|CERN -> ATLAS -> TU Vienna|
|High-Density Digital Links|
As indicated in the title of this thesis my work at CERN was dedicated to signal integrity and electromagnetic compatibility. The SI part mainly included modeling, simulating, and measuring twisted-pair cables and cable filters, and investigating possible repeater solutions. The EMC part was composed of redesigning the front-end-electronics boards in order to improve the noise performance, investigating and defining a grounding scheme for the TRT sub detector, and evaluating the compatibility with the surrounding sub detectors.
In order to develop a model of shielded-twisted-pair cables, I started at CERN using commercial electromagnetic-field simulation tools, but these tools reached their limits due to the double-helical structure and the small size of the cable. Finally, measurements in combination with spice simulations allowed the design of a proper compensation network which will be integrated into the final design.
We investigated several repeater designs including combinations of LVDS receivers or fast comparators with LVDS drivers and discrete transistor networks and concluded that it would be necessary to design a bipolar ASIC as LVDS repeater. For the intermediate system test, I designed repeater patch panels using the first solutions.
Furthermore, I explored the rest of the system in terms of signal integrity, such as the granularity of the distribution of the control signals.
Finally, we established a completely working transmission link from a sector prototype over the front-end electronics, twisted-pair cables, and repeater patch panels to the back-end electronics in the laboratory.
During the second half of my work I concentrated on electromagnetic compatibility. We examined different grounding schemes using the existing sector prototype and electronics. This, together with the estimation of possible noise sources of the TRT and neighboring sub detectors, led to the above mentioned approaches for controlled currents and lowest possible impedance inside the system.
I then turned my principal attention to the redesign and simulation of the front-end boards. Using the results of my research allowed us to decrease the noise level of the front-end electronics close to the level of the bare analogue chip.
Measurements at BNL showed that the electronics and the services of the TRT will not influence the LAr calorimeter and its front-end electronics - the most susceptible victim of the TRT.
However, only a bigger and more realistic prototype will show the response of the full system. Small systematic effects could pile up in a complex system like the TRT with thousands of analogue and digital chips.
|February 9, 2000 - Martin Mandl||Copyright © CERN 2000|