Technical Design Report Super Fragment Separator
Technical Design Report Super Fragment Separator
Technical Design Report Super Fragment Separator
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
DRAFT<br />
synchronization to the BuTiS clock with a defined phase. Different types of FEE will also provide<br />
their local clocks. One may consider the following scenarios:<br />
a) Self-made timestamps, where a specification of the latency of the front-end trigger with<br />
respect to the timestamps is needed in order to be able to synchronize the particular data to<br />
the overall system. The latency can be automatically determined e.g. by using generated<br />
synchronization events from the host DAQ system to the particular timestamp generating<br />
system. Another question is how to couple local systems to BuTiS. One BuTiS receiver per<br />
local timestamping system might be too expensive it is a subject to further R&D whether an<br />
additional (experiment-wide) time distribution layer is desirable.<br />
b) For the standardized NUSTAR DAQ one should consider to build a novel “Titris II”<br />
module in different form factors, to be used in VME crates or attached to the FEE, which<br />
can be coupled to the BuTiS system.<br />
We can mainly base our studies on the already existing techniques, use of time-stamping followed<br />
by software filters and triggers which were developed for GREAT. Within GREAT these items are<br />
handled by the Metronome and its interface to the ADC cards. Within the next years we need to<br />
specify latencies for trigger decisions depending on the physics of the trigger, taking into account<br />
the limited pre-processing buffer depths, and their associated acceptance window. Here dedicated<br />
input from the different experiments is needed to come up with a generalized and flexible scheme.<br />
2.4.A1.2 Data collection and storage<br />
Maximum data rates of 10-100 Mbytes/sec are estimated for the EXL and R³B setups. Thus the<br />
total data produced per day is in the order of 1-10 TBytes. This means, for the NUSTAR experiments<br />
we expect not more than 500TB/year of data. The requested band-width for data transfer is<br />
in accordance with the expected rates that can be transferred using standard network components.<br />
The demand to run independent NUSTAR DAQ modules that can be combined in a flexible<br />
manner poses certain requirements to the transport layer in between the experiments and to the<br />
mass storage:<br />
1. Standardized sub-event format; this is favourable in order to facilitate to event<br />
building from different data sources.<br />
2. Flexible event builder combinations; one may consider to use different parts of the<br />
data stream at different locations to perform different tasks: mass storage; online<br />
analysis; slow control feedback.<br />
3. Standardized event format; to be used for a common unpacking scheme for data<br />
analysis.<br />
The NUSTAR DAQ system will provide the necessary data collection procedures for standard<br />
electronics module (e.g. VME) and collection from FE-boards that follow the conventions for the<br />
standardized digital interfaces (e.g. GTB). The necessary R&D work will be done in parallel to the<br />
various electronic developments in the NUSTAR community in close collaboration with the respective<br />
groups, co-ordinated by the DAQ responsible for the particular experiments.<br />
158