fibres 45-55) may need to be redone.
5) Need to re-check layer 6 to make sure that the lack of light seen today was due to the LED issue, or if it is because of the crack. If so, using a diffuser plate between the bar and the waveguide may rectify this, although this would probably require further milling of the bar.
6) Should check the status of LED 24, to see whether it was actually a symptom of the larger LED problem, or if it is an independent problem that just occurred at a similar time.
Bar damage at end of day:
---------------------------------------------------------------------------------------------------------------------------------
17/11/14
we need to get the fibres to RAL for Tracker commissioning by the end of the week. So in light of the broken system and bar we have decided to determine an analysis structure to complete early without completing calibrations and determine and carry out essential tasks requiring waveguides in the early week. The tasks completed are below
Checking if the waveguides are attached to the bar the same way up each time:
It was observed during data taking that, given that the calibration piece has to be kept in roughly the same position/orientation when taking data, there wasn't always an obvious orientation to use in attaching the waveguide to the bar so that the ends would match. Accordingly it is possible that different orientations have been used throughout data taking. In order to determine how probable a problem this is, we ran three blind tests, attaching the bar as feels natural, and then recording if it was placed with the dot (mentioned above, and as seen in D0 diagram) up, denoted as 0 degrees, or down, denoted 180. A comparison of the 3 runs revealed 7 instances of disagreement, so a ~11% error. Given that this has been seen with 3 runs, we will do 7 more to see if this is an even larger potential source of error.
Recording fibre damage as seen by eye:
Whilst doing the above, we did an optical check of the fibres in all waveguides to check for breaks, making notes of which fibres were damaged as we went along.
------------------------------------------------------------------------------------------------------------------------------------
18/11/14
Checking effect of crack on light measured in bar
+Developing script to organise groups of fibre x, LED y, angle z
------------------------------------------------------------------------------------------------------------------------------------
19/11/14
+ Rotation of waveguide tests +
LED Calibration
- Use main selective
- Set WG = 2000
- Bar 2
- Use dummy calibrations
- Use round waveguide and bar 2 and fibre made by Kevin, carried by Geoff and Steve
FIBRE HAS COME OUT OF LED 15** We used the calibration file that we took earlier as we had a flase start but we think the old fibre 15 data was ok.
Taken over 30 minutes with no background.
Updated by George, Melissa almost 9 years ago ยท 87 revisions