CALICE MAPS Meeting, RAL, 31/03/08
==================================
Present: Jamie Crooks, Paul Dauncey, Anne-Marie Magnan, Yoshi Mikami,
 Marcel Stanitzki, Renato Turchetta, Mike Tyndel, Nigel Watson, John
 Wilson

Minutes: Paul


Minutes of previous meeting: No changes.


Upcoming talks: There is an SiD meeting at Coseners' House on 14-16 Apr,
 with the calorimeter session on Tue 15 Apr. At the end of that week,
 there is a CALICE-UK meeting at Birmingham on Thu 17 Apr, followed by
 an LCUK meeting on Fri 19 Apr. We can get a 15-20min MAPS talk at the
 SiD meeting and a ~30min talk at the CALICE-UK meeting. Nigel and
 Anne-Marie will attend both so either could give either talk. As Nigel
 may be busy with the organisation at Birmingham, it was thought better
 if he gave the SiD talk and Anne-Marie gave the CALICE-UK talk.


Political status: The funding outcome is not yet known and will be
 slower than stated in the last meeting. A consultation committee is
 looking through the Programmatic Review outcome and according to Phil
 Burrows, who is a member, it will give a first statement on Apr 24,
 with a final report due on May 16. There has therefore been no real
 feedback to CALICE.

 Paul will be talking with Janet Seed on Wednesday this week to discuss
 the funding. She indicated we were close to converging, which sounds
 hopeful. She also said the ILC detectors (CALICE and LCFI) would not
 have funding determined by the consultation committee.


Sensor studies: Jamie showed some slides of his work on sensors; see
 usual web page. After adjusting the trim bias resistors, he finds the
 full trim range shifts the shaper threshold curves by ~100 threshold
 units (TU), which implies 1 trim unit is ~7TU. This seems odd as this
 was roughly the ratio seen by Paul, shown in the trim plot in the
 meeting of 19/11/07 (see usual web page), without any resistor changes;
 this was sensor #4. One of these may have been an exception; Jamie will
 cross check his measurements on a different sensor. The ability to trim
 to within ~1TU is needed if the expected noise of 40e- ~ 5TU in the
 shapers is achieved, so ~8TU is larger than required. Note, DAC #8 and
 #11 are the two which control the trim bias currents.

 Mike suggested the odd shapes in the samplers could be due to some
 common mode noise picking up on many channels. It is not clear how
 to check this hypothesis.

 Of the three sensors mentioned in the last minutes, none are so far
 working. Jamie has had six more mounted since then, of which only two
 work. Paul will take three (sensors #10, #14 and #20) to Imperial for
 modifications. The one taken after the last meeting, sensor #9, works
 normally.

 Jamie also showed a spreadsheet indicating that many wafer #10 chips
 have a similar problem with the onward read enable (ORE) which prevents
 them reading out correctly. He has found some extra wafer #8 chips and
 will try mounting some of them to check the problems are not due to
 a change in the assembly or bonding procedure.

 Marcel showed slides of his work on checking the basics of the two
 sensors #6 and #16. It should be possible to check the read back
 configuration data from the PCB and sensor to see if it was corrupted
 in any of the odd cases seen; in particular the "row 7" run. He sees an
 increase in the rate of hits for samplers later in the bunch train.
 This agrees with the observation from Jamie Ballin previously; see
 the last slide in the talk by JamieC in the meeting on 08/02/08. It is
 not known if the dip in hits at timestamps < 200 is isolated to
 particular regions.

 [Note added after the meeting: Marcel distributed some slides with
  more information on some of these points; see usual web page.]

 Nigel showed a few slides from Owen of a first look at uniformity of
 the pixel turn-on point. He looks for the lowest threshold value which
 has zero hits in each pixel and plots the value. He sees two peaks with
 low statistics and looking at ~700 pixels. It would be good to try this
 for all pixels and with more statistics; Jamie already has data
 suitable for this.

 Paul showed some studies of pedestal uniformity. The odd behaviour
 seen makes defining the pedestal difficult. It would be useful to
 check the reproducibility of the effect with other sensors and also
 with the enabled pixels more widely separated.


Future measurements: Paul asked which measurements we should now be
 aiming to do. These should be aimed at understanding the sensor,
 finding what to change in the next round, and improving the data so
 they can be publically released.

 The laser system is probably the only place which can realistically
 do a (relative) gain measurement and hence uniformity check, for many
 pixels. With a focussed laser, spot sizes of 50x50mu2 at most can be
 obtained. Larger spots should be possible with a defocussed laser and
 Marcel is sure the intensity can be increased to give a reasonable
 signal in each pixel. Hence, by illuminating e.g. 1x1mm2 and then 
 masking individual pixels in turn, it should be possible to quickly
 get a handle on the gains of hundreds of pixels, which will give a
 good indication of the gain uniformity.

 An intense Fe55 source would in principle give a physical calibration
 value from its spectrum endpoint, but this would need to be done
 separately for individual pixels to be usable for a gain unformity
 measurement and this seems unlikely, given the extra inefficiencies of
 running in bunch trains compared to the analogue readout from the test
 structure. In addition, Konstantin is the only person registered to use
 the source in PPD and he is leaving within days. There is a stronger
 Fe55 source in TD and Renato will look into whether we can use it. This
 will need an additional USB_DAQ board at RAL and Paul will see about
 getting one from Imperial.

 Birmingham could also use any other USB_DAQs which are spare. Nigel
 stated that it was not clear if the cosmics setup would be useful. With
 the geometry factors of requiring hits in the PMTs, etc, he estimates
 rates of order 0.001Hz, before including efficiency factors for the
 sensors, which is an order of magnitude lower than previously thought.
 One alternative would be to return to a test beam. DESY will not have
 beam until Sep. CERN may have beam from May, but we would have to bid
 for this and would probably get only a short time. The ISIS beam line
 would be very convenient; it has muons with momenta of order 100MeV.
 These should be enough to pass through four silicon layers as MIPs lose
 a few 100keV per layer, although multiple scattering might be an issue.
 Mike will look into using the beam and Anne-Marie can check for the
 amount of scattering expected using the simulation.

 The test structure measurements should also be redone more thoroughly,
 as discussed at previous meetings. By automating the measurements, it
 would be possible to do a large systematic scan on deep p-well and
 non-deep p-well sensors and cover the whole set of test structure
 pixels. It is important to compare these to the sensor simulations,
 which did not agree well when presented by Giulio in the previous
 meeting. To agree with both the deep p-well and non-deep p-well just by
 smearing will not be easy. This is important if we want to show that
 the charge spread is understood. An alternative (or better, additional)
 way to do this is to measure the pixel hit rate as a tightly focussed
 laser is scanned around within a single pixel. The smallest spot size
 should be 2x2mu2, so this would allow a finely stepped grid to be used,
 similar to that used on the test structures.


Beam test analysis: Paul showed some slides from Jamie Ballin, who has
 now joined CMS and so will no longer attend MAPS meetings. Mike pointed
 out that his 18mu sigma (which is close to 50mu/sqrt(12) ~ 14mu) would
 potentially reject quite a few hits in the tails of the residuals. This
 could itself lead to part of the inefficiency seen.

 Marcel said the ~15% efficiency being seen by JamieB, Owen and Tim
 Martin agree with a rough estimate he made of the number of hits seen
 in the beam test when running with tungsten plates. The showers had
 many fewer hits than would be expected and 10-15% efficiency could
 account for this.

 Nigel stated that Tim Martin has found that for his events with two
 spatially-correlated clusters (see his slides from the last meeting),
 there is a 68% chance of having two coincident hits in the PMTs.


Conferences: Marcel had had to leave but he had updated the usual MAPS
 conference web page with most of the bigger conferences for 2008.
 Renato thinks he should also add the 8th International Conference on
 Position Sensitive Detectors (PSD8), which is in Glasgow from 1-5 Sep.

 There are three conferences with abstract deadlines in Apr (CALOR08,
 ICHEP and TWEPP) and then another in early May (IEEE). Of these, the
 most important for calorimeters is probably CALOR08, and for the sensor
 designs, it is IEEE and then TWEPP. We should concentrate on these 
 although Mike pointed out that we should have significant results to
 show before applying for too many talks.

 For the first of the more important conferences, CALOR08, then Paul and
 Anne-Marie are potentially available to give a talk.


Next meeting: Tentatively scheduled for Wed 23 Apr, with Tue 22 Apr as a
 fallback if Marcel cannot make the first date. Paul will not be able to
 attend but will try to call in if he can.