Hall D MOU addendum - Indiana University
ADDENDUM TO MEMORANDUM OF UNDERSTANDING BETWEEN
INDIANA UNIVERSITY
AND
THOMAS JEFFERSON NATIONAL ACCELERATOR FACILITY
March 22, 2002
The parties agree to continue the Research and Development effort for Hall D. This particular effort is directed toward prototyping and developing techniques to manufacture the highly specialized electronics needed to meet the high rate requirements of Hall D. We recognize that this effort is project specific and as such cannot start before DOE approval of Critical Decision 0 (CD-0) for the Jlab 12 Gev Upgrade.
A. The tasks to be performed, until completion of this R&D phase that covers FY2002 and FY2003, are the following:
1. Flash ADC prototyping: The design of a Xilinx gate array which performs the deadtimeless data buffering and feature extraction functions described in the Hall D Design Report was completed in 2001. A printed circuit for evaluating this gate array along with a suitable digitizer has been designed, and will be fabricated and tested. This was one of the R&D priorities recommended in the Cassel review. Important measurements include signal-to-noise ratio, speed, and clock feedthrough.
This first protoype is a single channel in PCI format. Subsequent prototypes will investigate issues of clock fanout within a module and between modules and crates. All channels in the final system must be synchonized so that “sample 0” refers to the same time in every channel. For the Forward and Barrel Calorimeters (FCAL & BCAL), a continuous energy sum of all channels must be calculated for use in the level 1 trigger.
Another important issue to be investigated is the optimal packaging for the Flash ADC system. VME and PCI are possible candidates. However, neither bus system provides for the required energy sum trigger. The physical packaging must accomodate the energy sum, as well as distribute a 250 MHz clock, and global synchronization signals. Processors of some sort must be included in the overall system to compress and format the data. [FY2002-FY2003].
2. FCAL PMT power supply prototyping: The Forward Calorimeter (FCAL) requires compact, low power computer controlled high voltage power supplies. The design of the FCAL bases is based on two generations of experience (BNL E852 and JLab RadPhi) in building or using such devices[1]. [FY2002-FY2003].
3. BCAL PMT power supply prototyping: The very high magnetic field in the location of the SciFi ends of the Barrel Calorimeter (BCAL) dictates either the use of field resistant PMTs with good response up to 2T and/or the use of fiber optic (FO) light guides to move the PMTs as far away as required to reduce the field down to working levels. In either case, it is likely that custom electronic bases will be required. The most promising candidates for PMTs are the new technology of hybrid PMTs (HPMTs) which combine high resolution with magnetic field tolerance up to 2T. The operating requirements require custom electronics for the voltage and pre-amp circuits. Commercial units are available but are expensive and savings must be implemented considering the large number of units required for the BCAL. The capability to manufacture these bases can mean the difference between affordability and cheaper solutions with degraded performance. [FY2002-FY2003]
4. Electronics assembly and testing procedures prototyping: The high rate, continuous photon beam proposed for Hall D requires front end electronics with no deadtime, programmable feature extraction, and large buffers. There are no commercial manufacturers of suitable electronics in the U.S. at this time. Indiana University has successfully designed and built a large fraction of the electronics for experiments at Fermilab, Brookhaven, and JLab. Hall D is larger and more complex than past experiments and will require the development of new techniques. High reliability is crucial to the success of Hall D. We plan to begin long term tests of assemblies as soon as they are produced giving rapid identification of problems. Waiting until a large batch of electronics has been commercially assembled risks learning about the problems after it’s too late. [FY2002-FY2003]
B. Completion of these tasks along with future R&D and construction will require:
In addition to developing prototype electronics systems for Hall D, Indiana University is committed to producing the final electronics for the experiment as well as its maintenance over the anticipated lifetime of the detector.
Surface mount technology is essential to achieve the required density and functionality. Simple scaling of the techniques used for the construction of the RadPhi bases and E852 ADCs is not feasible. The 750 bases built for RadPhi required one person day per base using skilled technicians. The labor cost would be prohibitive should this method be used again. Finding enough skilled workers for a limited period and managing them is unrealistic. Automated assembly of all or at least a large fraction of the electronics is required. Repair of electronics constructed using high density surface mount technology would require an extremely skilled technician standing by during the entire detector lifetime unless automated component replacement capability is also available.
We intend to purchase an automated assembly and rework system with capabilities ideally matched to our requirements. Current quotes place the price of this system at $160,000 including an educational discount of $40,000.
The Beamworks Spark 400 is a unique device which is well matched to task of producing custom electronic assemblies for the Hall D experiment. All phases of surface mount assembly are addressed: solder paste dispensing, component placement, laser soldering, and inspection. Other, competing machines only perform one of these operations. Matching these capabilities with separate machines would be more expensive. Larger, more expensive assembly machines are faster, but less flexible. Smaller and less expensive machines are not as automated, requiring skilled operators.Laser soldering offers several advantages over infrared or convection reflow. The lasers can be focused on a small area heating only the component leads and pads. Moisture sensitive components don't require prebaking. Fine-pitch and ball-grid components can be removed by the device, facilitating repairs and rework. Components can be installed selectively, allowing partial assemblies to be tested.
Experience has shown that the construction of Cockroft-Walton bases is not a simple application of "conventional" electronics assembly techniques. In process testing is required at each stage of the assembly process to avoid consuming large quantities of components to make bases that will ultimately fail in use. In house assembly, in stages and small batches, will allow this testing to be performed. Local expertise in design, construction and testing exists and can be profitably exploited to assure high quality standards are met. Staged, small quantity assembly done by commercial concerns is much more expensive than "batch" assembly because of minimum order and setup fee considerations.
Another consideration made apparent by previous experience is that component failure after long term "burn-in" is quite common. This problem could be addressed by specifying pre-tested components but would be prohibitively expensive. A more attractive solution is to exploit the rework capabilities of the Spark 400 system to automate component replacement.
Approximately 20,000 channels of FADC will be required by the experiment. Large scale prototyping becomes a realistic option if automated assembly capability exists in the collaboration. Demonstrations of suitability and manufacturability are possible. Manufacturing of the entire batch of required FADC channels is our preferred scenario and is possible using the proposed equipment.
Other manufacturing scenarios are also made possible with this technology. The rework and laser soldering capabilities of the proposed system allow the possibility that several different versions of FADC (suitable for different detector subsystems) could be built. In this scenario, "generic" FADC components (for example, communication drivers) would be placed on the entire order of PC boards. Detector specific components (for example, front end amplifiers or digitizers) would be placed in subsets of the order using the laser soldering capabilities specific to the proposed system.
Indiana University has nearly 10 years of experience with surface-mount electronics construction. Paul Smith and Eric Scott toured facilities at other universities and a commercial assembly contractor. They visited the BeamWorks factory in Portland, Oregon to see the system in action. Extensive searches in trade publications have failed to turn up a device with equivalent capabilities at lower cost.
C. The schedule is:
In FY2002: Complete first FADC prototype and begin testing. JLab will purchase the BeamWorks system and drop ship to Indiana University. Indiana University will install and get system running. Begin design of FCAL PMT bases.
In FY2003: Construct 100 FCAL PMT bases using BeamWorks system and test. Design next generation FADC prototype. Begin design of BCAL PMT bases.
D. The resources are:
Manpower:
Four faculty members: Dzierba, Heinz, Szcepaniak, and one to be hired.
One research scientist: Teige.
One postdoctoral fellow: Steffen.
Two engineers: Smith and Scott.
Two to three undergraduate students per year.
Indiana University Physics Department machinists.
Equipment:
Computers, tools, instrumentation, and test equipment at IU.
Requested materials (Section E).
Facilities:
IU Physics Department electronics and machine shops.
IU will provide a room for the Spark 400 with appropriate utilities.
E. Material requirements are:
Materials requested from JLab/DOE under FY2002 and FY2003 are:
a) Electronic components and printed circuit boards:Digitizer ICs / $1K
Gate Array ICs / $3K
Printed Circuits / $4K
Other active components / $1K
Passive components / $1K
b) BeamWorks Spark 400. / $160K
Prof. Alex Dzierba / Date
IU Contact Person & Hall D spokesman
Indiana University
Prof. James Musser / Date
Chair, Physics Department
Indiana University
Dr. Lawrence Cardman / Date
Associate Director for Physics
Jefferson Lab
Dr. Elton Smith / Date
Hall D Group Leader
Jefferson Lab
[1] Nucl.Instrum.Meth.A414:466-476,1998