|
Minutes of the working group on TEST of MODULES ( September 28, 2000) |
|
Present:
|
Agenda of the meeting :
|
|
Meeting
discussion:
Marco remembered that the working group has a very tight deadline of the EDR that will be on the 15-th of November. The WG has to prepare a precise scheme of the testing scheme and QA, with strategy and flow of components during the production. Driven by this deadline, the WG has to converge very rapidly on what tests to do, how to do them and define the set-ups for the testing. M.Meschini has been requested (by G. Rolandi) to
make a presentation during next TK week (Oct. 23-27) at TK general meeting
concerning the proposed scheme for module testing etc. In the same week the Tracker Institution
Board will take the decision if adopt the proposal or not.
Investigations on the laser and backplane pulsing tests were made by the Torino group, using a full VME based DAQ system with a RIO processor and the TTC system for trigger. Both tests are meant to determine channels that do not respond to a physical signal, as broken or unbonded lines. It is clear that a laser test can do this job very easily, but it is less obvious that the test can be done without big complications in the hardware and in a short time to be compatible with a fast test. In Torino has been used a very compact VME card that houses a 990 nm infrared laser and its driver. The card has been developed by G. Rizzi in Pisa and tested deeply in Torino, it has 2 outputs for 2 laser with 17 ns and 25 ns laser pulse; the latency of the laser from an input trigger can be programmed by VME and more pulses can be generated one after the other at 25 ns distance. The test was performed in a rough way. With the laser signal transmitted with a 100 micron diameter multimodal optical fiber, and placed naked at 0.5 mm from the surface the beam obtained was of about 200 micron diameter. In order to have a fast test the beam was intentionally left very large, leaving a distance of several mm from the silicon surface: a laser cluster of about 30 strip of 60 micron pitch was obtained. The fiber was moved with a joystick up and down along the transverse dimension of the silicon while data were collected. For every event only channels with signal above a certain threshold were recorded and from an occupancy plot bad channels were detected. The test took about 2 minutes. The backplane pulsing test was attempted for the first time with the APV6 chip. Results were shown and display that the test is able to detect unconnected and shorted channels. It is anyway recommended to use both this test and the laser to identify and study bad channels. The VME laser board can be produced and distributed to other interested groups: a cost was not evaluated precisely but it should be of about 2000 CHF, laser included. It is also under way the design of a VME board to
perform backplane pulsing in an easy way.
Laurent presented the status of the Lyon system. It has been successfully tried to use a PC/VME interface that enables to see the VME boards as a direct extension of the PC CPU memory. The interface is efficient enough to allow up to 100 Hz events/sec data taking. In this way the system is compatible with a full VME solution where if many FEDs are needed they can be used. Still, a full PC solution can be adopted for smaller systems where only few FEDs are needed. In any case is foreseen to maintain all the slow control on the PC directly. The proposal of a new sequencer and trigger card was presented. It can communicate with the CCU board and the FEC, and therefore is compatible with the final CMS electronics. The first version of the new card will be PCI, but later on also a VME board will be developed. The project is on the way, and is planned to be ready for beginning spring 2001. There were several comments to Laurent
presentation, and certainly the time delivery of the system is an essential
issue. Marchioro suggested to have the possibility in the system to use both
the new sequencer OR the TTC system, that already several institutes have and
are using. Verdini suggested to add a skewed clock to the sequencer to have
an important degree of freedom in the tuning of the timing.
A full report on the strategy used at FNAL for the testing of CDF modules was presented. Hybrids were fully tested outside FNAL and burned. At FNAL only visual inspection and functionality tests were performed. Sensors were visually inspected; half-ladders mechanical structures were also visually inspected. The test was made at each stage of production, after every bonding. Readout test and burn-in procedure was made at each level: module, half-ladder and full ladder. Burn-in procedure has showed that 1 day was normally enough. Lesson learned: 25% spare sensors and hybrids were not enough to avoid fixing (but complicated modules were adopted) therefore it was necessary to make repairs. No online qualification was possible, always physicists had to understand and qualify modules after testing distinguishing grade (high, medium, low) of modules. It is essential to have availability of trained
personnel and to have similar skills in all centers. It is also very important
to have a prototype/pre-production stage before entering full
production.
Comments from CMS USA groups to the module testing were presented. See Alice Bean proposal in the Module Testing Working Group Web page. If hybrids do not pass the tests in USA, they will be sent back to Europe. Sensors will be visually inspected, and a fraction probed in KSU. It is proposed to make only fast test that do NOT
include a laser test but DO include a backplane pulsing test. It is also
proposed NOT to make a thermal cycle.
Marco presented the proposal of the testing scheme for the module production. A copy of the presentation can be found in the Module Testing Working Group Web page, http://hep.fi.infn.it/CMS/moduletest/mod_00sep28.pdf. See there for all details. The full production chain includes: hybrids centers, Gantry centers, Bonding centers, Module burn-in centers, Rods/Petals/Disks centers and in the end Integration centers. Hardware: 2 DAQ set-ups are needed, a) for hybrid and Gantry centers; b) for bonding, burn-in and integration centers. Candidates are: for a) the Aachen set-up and for b) the Lyon set-up. Estimates of needed set-up are: 10-15 of type a) and 16-20 of type b). Both Aachen and Lyon systems will be supported by their respective institutions during CMS tracker production. There was a lively discussion on many points of the presentation. There was a general agreement on: ‘fast test’ procedure after each production step; need of thermal cycling (at least at low temperatures) and readout for large structures (petals and rods); thermal cycling (without readout) for modules right after bonding in addition to the electronics tests (to be checked with Bonding W.G.); red/infrared light test on modules or larger structures. Minutes taken from : Lino Demaria |
|
|