PSN-L Email List Message

Subject: Re: tin cries - dithering
From: Christopher Chapman chrisatupw@.......
Date: Mon, 22 Aug 2011 18:56:34 -0400 (EDT)






Brett Nordgren brett3nt@.............
Sent: Tue, 16 Aug 2011 18:14
Subject: Re: tin cries - dithering


Hi Chris,
****Thanks for your suggestion. I understand your=20
approach, though implementing it will be a bit difficult.
Hi Brett,
    That rather depends on how you go about it !!=20
    An alternative might be to fit two coils close to=20
the centre of the spring and just excite them ?=20
.
***Normally, to get a decent idea of long-period noise we=20
have to record for at least a day, and preferably several.=20
And to separate instrument noise unambiguously from true=20
ground noise we would need to record three instruments and=20
use correlation techniques to distinguish their individual=20
noise contributions. See:
leeman, Wettum and Trampert
Three-Channel Correlation Analysis: A New Technique to=20
Measure Instrumental Noise of Digitizers and Seismic Sensors"
SSA v96 n1 p258
.
***As it stands, at the lowest frequencies, the noise spectrum=20
we normally see rises smoothly as 1/f and looks almost exactly=20
like the best instruments except, of course, it's not as low. =20
It is entirely possible that some of the noise we see may be=20
true ground noise. We just don't have a quiet enough location=20
to tell without playing the correlation game, which could be=20
a pain.   =20
.
    You don't happen to have a well locally in which you could float the se=
ismometer?
=20
    That is if you want absolute readings, but they will still=20
be variable. If you just want to check if the noise is lower,=20
you might have a reference seismometer and an experimental=20
seismometer and look at the long period noise of both ?
=20
****If anyone could suggest what is the magnitude of the=20
effect on the spring characteristics, I could at least model=20
the instrument response to such stepwise force changes.=20
Someone must have studied this and come up with some real=20
numbers.
.
    You have traces of a sensor with an Annealed spring=20
and an As Supplied spring.

****Wouldn't the six-second microseisms be an effective=20
dither mechanism?  They're usually large compared with=20
everything else.
    I don't think so. You are using a FORCE FEEDBACK=20
SEISMOMETER ! The increase in the strain of the spring=20
may be << 1% of the indicated signal. You are trying=20
to keep the position of the mass constant with respect=20
to the frame.
    How much feedback are you using?=20

   Regards,

   Chris




= Brett Nordgren brett3nt@bnordgren= ..org
Sent: Tue, 16 Aug 2011 18:14
Subject: Re: tin cries - dithering

Hi Chris,

****Thanks for y=
our suggestion. I understand your 
approach, though implementing it wil=
l be a bit difficult.
.
Hi Brett,
    That rather depends on how you go about it !! 
    An alternative might=
 be to fit two coils close to 
the centre of the spri=
ng and just excite them ? 
<=
TT>.
****Normally, to get a decent idea of long-period=
 noise we 
have =
to record for at least a day, and preferably seve=
ral. 
And=
 to separate instrument noise u=
nambiguously from true <=
/TT>
ground noise we =
would need to record three instruments and 
use correlation techniques to distin=
guish their individual <=
/TT>
noise contributions. See:
Sleeman, Wettum =
and Trampert
"Three-Channel C=
orrelation Analysis: A New Technique to 
Measure = Instrumental Noise of Digitizers and Sei= smic Sensors" BSSA v96 n1 p258 .
.=

****As it stands=
, at the =
lowest frequencies, the =
noise spectrum 
we normal=
ly see rises smoothly as 1/f and=
 looks almost ex=
actly 
li=
ke the best instruments except, of course, it's not as low.  
It =
is entirely possible that some of=
 the noise we see may be<=
/FONT> 
true ground noise. We just don't have a quiet enough location 
=
to tell with=
out playing the correlation game, which<=
/FONT> could be 
a pain.
.
 
That is if you want absolute readings, but they will st= ill
be v=
ariable. If you just want to check if the noise is lower, 
you might have a referenc=
e seismometer and an experimental 
seismometer and look at the long period noise of =
both ?
 <=
/FONT>

****If anyone co=
uld suggest what is the magnitude=
 of the 
effect on the spring characteristics, I could at least model 
the instrument response to such stepwise force changes. 
Someone must have studied <=
FONT size=3D2>this and come up with some=
 real 
numbers.
.
    You have traces of a sensor with an Annealed spring 
and an As Supp=
lied spring.


****Wouldn't the six-second microseisms =
be an effective<=
/FONT> 
dither mechanism?  They're usually large compare=
d with 
SEISMOMET=
ER ! The increase in the strain of the sprin=
g 
may be=
 << 1% of the indicated signal. You are trying 
to keep th=
e position of the mass constant with respect 
to the frame.
    How much feedback are you using? 

    Regards,
=

    Chris

[ Top ] [ Back ] [ Home Page ]