Starting PbGl


Igor Yushmanov made some useful notes on starting PbGl from power-off mode. His notes reside here: http://www.phenix.bnl.gov/~yushman/Zametki_Sklerotika/ and this is just a pirate copy of them.
These are in the PEH:
ECN14 (E0, North, LED pulsers, FEM LV power)
ECN21 (E1, North, LED pulsers, REF FEMs North)
ECN31 (REF FEMs power, North; what about about South? I guess they are powered from supplys in ECN14)
ECB (It consists of ECB1, and ECB2, ECB2 holds our ARCnet hubs)
ECS13 (E0, South, 55 V supply, LED pulsers)
ECS22 (E1, South, LED pulsers, REF FEMs South)
First check the breakers in ECN and ECS power panels (close to ECN14 and ECS13). Then switch on all the fans and relevant crates in the racks. Then use Rack Monitoring and Control to switch the racks on. If something goes wrong - check fuses at the bottom of FEM racks or on the right side of REF FEMs racks; these are one per FEM, 15A on red cables and 5A on blue ones Check ARCNET connectivity to FEMS on given ARCNET subnet: (Or, in other words, to all FEMs included into given granule). Using phoncs account on phoncs1 host: setuponcs (not really needed) As described in EMCal logbook (with netmap /dev/arcX) ARCNET subnets, ascribed to EMC are the following: 
/dev/arc0 - EB (PbGl)
/dev/arc2 - ET (PbSc)
/dev/arc4 - WB (PbSc)
/dev/arc5 - WT (PbSc)
See PHENIX->Internal->EMCAL->Online documentation->Collecting Data for EMCAl All FEMs on the given subnet should be visible. Things to check, when experiencing problems with arcnet:
  1. Are you running on phoncs1?
  2. Is the LV for PbGl on? (including reference FEMs in separate racks)
  3. Is the LV for ARCNET HUB on?
  4. Is arcnet chain terminated? (I've seen part of chain inaccesible when not terminated)
  5. If you see some but not all FEMs - list the ones you see (arcnet addresses), map them to physical FEM location (as written on the doors). Fem E0-0 has arcnet address 144, FEM E0-31 has arcnet address 175. Fems 176,177,178,179 are reference ones and 180-211 are E1 from E1-0 to E1-31). If they are aligned in columns (as arcnet cable goes) - check the cable after the last FEM visible on the net. John Haggerty has the tool to press RJ-11 on cables.
    FYI: Arcnet cable is plain straight (do not twist) 1-pair phone cable. Feel free to cut off second (outher) pair. Both arcnet jacks on arcnet board are equivalent, there's no "input" and "output". (If you look throughly at PCB you'll see that the only used inner pair goes straight from one jack to other; outher pair looks asymmetric)
  6. Is arcnet server process stuck? See the yellow light on the backside of an arcnet server labled /dev/arc0 in.
    If constantly on - restart arcnet server process going to the phoncs1 as phnxrc (needed?) and using strt_arcnet_server.csh This script does kill all existing servers (if any) and restart them. I don't know how to restart just one (look into the script!)
    Just for information - there are 6 server processes serving arc0 thru arc7 running on phoncs1. There is another machine (phoncs5) serving /dev/arc8-arc9. phoncs1 is located in PRR1.3 rack and can be accessed directly from KVM console in the same rack.
  7. One ARCnet subnet may host and do often host more then just one subsystem granule. In particulair PbGL is sitting on arc0, but there are also other subsystems of East Carridge attached to arc0. It has been observed when one flaky FEM of such subsystem has corrupted all data transfers on the whole subnet. That showed up either as inability to feed PbGl or even as inaccessibility of subnet when checked with netmap. The good thing is to switch off the low voltage for all neigbors (who?) and check again.
    PbGl has arcnet addresses 144-211 (144-175 E0, 176,177 E0 REFs, 178,179, E1 REFs, 180-211 E1). There are also devices having adresses
    80-87 PC1.E.N
    96-103 PC3.E.N
    48-67 DC.E.N
    112-115 TOF
    121 TOT (Not TOF!) ???
    134-135 RICH
    ARCNET address assignment (and correspondence, between ARCNET addresses (thus geometrical position) and FEM MB number) can be found in $ARCNET_DATA/gdb.dat
PbSc granule/subnet/2 sectors have 36 FEMS for data (typically having ARCNET addresses 192-227) and one FEM for monitoring (228 for W.B and E.T, 229 for W.T) PbGl has 32 FEMS for data and 2 for REF signals for each sector (144-175+176,177 for E) and 180-211+178,179 for E1) Their addresses on ARCnet are given on the web: see PHENIX->Internal->EMCAL->Online documentation->Run3 That points to Run 3 Hardware Map. Go to phnxrc@phoncsc and run feed.pl. Do a complete feed for given part of calorimeter. (fast feed just sends reset, it DOES NOT load serial data to ASICs and DSP programs to HMs and DMs

Feed will fail if some FEMs are not acessible via ARCNET. (Or if something else is wrong, e.g. CORBA stuff of wrong machine choosen)

Feed may also fail in case, e.g. one of the files to be fed is missing. The decryprion of the feed error messages can be found under /export/software/oncs/R6.0.1/online_distribution/Arcnet/arc_node.h:
SUCCESS 0
DEVICE_OPEN_FAIL 1
DEVICE_IOCTL_FAIL 2
RESPONSE_TIMEOUT 3
HEXFILE_OPEN_FAIL 4
ERROR_READING_RESPONSE 5
INVALID_LENGTH_RESPONSE 6
PUT_PCKT_FAILED 7
ILLEGAL_TRANSMISSION_LENGTH 8
INCORRECT_RESPONSE 9
BROADCAST_REPLY_ERRORS 10
BAD_HEX_CHARACTER 11
MYSQLERROR 12
NOACFFILE 13
CORBA_FAILURE 14
NODE_ERROR_START 0x80000000

In case of error 3,7 - look for dead FEMs, error 4 - software, look through all the files used by feed for your granule.

Arcnet history can be found at http://logbook.phenix.bnl.gov/arcnet
Now it's time to start data collection. Before that:

Check if the VME crate, hosting DCMs and icondev for the granule(s) (W.B,W.T,E.B,E.T) you are looking at are powered. Just the automatic switsh at the top of the rack. Feel free to switch on/off.

To see what's going on on appropriate iocondev do 
ts iocondev4c
That opens the window on terminal server, to which iocondev is connected via serial port and attach this window to that serial port. This window is NOT on iocondev itself, thus it survives iocondev reboot. You can both see iocondev output messages and tell it to do something (shutdown :) via this window. To reboot an iocondev use Ctrl+X in that window.

Go to phnxrc@phoncsc and start runcontrol. 
   rc -standalone &
a) Select granule, eg. EMC.W.T
b) Download
c) Start
Last update:
Andrey Kazantsev