Overview
There are currently two main test systems in SR1:
- H8 test box setup. Host machine srsctdaq1, SBC ctatsct01 and LTP lnxpool17
- Sector test setup. Host machine pcphsctr05, SBC lnxpool10.
The SR1 server is pcatl-idsrv01, which is used to host the SctRodDaq installations and the tdaq software.
For the combined run we are using a single ROS machine (pcphsctr01) with 3 FILARs. The second ROS machine (pcphsctr02) has no FILARs and is used as the second temrinal in the control room.
We have two SuperMicro? servers, pcphsctr03 and pcphsctr04. Each have a 100GB /work directory, nfs mountable to the other systems as required.
Cooling Issues
The H8 box and endcap sector is water chilled, ie can be switched on at short notice. The barrel sector is C3F8 chilled. Only trained cooling operators are permitted to operate the cooling. The SCT cooling operators are:- Dave
- Jorgen
- Ewa
- Forest
- Lewis
SBCs
nfs mounts and vme setup is performed in the post boot script in /clients/lnxpool10 etc subdirectory on the SBC, on the SR1 server pcatl-idsrv01. ssh to the SBC to edit this.- lnxpool09 (SR1) used by pcatsct06 endcap area
- lnxpool10 (SR1) used by pcphsctr05 SLC3 endcap area
- ctatsct01 (SR1) used by srsctdaq1 SLC3 racks
- lnxpool17 (SR1) LTP crate racks
Software
- Release Candidate 4.3
- From srsctdaq1, setup with /daqsoft/sct/setup_RC43.sh
- Tip of TDAQ14_BRANCH
- from srsctdaq1, setup with /daqsoft/sct/setup_tdaq14_branch.sh
- from pcphsctr05, setup with /daqosft/sct/setup_pcphsctr05.sh
There are 3 100GB scratch disks at
- /work/srsctdaq1
- Used for the H8 testbox system running on host srsctdaq1
- /work/pcphsctr03
- Used for the barrel sector system running pon pcphsctr05
- /work/pcphsctr04
- Not currently used
Example - To start daq with the H8 test box on srsctdaq1:
cd /daqsoft/sct
- source setup_RC43.sh to run with Release Candidate 4.3
- source setup_tdaq14_branch.sh to run with TDAQ14_BRANCH
We currently have 12 ROD/BOCs in the main daq crate in racks. The ROD in slot 6 has an Slink connected to the ROS machine pcphsctr01. The ROD in slots 10/11 is used for H8 box calibration tests.
Eg, for DSP code testing:
cd /daqsoft/sct source setup_RC43.sh ./start To flash MDSP: ssh ctatsct01 cd /daqsoft/sct source setup_RC43.sh cd RodDaq/RodUtils
Combined Running
How to start a combined run, using stabndalone SCT partition. Partition is part_SCT.xml in /daqsoft/combined/partitions.This setup is for triggering with both (TRT and SCT) LTPs, with SCT LTP as a slave (assumes TRT pulse generator is running), and includes the SCT ROS and event builder segments.
- on srsctdaq1, login as 'sr1daq'
- cd ~
- source setup.SCT.sh
- RUNDAQ
Tip: if SctRodDaq applications do not start, check you have the correct tag, eg, in setup script 'export CMTCONFIG=i686-slc3-gcc323-dbg' and in SctSoftware.data.xml: <rel name="Tags" num="1"> "Tag" "i686-slc3-gcc323-dbg" </rel>
In order to run without L1ID errors, triggers must be vetoed for the duration of the ORBIT signal (as generated by the TTCvi).
The following refer to old procedures, no longer necessary:
SI.getSctApi().changeRunMode(Sct_SctApi.RunType.PHYSICS_RUN_TYPE) SI.getSctApi().timWriteRegister(0,0,(short)0x18,(short)0x62f) To avoid L1ID errors, you need to send a soft reset that goes through the TIM, the best way to do this is from the LTP. If this isn't possible, it can be sent directly by the TIM: SI.getSctApi().timWriteRegister(0,0,(short)2,(short)4) SI.getSctApi().timWriteRegister(0,0,(short)2,(short)0) SI.getSctApi().timWriteRegister(0,0,(short)2,(short)8) SI.getSctApi().timWriteRegister(0,0,(short)2,(short)0) To get rid of BC errors, write the 0xfe to the top byte of the MDSP register: 0x00402210 so 0x000000e0 should become 0xfb0000e0. A similar effect is achieved by doing SI.getSctApi().timWriteRegister(0,0,(short)0x14,(short)0x5000) followed by a L1SR. But this results in a different BCID in the event byte stream. The correct thing to do is the following: write the 0xfb0000e0 to the MDSP register: 0x00402210 AND do <pre> SI.getSctApi().timWriteRegister(0,0,(short)0x14,(short)0x0000) followed by a L1SR.
In current CVS the above should be done automatically by changeRunMode(). </pre>
The data is written to /home/DATA/sct
There are several ways to see what is in this file:
- Dump the file in octal format by doing
od -t x4 daq_SCTEB__0001009_file01.writing | less
- Use the tdaq event analysis tools:
StorageReader $filename dump-eformat-in-data filename -o outputfile -f event_number(above utility in tdaq-common-01-01-00)
- Use example executable from TDAQ:
SimpleTBA $filename
eventStreamDecoder $filename
More on eventStreamDecoder
Here is a link for on-ROD histogramming: onRODhist
Multiple FILARs
more /proc/filar lists status of all FILARs in ROS filar_slink_dst -o channel number (to run utility for channel number) where Card0 1 2 3 4 (physical middle) Card1 5 6 7 8 (physical down) Card2 9 10 11 12 (physical up) Software map in oks database: Card0 0 1 2 3 Card1 4 5 6 7 Card2 8 9 10 11 These are physically connected to BOCs in the following rod crate slots: Card0 10 11 14 15 (physical middle) Card1 6 7 8 9 (physical down) Card2 16 17 18 20 (phsical top) All FILAR Input Channels (0..11) should belong to the same FilarReadout Module To run a test program, over Slink channel 2 (say): on ROS, filar_slink_dst -o 2 -p on BOC, SlinkTest -s11