Reports until 12:54, Monday 04 August 2014
H1 ISC (CDS, DAQ)
david.barker@LIGO.ORG - posted 12:54, Monday 04 August 2014 - last comment - 14:23, Monday 04 August 2014(13190)
new LSC model installed, DAQ restarted

Kiwamu and Dave. Installed new LSC model, then restarted the DAQ. We had an NFS caching issue, even giving a period of at least 15 seconds between the  make-install and the restart, the model had cached the old INI file and we did not get the expected DAQ status of 0x2000. After verifying that the INI had indeed changed, we restarted the model once more and the new INI file was picked up by the model. 

Comments related to this report
kiwamu.izumi@LIGO.ORG - 14:15, Monday 04 August 2014 (13192)
Link

I have edited both /h1/models/h1lsc.mdl and /common/models/lsc.mdl. The motivation of this update is for us to keep up with Livingston's model.

Background:

Currently, the Livingston LSC model is completely disconnected from the common libraries and therefore it can be regarded as a prototype at this point. On the other hand, our Hanford model is more like a production model and therefore we use and maintain the common LSC libraries on the basis of the LLO prototype.

The modifications:

  • I added several signal processing blocks for the transmission signals (i.e. TRX(Y) and TR_X(Y)_QPD_B_SUM).
  • I did not add the TrDARM process path as this turned out to be not so useful in Livingston. In fact they are not using this.
  • I expanded input matrix (i.e. PD_DOF_MTRX) from 34 inputs to 37 inputs. And I newly appended TrCARM, TrREFL9 and REFL_A_LF_NORM to the input matrix.
  • I deleted the most of the auto-locker blocks which had been introduced historically because Livingston did not have a guardian when locking DRMI. Though, I left the OMC_OFFSET block because the purpose of it was not clear to me. I will confirm the (un)necessity.
  • Added the MICH and SRCL feedforward paths. This required expansion of the output matrix such that it takes another two channels. The MEDM screen needs to be fixed.
  • Added two new inputs for the LSC block. They are reserved for the TRX(Y)_QPD_B_SUM signals.
  • I put two new RFM receivers in h1lsc.mdl in order to grab the TRX(Y)_QPD_B_SUM data. Note that I did not have to add an RFM sender because they were already in the models (i.e. h1iscex and h1iscey).

Discrepancies between the sites (as far as I know):

  • H1 uses a ramping matrix for the input matrix, but L1 does not. Not clear which is better.
  • L1 uses a ramping matrix for the output matrix, but H1 does not. Not celar which is better.
  • L1 has the auto-locker blocks, but H1 does not. L1 should get rid of these blocks if possible.
  • L1 has the TrDARM signal path, but H1 does not. L1 should get rid of this path. This requires a change in the input matrix as well.
  • H1 has a trigger and filters for REFL_BIAS, but L1 does not. Probably H1 should get rid of these blocks, or even we don't need REFL_BIAS at all.
    • I want to check this path out during the future locking trials at Hanford. If this is really useless signal, we will get rid of them.
kiwamu.izumi@LIGO.ORG - 14:23, Monday 04 August 2014 (13194)
Link

Also, I checked the connection of the new RFM recievers by sending some values from the end station over to the LSC model. And I confirmd that they are working fine. Both H1 and Common models are now checked in SVN.