Brett
Yea, the best I can figure out I am clipping at about 5 mm/sec.=A0 This is =
what I was wondering . I should=A0maybe lower the gain by about 5.
Barry
=A0

--- On Thu, 9/18/08, Brett Nordgren  wrote:

From: Brett Nordgren 
Subject: Re: Vertical Seismometer with Feedback, Transducer Question
To: psn-l@..............
Date: Thursday, September 18, 2008, 8:05 AM

Hi Barry,

I had been wanting to compare your clipping levels with the STS-2, which is=
=20
spec'd at 26mm/sec p-p velocity up to 20Hz and 0.34g p-p acceleration above

20 Hz.  I am wondering how that compares with what you are getting?  If you=
=20
are clipping much below that, you might want to look at whether you can=20
implement some circuit changes to make it higher.  Having made the clipping=
=20
level as high as possible with full sensitivity, you could then look at an=
=20
alternate reduced-gain output for the more local events.  If you are=20
wanting to get much higher clipping levels than the STS-2 numbers, I=20
suspect that you will probably have to give up on maintaining good=20
teleseism sensitivity.

At 07:33 PM 9/17/2008 -0700, you wrote:
>Hi Brett
>After my email I played around with the various components. I have
STM's=20
>Mcad program, but I was able to present the data in Excel as you did.

I'm sure you had noticed that all the Mathcad seismo. analyses appear to=20
contain the line:

"Td =3D Displacement Xducer time constant (~=3D 0)",

which actually means that all frequency dependent elements in the=20
forward-directed portion of the loop have been ignored.  Since these are=20
elements which strongly affect the tendency of the loop to oscillate, and=
=20
which also include the 'inverse filter', required to obtain adequate
loop=20
gain near the low frequency corner,  they very much do need to be=20
considered.  That could probably be done by adding complexity to the=20
Mathcad equations, but I agree that including them in an Excel analysis=20
should be easier to do.

>I currently have a max output of around 2000-2500 v/m/s for my sensors. I=
=20
>was able to reduce this down to around 400 by reducing the mass to around=
=20
>0.2 kg and reducing r from ~240000 to around 100000. I then adjusted the=
=20
>other feedback components to get a good response at the ends of the pass=
=20
>band and an appropriate damping value. I then could use this as one=20
>output, increase output back up by ~6x  or more and make this my second=20
>output. I'm not sure if a gain reduction of 6 is enough to make much of
a=20
>difference. Right now I'm "clipping" with a magnitude of 7.0
@ about 4000=20
>km or 5.0 @ ~200 km. Looking at my log-log graph of distance vs counts a=
=20
>gain reduction of 6 would correspond to an increase in non- clipped=20
>magnitude of ~ 0.7 at a given distance. Hardly seems worth the bother.

Regards,
Brett


__________________________________________________________

Public Seismic Network Mailing List (PSN-L)

Brett
Yea, the best I can figure out I am clipping at about 5 mm/sec.  This is what I was wondering . I should maybe lower the gain by about 5.
Barry
 

--- On Thu, 9/18/08, Brett Nordgren <brett3nt@.............> wrote:

From: Brett Nordgren <brett3nt@.............>
Subject: Re: Vertical Seismometer with Feedback, Transducer Question
To: psn-l@..............
Date: Thursday, September 18, 2008, 8:05 AM

Hi Barry,

I had been wanting to compare your clipping levels with the STS-2, which is 
spec'd at 26mm/sec p-p velocity up to 20Hz and 0.34g p-p acceleration above

20 Hz.  I am wondering how that compares with what you are getting?  If you 
are clipping much below that, you might want to look at whether you can 
implement some circuit changes to make it higher.  Having made the clipping 
level as high as possible with full sensitivity, you could then look at an 
alternate reduced-gain output for the more local events.  If you are 
wanting to get much higher clipping levels than the STS-2 numbers, I 
suspect that you will probably have to give up on maintaining good 
teleseism sensitivity.

At 07:33 PM 9/17/2008 -0700, you wrote:
>Hi Brett
>After my email I played around with the various components. I have
STM's 
>Mcad program, but I was able to present the data in Excel as you did.

I'm sure you had noticed that all the Mathcad seismo. analyses appear to 
contain the line:

"Td = Displacement Xducer time constant (~= 0)",

which actually means that all frequency dependent elements in the 
forward-directed portion of the loop have been ignored.  Since these are 
elements which strongly affect the tendency of the loop to oscillate, and 
which also include the 'inverse filter', required to obtain adequate
loop 
gain near the low frequency corner,  they very much do need to be 
considered.  That could probably be done by adding complexity to the 
Mathcad equations, but I agree that including them in an Excel analysis 
should be easier to do.

>I currently have a max output of around 2000-2500 v/m/s for my sensors. I 
>was able to reduce this down to around 400 by reducing the mass to around 
>0.2 kg and reducing r from ~240000 to around 100000. I then adjusted the 
>other feedback components to get a good response at the ends of the pass 
>band and an appropriate damping value. I then could use this as one 
>output, increase output back up by ~6x  or more and make this my second 
>output. I'm not sure if a gain reduction of 6 is enough to make much of
a 
>difference. Right now I'm "clipping" with a magnitude of 7.0
@ about 4000 
>km or 5.0 @ ~200 km. Looking at my log-log graph of distance vs counts a 
>gain reduction of 6 would correspond to an increase in non- clipped 
>magnitude of ~ 0.7 at a given distance. Hardly seems worth the bother.

Regards,
Brett


__________________________________________________________

Public Seismic Network Mailing List (PSN-L)


[ Top ] 
[ Back ] 
[ Home Page ]