<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1">
<title></title>
</head>
<body text="#000000" bgcolor="#ffffff">
Martin Shepherd wrote:<br>
<blockquote type="cite"
cite="midPine.LNX.4.58.0403121139080.24019@goblin.caltech.edu">
<pre wrap="">
On Thu, 11 Mar 2004, Brian Mason wrote:
</pre>
<blockquote type="cite">
<pre wrap="">...
I see. So is it correct then that one overflowed ADC sample would set the
entire integration to 2^32-1?
</pre>
</blockquote>
<pre wrap=""><!---->
Correct. I can't think of any reasonable alternative. You can't just
drop overflowed samples from the parent integration, since this would
give that integration a different duration than the integrations in
the other phase-switch states, and thus compromize the cancellation
properties of phase-switching. Similarly, if you ignore the ADC
overflow, and simply add in whatever number it comes up with, then the
final integration will have an incorrect value. In other words,
whatever one does, the final integration will be corrupted, so it
makes sense to flag it with an obviously corrupt magic-value, rather
than present a subtly corrupted value that might go undetected.
Martin
_______________________________________________
gb-ccb mailing list
<a class="moz-txt-link-abbreviated" href="mailto:gb-ccb@listmgr.cv.nrao.edu">gb-ccb@listmgr.cv.nrao.edu</a>
<a class="moz-txt-link-freetext" href="http://listmgr.cv.nrao.edu/mailman/listinfo/gb-ccb">http://listmgr.cv.nrao.edu/mailman/listinfo/gb-ccb</a>
</pre>
</blockquote>
This sounds reasonable to me as well. It seems much more desirable to
"flag" and discard<br>
bad integrations than to "mask" or "massage" them. The hope here is
that such occurrences<br>
would be rare and would not significantly degrade the instrument's
overall performance in<br>
terms of the "integration failure rates" we might experience...<br>
<br>
Also, it would seem that keeping some statistics on such occurrences
might be beneficial in <br>
terms of monitoring the instrument's overall health???<br>
<br>
Randy<br>
</body>
</html>