PSN-L Email List Message

Subject: Re: Verticals
From: Robert McClure bobmcclure90@.........
Date: Sun, 27 Dec 2009 18:03:20 -0500


  I completely agree with Brett. Although my SpringCalc.exe program
can predict the attainment of long periods, the usual materials for
amateur sensor construction make the temperature sensitivity of a long
period adjustment too unstable to be practical. My vertical sensor is
easily adjusted to have a period of 4.4 seconds. This period is too
long to be stable over even minor ambient temperature changes. I have
to keep adjusting a movable balance weight over the seasons to keep
the pendulum position where I want it, even though the sensor is in
the basement where temperature is fairly constant, but colder in
winter than summer.

  I use WQFilter.exe to extend the response to 32 seconds, with
additional 32 second corner high pass backfiltering. I agree with
Chris that a two-stage Roberts circuit amplifier would perform better
under ordinary circumstances. The problem with the Roberts circuit is
that you must change circuit components to match the natural period if
it changes, and that it cannot compensate for over- or under-damping.
Optimal use of WQFilter requires more than minimal amplifier gain to
overcome quantization noise in the A/D converter. This is feasible,
because the largest signal from a teleseism is usually the L wave,
whose amplitude is suppressed by the original short period sensor
response, so more gain can be used without overloading the
amplifier/A/D.

  By all means, construct a feedback sensor if you can, but you will
likely still have problems with zero drift to cope with.

Bob


On 12/27/09, Brett Nordgren  wrote:
> Hi Ted,
>
> At 06:59 AM 12/27/2009 -0700, you wrote:
>>Hi Folks,  I have downloaded Bob McClure's great program
>>SpringCalc.exe.    I have put in some numbers and see it calculates very
>>long periods under certain conditions, periods of >20.
>>
>>My question is can long periods like these really be achieved?
>
> Then again, maybe.  I took a look at Bob's results and he seems to have
> been quite successful at lowering the apparent low corner frequency of a
> vertical by a factor of 8 by using an inverse filter.  I would be
> interested in knowing what the noise of a 2.5second vertical would be in
> the neighborhood of 20 seconds since, in the process of extending the
> period, that noise also gets multiplied by 8.  If the 20-second noise is
> low relative to the 20-second ground motions you are expecting, then it
> should work fine--and Bob's results would tend to bear that out.
>
> I'd like to look at the differences between extending the period by using
> feedback and doing it with an inverse filter.  May prove interesting,
> though I'm convinced that using a feedback design will have some
> significant advantages.
>
> Brett
>
> Watch our wiggles
> http://bnordgren.org/seismo/gif_images.htm
>
> or watch some very very good wiggles
> http://aslwww.cr.usgs.gov/Seismic_Data/telemetry_data/ANMO_24hr.html
>
>
> __________________________________________________________
>
> Public Seismic Network Mailing List (PSN-L)
>
> To leave this list email PSN-L-REQUEST@.............. with
> the body of the message (first line only): unsubscribe
> See http://www.seismicnet.com/maillist.html for more information.
>
__________________________________________________________

Public Seismic Network Mailing List (PSN-L)


[ Top ] [ Back ] [ Home Page ]