Reports until 16:45, Sunday 27 August 2017
H1 CDS (DAQ, ISC, SUS)
david.barker@LIGO.ORG - posted 16:45, Sunday 27 August 2017 - last comment - 20:27, Sunday 27 August 2017(38388)
h1lsc, h1omc, h1susitmx and h1susitmy model changes, DAQ restart

WP7120 Sheila and Dave:

The h1lsc model was modified to add two new SHMEM IPC senders. The h1omc model was modified to add two new SHMEM receivers (for the new h1lsc channels) and two new PCIE (Dolphin) sender channels. Each of the SUS ITM models were modified to add one new PCIE receiver channels to receive from h1omc.

The models were compiled and installed, the modifications to H1.ipc is shown below.

The models were restarted in the order: h1lsc, h1omc, h1susitmx, h1susitmy.

Because IPC receivers had been added, their slow channels were added to the DAQ INI file, requiring a DAQ restart.

After the DAQ was restarted, I discovered two things:

1. I forgot to check the write status of h1tw1 before restarting. I have checked that h1tw1 was writing raw minute trends every 5 minutes.

2. The EDCU did not come back to a GREEN status. The reason is that a new Guardian node ALS_DIFF_ETMY_ESD had been created (and added to the Guardian DAQ INI file) but is not being ran. We'll fix this tomorrow, for now the EDCU is RED, and it is missing these 19 guardian channels. The disconnected channel list is shown below.

ALS_DIFF_ETMY_ESD Guardian DAQ channels (disconnected):

H1:GRD-ALS_DIFF_ETMY_ESD_VERSION

H1:GRD-ALS_DIFF_ETMY_ESD_EZCA

H1:GRD-ALS_DIFF_ETMY_ESD_OP

H1:GRD-ALS_DIFF_ETMY_ESD_MODE

H1:GRD-ALS_DIFF_ETMY_ESD_STATUS

H1:GRD-ALS_DIFF_ETMY_ESD_WORKER

H1:GRD-ALS_DIFF_ETMY_ESD_LOAD_STATUS

H1:GRD-ALS_DIFF_ETMY_ESD_ERROR

H1:GRD-ALS_DIFF_ETMY_ESD_CONNECT

H1:GRD-ALS_DIFF_ETMY_ESD_EXECTIME

H1:GRD-ALS_DIFF_ETMY_ESD_STALLED

H1:GRD-ALS_DIFF_ETMY_ESD_NOTIFICATION

H1:GRD-ALS_DIFF_ETMY_ESD_NOMINAL_N

H1:GRD-ALS_DIFF_ETMY_ESD_REQUEST_N

H1:GRD-ALS_DIFF_ETMY_ESD_STATE_N

H1:GRD-ALS_DIFF_ETMY_ESD_TARGET_N

H1:GRD-ALS_DIFF_ETMY_ESD_OK

H1:GRD-ALS_DIFF_ETMY_ESD_ARCHIVE_ID

H1:GRD-ALS_DIFF_ETMY_ESD_TIME_UP

H1.ipc additional channels:

[H1:LSC-OMC_ITMX]

ipcType=SHMEM

ipcRate=16384

ipcHost=h1lsc0

ipcModel=h1lsc

ipcNum=59

desc=Automatically generated by IPCx.pm on 2017_Aug_27_13:36:33

 

[H1:LSC-OMC_ITMY]

ipcType=SHMEM

ipcRate=16384

ipcHost=h1lsc0

ipcModel=h1lsc

ipcNum=60

desc=Automatically generated by IPCx.pm on 2017_Aug_27_13:36:33

 

[H1:OMC-ITMX_LOCK_L]

ipcType=PCIE

ipcRate=16384

ipcHost=h1lsc0

ipcModel=h1omc

ipcNum=385

desc=Automatically generated by IPCx.pm on 2017_Aug_27_16:09:43

 

[H1:OMC-ITMY_LOCK_L]

ipcType=PCIE

ipcRate=16384

ipcHost=h1lsc0

ipcModel=h1omc

ipcNum=386

desc=Automatically generated by IPCx.pm on 2017_Aug_27_16:09:43

 

Comments related to this report
david.barker@LIGO.ORG - 16:47, Sunday 27 August 2017 (38389)

Here is the start log for this afternoon's changes:

2017_08_27 16:16 h1lsc

2017_08_27 16:18 h1omc

2017_08_27 16:18 h1susitmx

2017_08_27 16:20 h1susitmy

2017_08_27 16:23 h1broadcast0

2017_08_27 16:23 h1dc0

2017_08_27 16:23 h1fw0

2017_08_27 16:23 h1fw1

2017_08_27 16:23 h1fw2

2017_08_27 16:23 h1nds0

2017_08_27 16:23 h1nds1

2017_08_27 16:23 h1tw1

 

sheila.dwyer@LIGO.ORG - 20:27, Sunday 27 August 2017 (38391)

These model restarts were intended to allow us to send the DARM signal to the ITMs.  Previously, the LSC model had PCIE IPCs to the ITMs, while for the ETMs the LSC had shared  memory IPCs to send the signals to the OMC model, where they are summed with the DARM signals and send to the end stations using RFM IPC.  

Today Dave and I modified the models so that the ITM signals would be routed in a way more similar to the ETMs, so that the LSC has PCIE SHM IPCs to send the signals to the OMC model, then PCI IPCs to send the signals to the ITMs.  

We were able to relock fine after this, but the LSC feedforward which is routed through the new IPCs is not well tuned.  For MICH,  we used to operate with a filter gain of -15.9, today I got some decent MICH subtraction (not well tuned) with a gain of +600.  For SRCL we started to get a small amount of subtraction with a gain of around 22, while our nominal gain was -1.  

I don't understand why this is happening, but will leave the IFO to Thomas Vo for some Hartman tests.  Hopefully we look at what happened early in the morning.