[evla-sw-discuss] VCI schema changes

[Ken Sowinski]

See the added test below.

The next step is to make a list of all the changes to CM that
we can make after the schema update.  I still worry that we
might break something so a careful look at these schema changes
with usage of m2s. executor and CM in mind would not hurt.

Ken


On Wed, 20 Feb 2019, Ken Sowinski wrote:

> Below are proposed changes to support more flexible control of the
> summed array AGC circuitry and the mixer in the filter chip.   I
> intend to make these change to the VCI scehma in the next week or
> so, so speak now if you have any reservations or suggestions.
>
> Ken
>
>
> This is a proposal for what summed array AGC control for VDIF summed
> array output should look like at the VCI level.  The cc element contains
> the same attributes, they behave identically and will be ignored here.
>
> The current definition of the vdif element includes three attributes related
> to AGC:
>
>    - requantGain    int     0 < requantGain < 255
>      To explicitly set the requantizer gain.  Not very useful without
>      control of the "window" setting as well.
>
>    - agcRms         float
>      Desired setpoint for summed array output power RMS.
>
>    - agcEnabled     boolean
>      Chooses between setting requantizer gain explicitly (false), or
>      servoing the RMS to the desired setpoint (true).  The default
>      is true.
>
> These are all optional with defaults enforced in CM rather than XML.
> I propose to deprecate agcEnabled, and to add optional attributes with
> default values enforced by CM.
>
>    - agcMode        enumeration
>      To control how the AGC behaves.
>
>    - window         int
>      Which may be used in the future to specify the "window" setting.
>      Currently there is no direct control over this hardware setting.

I also would like to add:

   agcDelay int

which will specify the delay in seconds that should elapse between
thge setting of requantizer gains and the associated re-setting if
the VDIF or CC summaed array AGC.  This should be an optional
attribute with a default falue of "3" enforced by the schema.
Currently this is expressed by asking CM to subverting the
"integTime" attribute of the summedArray element.

> agcMode may take on several values:
>
>    - dontChange   Do nothing, nothing is sent to the BL boards.
>    - setRq        Equivalant to agcEnabled=false.
>    - servoRms     Equivalent to agcEnabled=true.
>    - setRms       Equivalent to servoRms which freezes when stabilized.
>    - table        Calculate RQ gain from window setting chosen by hardware
>                   and number of antenna included in the sum.  This is not
>                   yet implmented in the BL board software.
>
> If agcMode is not present, agcEnabled is transformed to "setRq" or 
> "servoRms".
> agcMode may be an enumeration or string.  I prefer the simplicity and
> extendability of a string.  agcMode, like agcEnabled, is modifiable.
>
>
> Usage of the mixer in the filter FPGA of the station board is controlled
> by two attributes of the subband element.
>
>    - useMixer             "yes" to use the mixer to center a subband at
>                           an arbitrary non-grid frequency.
>
>    - mixerPhaseErrorCorr  Always ignored by CM, usage controlled internally
>                           by the CMIB.
>
> Both are optional with a default of "no" enforced by the XML.  I propose
> to deprecate both these attributes, enforce the defaults in CM rather than
> the XML, and add two new attributes to describe how to use the mixer.
>
>    - centerFreqInSubband       "yes" to use the mixer to center a subband
> 			                                      at an arbitrary
> 			        non-grid frequency.  This
> 			        is the same as the old "useMixer"
>
>    - fringeRotateInFilterChip  Do fringe rotation (and LO offset removal)
> 			                                      in the filter
> 			        chip
> 			        Zeros will be sent to
> 			        the correlator chip and RXP summing
> 			        circuitry.
>
> Both will have default value "no".  The new attributes will be ignored
> unless neither of the old attributes is present, so that old scripts
> will always work.  The two attributes may be requested independently or
> in combination.
>
> Both these changes are designed to allow old scripts to still work without
> change.  In both cases if an old attribute is present any new ones will be
> ignored.
>
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/evla-sw-discuss
>