PSN-L Email List Message

Subject: Re: 24bits A/D converters & seismology
From: Jim ODonnell jimo17@........
Date: Thu, 4 Jul 2002 17:05:07 -0700


In a message dated 04/07/02, furansowa@........... writes: 


could someone explain me why you need 24bits AD converters in seismology.

Indeed it corresponds to 16.8millions points on a 2g scale, which thus
corresponds to a resolution of about 120 nano-g (10^-12g !) 
Isn't the ambient noise much higher than this ?
Hi Francois, 

Looks like you got our attention- Here is my shot at why!
Comments on why 24 bit digitizers are needed.
Even with 24 bit digitizers and 140 db BB seismometers, the seismographs
go off scale.  Look at the records for any large EQ M>7 and you will see
the seismographs go off scale if they are within a few hundred km. of the
epicenter; Taiwan and Hector Mine come to mind.  The solution is SM
seismographs which clip ~2g, I believe the largest acceleration recorded
are ~1.7 g from the Northridge M6.7 in 1994.  Note Larry Cochraneís
PSN-accelerometer chip uses a Gain Ranging scheme to record High and Low
gain, utilizing the inexpensive 16 bit A/D, to stay on scale for the big
one (clips ~2g); but is still able to record some smaller events (clips
~70mg) which depends on the inherent noise of the chips, which are only
going to improve. 
See http://psn.quake.net/psnaccel/
 
 
Digitizer dynamic range ~1.0e07
10 bit=1024 or ~1.0 e03, 20 bit ~1.0 e06, and 24 bit ~16 e06  ~1.0e07 
 
Seismometer dynamic range for Good BB 1.0e07 (Best Guess for now)
140 db, 140/20 =7 orders of magnitude that is 1.0e07 
 
Earthquake amplitude  range: 1.0 e10
From M-1 to M8 we cover ~10 orders of amplitude velocity variation for
periods of 0.01 to 10 Hz. (See Chuck Ammonís class notes from SLU for
plot on seismic ground amplitudes for earthquakes and comments, given
below:
http://www.eas.slu.edu/People/CJAmmon/HTML/Classes/IntroQuakes/Notes/seis
mometers.html
Seismic Signals
The range of ground motions that are interesting to seismologists is very
large because the process of earth deformation occurs at many different
rates and scales.
The amplitude range of interesting signals in earthquake studies as a
function of frequency compared with a similar range of physical
dimensions of some common items. Since I am comparing the "spectral"
amplitude as a function of frequency with physical dimensions of the
common items, the analogy is not perfect, but the range of variation in
size is well represented. "D" represents the distance from the
earthquake. We usually specify large distances over Earth's surface in
units of degrees and 1 degree = 111.19 km.

The large range of amplitudes we are interested in exists because we are
interested in all the processes occurring in Earth, from small rock
fractures that form in mines to the great earthquakes that occur each
year. The amount of energy released by these different processes is
enormous, and the large range of interesting amplitudes reflects this.





In a message dated 04/07/02, furansowa@= ifrance.com=20 writes:

could someone explain me why you need 24bits AD converters in= =20 seismology.
Indeed it corresponds to 16.8millions points on a 2g = scale,=20 which thus corresponds to a resolution of about 120 nano-g (10^-12g !)=20
Isn't the ambient noise much higher than this ?
Hi Francois,

Looks like you got our = attention- Here=20 is my shot at why!

Comments on why 24 bit = digitizers are=20 needed.

Even with 24 bit digitizers and = 140 db BB=20 seismometers, the seismographs go off scale.  Look at the records for any large= EQ=20 M>7 and you will see the seismographs go off scale if they are within a = few=20 hundred km. of the epicenter; Taiwan and Hector Mine come to mind.  The solution is SM seismographs = which=20 clip ~2g, I believe the largest acceleration recorded are ~1.7 g from the=20 Northridge M6.7 in 1994.  = Note Larry=20 Cochrane=92s PSN-accelerometer chip uses a Gain Ranging scheme to record = High and=20 Low gain, utilizing the inexpensive 16 bit A/D, to stay on scale for the = big one=20 (clips ~2g); but is still able to record some smaller events (clips ~70mg) = which=20 depends on the inherent noise of the chips, which are only going to improve= ..=20

See http://psn.quake.net/psnaccel/<= /P>

 

 

Digitizer dynamic = range=20 ~1.0e07

10 bit=3D1024 or ~1.0 e03, 20 = bit ~1.0 e06,=20 and 24 bit ~16 e06  ~1.= 0e07=20

 

Seismometer dynamic range for = Good BB=20 1.0e07 (Best Guess for now)

140 db, 140/20 =3D7 orders of = magnitude=20 that is 1.0e07

 

Earthquake amplitude  range: 1.0=20 e10

From M-1 to M8 we cover ~10 = orders of=20 amplitude velocity variation for periods of 0.01 to 10 Hz. (See Chuck= =20 Ammon=92s class notes from SLU for plot on seismic ground amplitudes for= =20 earthquakes and comments, given below:

http://www.eas.slu.edu/People/CJAmmon/HTML/Classes= /IntroQuakes/Notes/seismometers.html

Seismic Signals

The range of ground motions that are interesting to seismologists is = very=20 large because the process of earth deformation occurs at many different = rates=20 and scales.

The= =20 amplitude range of interesting signals in earthquake studies as a = function=20 of frequency compared with a similar range of physical dimensions of = some=20 common items. Since I am comparing the "spectral" amplitude as a = function=20 of frequency with physical dimensions of the common items, the = analogy is=20 not perfect, but the range of variation in size is well=20 represented. "D" represents the distance from the earthquake. We = usually=20 specify large distances over Earth's surface in units of degrees and = 1=20 degree =3D 111.19 km.

The large range of amplitudes we are interested in exists because we are= =20 interested in all the processes occurring in Earth, from small rock = fractures=20 that form in mines to the great earthquakes that occur each year. The = amount of=20 energy released by these different processes is enormous, and the large = range of=20 interesting amplitudes reflects this.

Jim O'Donnell/ UNLV


[ Top ] [ Back ] [ Home Page ]