[evla-sw-discuss] Simple subarrays: a modest proposal

[Bryan Butler]

Michael Rupen wrote, On 4/2/14 10:53 :
> Hi Sonja --
>
>> Hi Michael,
>> the quick test would be to modify Executor to always add the mappingTime to
>> the Activation Trigger.
>> Executor could calculate  mappingTime=activationTime-x, where x is a
>> constant,  long enough interval to transmit two or three subarrays (and
>> associated Activation Triggers).
>> This would require code modification, but should be rather straight forward.
>
> We'll have to chat locally about the easiest way to check this. Ken/Bryan
> (or Barry) are the guys who would actually do this.

it'd be a barry thing.


>> By the way, I know that the correlator configuration is usually transmitted
>> in advance, but how much in advance ?
>> For the configuration with multiple parallel subarrays that lead time
>> probably should be increased.
>
> Currently I believe we transmit at the beginning of the preceding scan,
> except for the 1st scan of course.  For scans beyond #1 this generally means
> at least 10sec and usually a minute or more.  We could add a more stringent
> requirement on scan lengths for subarrays of course.

yes.  but an important exception to this is OTF, where scans are short 
(5 sec or so).  we may have to exclude OTF from subarrays if we take 
this approach.


> For the first scan, I believe we transmit the configurations for scans 1 and 2
> (if 2 is different than 1) at the beginning of the script, so there's
> basically no time between command and configuration.  As you know it's this
> first scan that has given most of our problems.
>
> Cheers,
>
>        Michael
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss
>