[daip] Re: EVNtech: Sampler Statistics

[Leonia Kogan]

Hi Steve,

I want to add something to the Jim's e-mail.
You erroniously suppose that 
>the high counts should, nominally, amount to ~37% of the  total.
The probabilities of the levels staring with the highest positive one are:
Level I  +n   0.163
Level II +1   0.337
Level III-1   0.337
Level IV -n   0.163

The calculations of the probabilities can be found elsewhere, in particualar
at the Appendix 3 of my VLBA scientific memo #9, February 1995 
"Effect of Digitizers Errors on the Cross and Auto Correlation..."
At the same memo you can find the whole concept of the cross correlation 
correction using auto correlation. This concepr is used at the AIPS task ACCOR.

Regards,  Leonia
----- Begin Included Message -----

>From julvesta at zia.aoc.NRAO.EDU Wed Jun 13 08:03 MDT 2001
Sender: julvesta
Date: Wed, 13 Jun 2001 08:03:01 -0600
From: Jim Ulvestad <julvesta at aoc.nrao.edu>
Reply-To: julvesta at aoc.nrao.edu
Organization: National Radio Astronomy Observatory
X-Mailer: Mozilla 4.77 [en] (X11; U; SunOS 5.6 sun4u)
X-Accept-Language: en
MIME-Version: 1.0
To: Steve Parsley <parsley at jive.nl>
CC: lkogan at zia.aoc.NRAO.EDU
Subject: Re: EVNtech: Sampler Statistics
References: <sb275a53.075 at server4.nfra.nl>
Content-Transfer-Encoding: 7bit
X-Lines: 50
Status: RO

Hi Steve,

On the face of it, it sounds like if you are not doing the
sampler-statistics correction at the correlator, then
your users need to run ACCOR in AIPS to correct their
data amplitudes.  Is it as simple as that, or am I
missing something?

Best,

Jim Ulvestad

Steve Parsley wrote:
> 
> Dear Evntechers
> 
> Whilst investigating a problem with anomalous autocorrelations
> we have noticed significant variations in sampler statistics
> across telescope channel combinations. In two bit mode the
> high counts should, nominally, amount to ~37% of the  total.
> We are finding cases where this number is as high as 50% and
> others where it is as low as 10%. This suggests that the amplitude
> of signals presented to the samplers is not well controlled in some
> cases.
> 
> Locally we learn that there is no user adjustment to set this level
> in the acquisition rack or routine procedure to check it. Is there
> something we should investigate here?  Are the expected input
> levels for the MkIV and VLBA formatters the same and does this
> have a consequence for the stations that have changed from
> VLBA to MkIV formatter?
> 
> At the correlator we are currently not yet correcting for sampler
> statistics, and offsets from the expected distribution affect the data
> quality, especially for the autocorrelations.  Maybe this is something
> to discuss at the TOG, may we sugggest, before any attempt is made
> to adjust the existing setup.
> 
> Regards, Steve.
> 
> =============================================================
> S M Parsley
> Technical Manager
> Joint Institute for VLBI in Europe
> 
> Postbus 2, 7990 AA Dwingeloo, Netherlands
> +31-(0)521-596519 (direct), +31-(0)521-596500 (switch),
> +31-(0)621 201418 (mobile), +31-(0)521-597332 (fax)
> parsley at jive.nl
> =============================================================


----- End Included Message -----