CALICE MAPS Meeting, RAL, 31/08/07
==================================
Present: Jamie Ballin, Jamie Crooks, Paul Dauncey, Anne-Marie Magnan,
 Yoshi Mikami, Matt Noy,  Marcel Stanitzki, Renato Turchetta,
 Giulio Villani


Minutes: Paul


Minutes of the previous meeting: Regarding naming the foundry, Renato
 said they have not yet given permission to be publically named. They
 would like to be sure of a positive story beforehand and so we will
 probably need to have "proof of success" first.

 Matt sent three USB_DAQ-to-sensor PCB cables to Marcel and Giulio and
 they have checked that they do indeed fit into the laser darkbox.


Sensor design: JamieC showed some slides on the testing status and
 options for running the sensor; see the usual web page.

 The problem with running the monostables has been found to be due to
 one of the power connections being tied to VDDO rather than the same
 line as the other parts of the monostable. The other parts would
 normally operate at 1.8V while VDDO is the voltage for the I/O section
 of the sensor and hence was expected to be at 3.3V.

 This was not shown by the simulation as the separate parts of the
 circuit were studied independently with the power wired to each by
 hand. A check of the design back to the schematic should have revealed
 this also. It is clear that checks to prevent similar problems in the
 second fabrication round will be needed.

 The issue is now how to run the sensor for testing. Of Jamie's six
 options, then option 5a, to repair the fault on the sensors, was
 thought to be the best, assuming the price was around the expected few
 kpounds. Renato indicated that this technique was quite standard and
 widely used and so had little risk to the sensors. We would want to
 repair a significant number of sensors as the yield is not yet known;
 around 30-40 would be useful. The fallback plan was option 3, which
 required running parts of the sensor PCBs with a reduced voltage (2.1V
 rather than 3.3v).

 There are also problems with overlapping phi_1,2,3 clocks in the data
 logic. Reducing the speed by a factor of two has not cured the problem
 so Jamie suspects one clock may be being inverted. Matt will produce a
 firmware version with a much slower clock so that edge effects can be
 neglected. This issue has prevented a full check of the data logic so
 far.

 Jamie sees a linear response in the test structure analogue circuit
 but with a factor two less gain than the simulation indicates. However,
 the difference could be due to parasitic capacitance and buffering 
 between the circuit and the measuring point which is not included in
 the simulation. Jamie will try to estimate the latter. He does not yet
 have a noise measurement from the analogue output but this should be
 possible when the sensor is run in the laser system. This should be
 done when Giulio returns from TWEPP in two weeks.

 Giulio has inspected wafers under the laser microscope. The epitaxial
 layer is not visible and so its thickness cannot be confirmed by this
 method. Marcel will distribute photos taken of these tests.


Sensor simulation: Giulio is continuing with the simulations. He has
 done roughly half the points and now has a complete half sensor (albeit
 without any diodes outside the pixel so the edges are inaccurate). The
 two quadrants done so far look reasonably symmetric so the previous
 21-point approximation may still be valid. He has sent the latest data
 to Anne-Marie for the physics simulation.


Sensor testing: All three of the sensor PCBs have now had sensors
 mounted onto them. The third was a non-deep p-well sensor and this was
 found to have an error in one of the four readout columns. This should
 therefore be removed and the PCB used as a test of replacing a sensor.
 To run with the reduced power voltage (as per option 3 above) then the
 sensor PCBs need a straightforward modification to the regulator. One
 PCB has already been done and the other two should be changed when
 convenient. The PCB has been quite thoroughly tested now, including the
 test structure analogue output (coaxial connectors and buffers). As
 Valdimir is no longer around, Matt will order 20 more PCBs. Components
 for 12 have been ordered already and these will be assembled but
 without the power regulator being fitted. If the sensor repair works,
 these can be added at Imperial; if not, then a pin-compatible 2.1V
 replacement will be mounted instead. The only other change will be
 that a faster inverter will be used to allow higher speed switching.
 The PCBs will be ordered with a two-week fabrication and two-week
 assembly turnaround, which should match the timescale for repairing
 the sensors.

 Three sets of USB_DAQ adapter-to-sensor PCB cables have been made and
 two sets have been tested so far. A set takes about an hour to make
 and another set will be produced for each USB_DAQ board made. The laser
 system will need a cable for the test structure analogue coax output
 and PPD will buy this. The USB_DAQ-to-laser connection will require
 four IDC-to-BNC cables (two input and two output) which will probably
 need to be made up by hand; Matt will take care of this. Matt will also
 order the micro-USB-to-USB cables for the PC-to-USB_DAQ connection.

 Matt has ordered 15 PCBs of the USB_DAQ adapter boards and these are
 expected back within a few days. All components are in hand to make
 up 10 of these and they can be assembled at Imperial at a rate of
 around one per day.

 Matt has 13 PCBs of the USB_DAQ (V2) and 5 sets of components. These
 can be assembled at Imperial with a rate of around one every 1.5 days
 although this will be done by the same person as the adapter boards so
 they will alternate assembly of these. This would allow two pairs of
 the boards to be produced in a week. If Imperial have problems with the
 BGA components, then it would be possible to have the assembly done
 commercially but Matt would prefer it to be done in-house if possible.

 Vladimir has produced a layout for the USB_DAQ memory extension card.
 In principle this is ready to be fabricated but Matt would like another
 iteration first. Vladimir has the required design tools at Belgrade to
 allow him to work on this from there. Matt will contact him to ensure
 this is done.

 Marcel showed some slides (see first half of talk on usual web page)
 on the laser status. The handling of asynchronous processes in LabView
 is slowing down the DAQ laser control progress so this is not yet
 working. He has also got some more information on tungsten sheets and
 input from Birmingham on the mechanical structure and hence size of the
 plates is now needed.

 Paul has updated the coordinate system definition diagram, see usual
 web page, following discussion. The coordinate system now coincides
 with the usual view from above the PCB in a test system (which is the
 substrate side of the sensor) and with the row/column markings on the
 PCB itself. The x coordinate and column number are directly related,
 as are the y coordinate and the row number.


Physics simulation: Yoshi showed some slides on Mokka studies; see talk
 on usual web page. He finds the clustering improves the resolution but
 degrades the linearity. It would presumably be possible to optimise the
 clustering for linearity instead but at some cost in resolution. There
 is a large study to be done on the best way to use the information.

 Anne-Marie showed some slides on various topics. She finds the energy
 deposit in the sensor varies depending on the GEANT4 options used
 although most of the differences are well above the threshold we expect
 to use. In particular, no options show a large increase in the spectrum
 at low energies, which would give a big sensitivity to the exact
 threshold value. She also has fixed the bug reported by Marcel
 previously and the two of them will meet at Imperial in the next few
 days so as to plan out the next steps. Anne-Marie has got a beam test
 geometry for Mokka running. It is not clear if this or Fluka would be
 the better approach and she and Nigel should come up with a proposal,
 as discussed at the previous meeting.

 Marcel showed the second part of his talk on his PFA studies. He noted
 that PandoraPFA is tuned for LDC00 rather than LDC01. The former has a
 much deeper calorimeter and so is intrinsically superior for PFA. He
 has a ttbar study running but really determining the origin of the
 different resolutions seen will be hard.


Conferences: Giulio will talk at TWEPP next week and so had produced a
 draft of the talk, which he showed to the meeting. It was quite long
 given the talk time of 20mins and so suggestions of what to remove
 would be welcome. He can take comments by email over the weekend and
 the talk will be posted on the meetings web page so people can look
 through it again. The final version should be put on Marcel's talks
 web page.

 The IEEE talks (JamieC and Marcel) are both 12+3mins. We would hope to
 have some concrete results from at least the laser and analogue test
 structure for these talks.

 JamieB has produced a MAPS poster for the RAL summer school. Marcel 
 should add this to the talks web page also.

 We should give a MAPS status report at the CALICE-UK meeting in
 Cambridge on Thu 20 Sep. JamieB volunteered to give this.

 The US ILC meeting (ALCPG) on Oct 22-26 is around the time of the IEEE
 and at least Marcel can attend. We could ask for two talks, one on the
 sensor tests and the other on physics studies with MAPS. Marcel will
 contact the organisers.


Next meeting: Mon 17 Sep at 1pm in a PPD meeting room.