From mayo@apollo.eeel.nist.gov Mon Dec  1 12:55:58 1997
Return-Path: <mayo@apollo.eeel.nist.gov>
Received: from x-files.cam.nist.gov by fs1.cam.nist.gov (4.1/SMI-DDN)
	id AA23116; Mon, 1 Dec 97 12:55:57 EST
Received: from tiber.nist.gov by x-files.cam.nist.gov (SMI-8.6/SMI-SVR4)
	id MAA22031; Mon, 1 Dec 1997 12:55:56 -0500
Received: from email.nist.gov by tiber.nist.gov (8.6.4/10.1)
	id MAA01884; Mon, 1 Dec 1997 12:55:55 -0500
Received: from piper (piper.eeel.nist.gov [129.6.65.1])
	by email.nist.gov (8.8.5/8.8.5) with SMTP id MAA18996
	for <n.heckert@nist.gov>; Mon, 1 Dec 1997 12:54:38 -0500 (EST)
Received: from smayo by piper (SMI-8.6/SMI-SVR4)
	id MAA21167; Mon, 1 Dec 1997 12:52:14 -0500
Message-Id: <3.0.32.19971201125558.00693a54@piper.eeel.nist.gov>
X-Sender: mayo@piper.eeel.nist.gov
X-Mailer: Windows Eudora Pro Version 3.0 (32)
Date: Mon, 01 Dec 1997 12:55:59 -0500
To: n.heckert@nist.gov
From: Santos Mayo <mayo@apollo.eeel.nist.gov>
Subject: noisy data
Cc: mayo@piper.eeel.nist.gov
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Status: R
Content-Length: 1511

Dear Alan:

As I said in today's conversation, I need help to device a practical way of
reducing a large amount of data 
collected from electromigration experiments. As an example, I am mailing to
you two files, ACR10.7d and ACR10.89,
showing the voltage across a thin aluminum film resistor at about 150 C
with  2E6 Amp/sqcm current density. These data are 2500 points spaced at
0.4 ms time intervals. In order to plot this information, the horizontal
axis is time (0 to 1s)
and the single column in the file is the voltage data which include
background electrical noise plus abrupt changes in resistance (ACR) created
by some rather complex mechanisms induced by electron-lattice interactions.
The random noise signal is expected to be present at all times,  while the
ACR signal, which also has random shape, appears at random times.

In order to separate noise from the ACR signal, the data in each file need
to be autocorrelated at various lag
values. The true random noise has zero autocorrelation, while the ACR
signal is not correlated to any other
signal. I am looking for a good way using Dataplot, or any other software
you may suggest, to isolate the ACR from the noise. The analysis I need to
do is somewhat similar to determining if seismographic data observed at
different locations A and B, are correlated.  In a typical electromigration
experiment some 200 files like these are generated. Clearly I need some
software capable of reducing these data. 
Thanks for your attention to this problem.

