RCC Essential Model
Draft 0.2
16 July 1999
John Hill (hill@hep.phy.cam.ac.uk)
This is a preliminary attempt to describe the essential model of the ROD
Crate Controller. While it is certainly not complete, I hope the information
in this document is correct. Comments/corrections/omissions will be gratefully
received.
Introduction
A ROD Crate Controller (RCC) is a processor which is housed in the ROD
crate. There will be one processor per ROD crate. Its basic task is to control
the RODs in the crate, but it will also interact with the Back of Crate cards
(BOC) (indirectly via the RODs), TTC Interface Module (TIM) and also the
global Trigger/DAQ.
The details of the implementation are not well defined, but the choice
will be influenced by the recommendations of the Trigger/DAQ group.
Context
See the diagram at http://www.hep.phy.cam.ac.uk/atlas/hill/rcc_essential.ps
or .gif - this shows the context in which the RCC works. In particular:
- The RCC responds to Run Control commands from central DAQ.
- The RCC will read the appropriate database information from the central databases
when necessary.
- The RCC issues commands to the RODs (including BOCs) and the TIM as necessary.
- The RCC can read register information from the RODs (including BOCs) and the TIM
over the VME backplane.
- The RCC can read event data, histograms and error counters/flags from the RODs over
the VME backplane.
- The RCC will send summary histograms, error counters/flags and database updates to
the central DAQ.
Notes
- The RCC may be a VME-based single board computer, or an interface module plus a PC.
The choice will depend on recommendations from the Trigger/DAQ group. For planning
purposes the choice has no bearing on the Essential Model.
- The RCC must be capable of the following VME functions: block transfers, programmed
i/o, broadcasts and interrupt handling.
- The interface to the global Trigger/DAQ is likely to be ethernet or similar - the
RCC must support whatever choice is made.
- The RCC must be able to operate in standalone and global mode. It is likely that a
standardised (or close) Local DAQ will run in the RCC and that this will cater for
both modes of operation.
- The RCC must be able to handle event data from the RODs over VME at up to 1kHz. It
must be able to either analyze the data, or write it to local storage.
- The RCC must be able to read histogram data from the RODs over VME. It must be
able to process these data and generate a summary, which will then be sent to
the central DAQ.
- The RCC must be able to read and write the registers in the RODs and the TIM.
- The RCC must initialise the RODs as appropriate. In global mode, the ROD-ROB link
must be enabled, the Front Ends to be enabled are read from the configuration
database and the RCC will spy a sample of events over the VME backplane.
In standalone mode, the ROD-ROB link must be disabled and it is likely that the
information about enabling Front Ends will come from information held locally. All
events will be read over VME.
- Similar considerations apply to the initialisation of the TIM - it must be set up
correctly for either standalone or global mode of running.
- The RCC must handle Run Control commands from the central DAQ and issue the relevant
instructions to the RODs and/or the TIM. It must issue a Run Control acknowledgement
or an error to the central DAQ on successful completion or failure as appropriate.
- The RCC must be able to initiate a calibration of the Front Ends. The TIM will be
signalled to issue the appropriate command to the RODs. The RCC must read the data
back from the RODs and handle it in the correct way.
- The RCC must be able to read from and write to the central databases. The exact
details of which databases will be accessed are not yet worked out.
- The RCC must monitor the event counters and flags on the RODs, and transmit them
(or a summary) to the central DAQ.
This file is http://www.hep.phy.cam.ac.uk/atlas/hill/rcc_essential.html