PSN-L Email List Message

Subject: Re: Lehman advantages
From: ChrisAtUpw@.......
Date: Sun, 24 Jun 2007 17:46:42 EDT


In a message dated 2007/06/24, tchannel@.............. writes:

> Subj:Lehman advantages 
> 
> Hi Folks, I am thinking about building another Lehman style Horz Pendulum 
> sensor.  I have some construction ideas I wanted to try.  Before I start, could 
> you describe the benefits of these points.
>  
> 1 FIRMLY ATTACHED THE SENSOR THE EARTH.  I wish to make the contact between 
> the earth, and the sensor as firm as possible, in this case the concrete slab 
> setting on the earth and the sensor.  Presently the sensor has three feet 
> which just set on the concrete slab. I know some people use adhesive to the 
> concrete.  If I found a way to bolt all three feet into the concrete, and a new 
> way to make the necessary adjustments, what benefits would be derived?
> I understand that even the concrete floor floats on the earth.  I am just 
> talking about the benefits of a tighter connection between the sensor and the 
> floor.

Hi Ted,

    Bolting the seismometer mounts to the floor may give problems when the 
seismometer expands with temperature at a different rate or at a different time 
to the floor. They are very unlikely to match. 

       I use three 2" square x 1/8" SS squares glued to the concrete floor. 
You can also use >5mm glass or even glazed tiles. The benefit is that you have 
a grit free, dead flat surface. Your level settings should not show drift 
either with temperature or over time, or be effected by large quakes. You can use 
pool cement to glue the plates.

       The mounting bolts need to be rigidly attached to the frame. To avoid 
thermal drift, I glue SS nuts to the underside of the arm with acrylic glue. 
On top of the arm I glue a 1/2" SS tube pillar and add a wavy washer. The SS 
set bolt has a SS ball bearing glued to a V in the end, to provide a central 
rotating contact with mounting plate. The set bolt also has a nut at the top end. 
After setting the correct height, I partly compress the wavy washer with the 
top nut. This keeps the thread in tension. The vertical alignment / side slop 
is controlled by the SS pillar and the tension.

>  2 A RIGID VERTICAL SUPPORT FOR THE UPRIGHT.  I know that on a typical 
> Lehman the vertical needs to be rigid and minimise the flex between the vertical 
> and the horz members.  If I found a way to minimise this flex, what benefits 
> might I see?  The one I have has no flex that I can see, but If I added 
> addition braces so the vertical was at 90 to the horz with the minimum of flex, 
> What benefit would there be?

       The vertical and horizontal arms need to be connected quite rigidly. 
This can be done conveniently with large triangular reinforcing plates at the T 
joint or ~ 45 deg bracing members to both the main beam and to the cross 
beam. This will minimise any cross alignment drift and tend to suppress arm 
oscillations, due to the vertical + arm + mass flexing. Unless you do this you are 
likely to pick up spurious resonant signals. The original Lehman design was 
inadequate in this respect.

       Thump the mass vertically and what do you see on the output? You need 
to eliminate any oscillations.

> 3 USING A LONGER ARM.  I used a normal length arm, and I understand if 
> space was not an issue a very long 100 meters arm would result in a longer 
> period.  I am just asking if space was avail would a 5 foot arm result in any 
> benefits, over a 3 foot arm?

       You can provide reasonable temperature and air motion control for a 2 
to 3 ft arm, but not for anything much larger. A 1 m long pendulum has a 
period of about 2 sec. To get a 4 sec period you need a 4 m pendulum. A 20 sec 
period would require a 100m pendulum. 
       The main factor you need to consider is the ratio between the natural 
period of an arm of length L and the desired seismometer period - the 1/sinA 
factor. If you try to get greater than x10 period extension, A becomes a very 
small angle. You may need fine thread adjustment screws or a slow motion drive. 
 
       A folded pendulum design is likely to be more satisfactory / easier to 
construct for mechanical periods over about 30 sec.
       An alternative method is to provide position and velocity force 
feedback to stabilise the position of the arm, but the electronics gets more 
complicated. Using electronic feedback control can run into noise and stability 
problems, but you can turn a 20 sec pendulum into a 200 sec sensor. See 
http://www.keckec.com/seismo/

       What period do you want?  The Rayleigh and Love surface waves tend to 
have periods of about 20 seconds and few are over 40 sec. For very long 
extension periods you need to measure the position of the arm, not it's velocity, or 
you just see noise.

> The last question is, if I had a sensor which was firmly attached to the 
> floor, with a very ridged vertical, and a longer arm.
> (with all the other important factors aside) What kind of improvements might 
> I expect?  I think I could build a new and improved sensor, addressing these 
> three issues.  But would these three issues make much different.  If I 
> would, see improvements would they only be for teleseismic events, or would the 
> improvements be evident in recording regional as well.

    You should see the true ground motion. There should be NO artefacts from 
the apparatus. You are more likely to be bothered by short period signals, but 
the P and S waves that you want to detect are above 0.5 Hz, often 1 to 5 Hz.

    Remember that IT IS THE EARTH WHICH MOVES ---> NOT THE SEISMOMETER ARM !!

    You have missed out some important considerations. You need to suppress, 
damp, or be insensitive to the natural oscillations / modes of the apparatus. 
Earthquakes are transient pulse type signals and can excite any natural 
oscillation modes.
    The arm and the suspension need to be rigid. The arm should be prevented 
from rotating around it's long axis. There will inevitably be some vertical 
bounce at the end of the arm, but the frequency should be above that of the low 
pass electronic filter and the sensor should be designed to have a low 
sensitivity to vertical motion. You also need to design the sensor to have a constant 
and linear sensor voltage output for mass position drifts of ~ +/-1/2". You 
ALWAYS get some position drift with a Lehman. They are very sensitive to tiny 
shifts in the local ground plane due to temperature, rain and seasonal changes. 
You need NdFeB bar magnets and rectangular coils to do this, or alternatively 
long cylindrical coils with many turns + magnets, similar to a loudspeaker, 
but with clearance gaps and coil lengths which allow for a 1/2" mass drift.
    The damping force should act ~on the line between the centre of mass and 
the lower bearing, otherwise it will try to rotate the arm about it's long 
axis. Also, place the centre of the pickup coil close to this axis. 
    Have a look at 
http://jclahr.com/science/psn/chapman/school/MKII/index.html The top wire suspension was changed to either a V cable or to a 1/2" tube. 
Both worked OK. Both hinges were altered to be crossed rods, although a ball 
on a plane works equally well. You mount the vertical rods or balls on the 
vertical support column, NOT on the arm.  The horizontal rods or the flats are 
mounted on the moving arm.
    Also have a look at the Sprengnether at 
http://www.geocities.com/meredithlamb/

       Have a look at 416 SS 'shoulder screws' 93985A205 or similar from 
www.mcmaster.com. 
       Alternatively, buy solid tungsten carbide drills and use the shank. 
They are sold for drilling fibreglass circuit board and other hard materials. 
See www.DigiKey.com or www.smallparts.com. Smallparts also sell bearings. You 
can buy flat triangular Tungsten Carbide tips for lathe tools quite cheaply with 
~ 0.3" sides. Alternatively, you can use a bit of a SS knife blade glued to 
the end of the arm.

    Regards,

    Chris Chapman   
In a me=
ssage dated 2007/06/24, tchannel@.............. writes:

Subj:Lehman advantages <= BR>
Hi Folks, I am thinking about building another Lehman style Horz Pendulum se= nsor.  I have some construction ideas I wanted to try.  Before I s= tart, could you describe the benefits of these points.


1 FIRMLY ATTACHED THE SENSOR THE EARTH.  I wish to make the contact be= tween the earth, and the sensor as firm as possible, in this case the concre= te slab setting on the earth and the sensor.  Presently the sensor has=20= three feet which just set on the concrete slab. I know some people use adhes= ive to the concrete.  If I found a way to bolt all three feet into the=20= concrete, and a new way to make the necessary adjustments, what benefits wou= ld be derived?
I understand that even the concrete floor floats on the earth.  I am j= ust talking about the benefits of a tighter connection between the sensor an= d the floor.


Hi Ted,

    Bolting the seismometer mounts to the floor may give pro= blems when the seismometer expands with temperature at a different rate or a= t a different time to the floor. They are very unlikely to match.

       I use three 2" square x 1/8" SS squares= glued to the concrete floor. You can also use >5mm glass or even glazed=20= tiles. The benefit is that you have a grit free, dead flat surface. Your lev= el settings should not show drift either with temperature or over time, or b= e effected by large quakes. You can use pool cement to glue the plates.

       The mounting bolts need to be rigidly a= ttached to the frame. To avoid thermal drift, I glue SS nuts to the undersid= e of the arm with acrylic glue. On top of the arm I glue a 1/2" SS tube pill= ar and add a wavy washer. The SS set bolt has a SS ball bearing glued to a V= in the end, to provide a central rotating contact with mounting plate. The=20= set bolt also has a nut at the top end. After setting the correct height, I=20= partly compress the wavy washer with the top nut. This keeps the thread in t= ension. The vertical alignment / side slop is controlled by the SS pillar an= d the tension.


2 A RIGID VERTICAL SUPPORT FO= R THE UPRIGHT.  I know that on a typical Lehman the vertical needs to b= e rigid and minimise the flex between the vertical and the horz members.&nbs= p; If I found a way to minimise this flex, what benefits might I see? =20= The one I have has no flex that I can see, but If I added addition braces so= the vertical was at 90 to the horz with the minimum of flex, What benefit w= ould there be?


       The vertical and horizontal arms need=20= to be connected quite rigidly. This can be done conveniently with large tria= ngular reinforcing plates at the T joint or ~ 45 deg bracing members to both= the main beam and to the cross beam. This will minimise any cross alignment= drift and tend to suppress arm oscillations, due to the vertical + arm + ma= ss flexing. Unless you do this you are likely to pick up spurious resonant s= ignals. The original Lehman design was inadequate in this respect.

       Thump the mass vertically and what do y= ou see on the output? You need to eliminate any oscillations.


3 USING A LONGER ARM.  I=20= used a normal length arm, and I understand if space was not an issue a very=20= long 100 meters arm would result in a longer period.  I am just asking=20= if space was avail would a 5 foot arm result in any benefits, over a 3 foot=20= arm?


       You can provide reasonable temperature= and air motion control for a 2 to 3 ft arm, but not for anything much large= r. A 1 m long pendulum has a period of about 2 sec. To get a 4 sec period yo= u need a 4 m pendulum. A 20 sec period would require a 100m pendulum.
       The main factor you need to consider is= the ratio between the natural period of an arm of length L and the desired=20= seismometer period - the 1/sinA factor. If you try to get greater than x10 p= eriod extension, A becomes a very small angle. You may need fine thread adju= stment screws or a slow motion drive. 
       A folded pendulum design is likely to b= e more satisfactory / easier to construct for mechanical periods over about=20= 30 sec.
       An alternative method is to provide pos= ition and velocity force feedback to stabilise the position of the arm, but=20= the electronics gets more complicated. Using electronic feedback control can= run into noise and stability problems, but you can turn a 20 sec pendulum i= nto a 200 sec sensor. See http://www.keckec.com/seismo/

       What period do you want?  The Rayl= eigh and Love surface waves tend to have periods of about 20 seconds and few= are over 40 sec. For very long extension periods you need to measure the po= sition of the arm, not it's velocity, or you just see noise.


The last question is, if I had= a sensor which was firmly attached to the floor, with a very ridged vertica= l, and a longer arm.
(with all the other important factors aside) What kind of improvements migh= t I expect?  I think I could build a new and improved sensor, addressin= g these three issues.  But would these three issues make much different= ..  If I would, see improvements would they only be for teleseismic even= ts, or would the improvements be evident in recording regional as well.

   
You should see the true ground motion. There should b= e NO artefacts from the apparatus. You are more likely to be bothered by sho= rt period signals, but the P and S waves that you want to detect are above 0= ..5 Hz, often 1 to 5 Hz.

    Remember that IT IS THE EARTH WHICH MOVES ---> NOT THE= SEISMOMETER ARM !!

    You have missed out some important considerations. You n= eed to suppress, damp, or be insensitive to the natural oscillations / modes= of the apparatus. Earthquakes are transient pulse type signals and can exci= te any natural oscillation modes.
    The arm and the suspension need to be rigid. The arm shou= ld be prevented from rotating around it's long axis. There will inevitably b= e some vertical bounce at the end of the arm, but the frequency should be ab= ove that of the low pass electronic filter and the sensor should be designed= to have a low sensitivity to vertical motion. You also need to design the s= ensor to have a constant and linear sensor voltage output for mass positi= on drifts of ~ +/-1/2". You ALWAYS get some position drift with a Lehman= .. They are very sensitive to tiny shifts in the local ground plane due to te= mperature, rain and seasonal changes. You need NdFeB bar magnets and rectang= ular coils to do this, or alternatively long cylindrical coils with many tur= ns + magnets, similar to a loudspeaker, but with clearance gaps and coil len= gths which allow for a 1/2" mass drift.
    The damping force should act ~on the line between the cen= tre of mass and the lower bearing, otherwise it will try to rotate the arm a= bout it's long axis. Also, place the centre of the pickup coil close to this= axis.
    Have a look at http://jclahr.com/science/psn/chapman/scho= ol/MKII/index.html The top wire suspension was changed to either a V cable o= r to a 1/2" tube. Both worked OK. Both hinges were altered to be crossed rod= s, although a ball on a plane works equally well. You mount the vertical=20= rods or balls on the vertical support column, NOT on the arm.  The=20= horizontal rods or the flats are mounted on the moving arm.
    Also have a look at the Sprengnether at http://www.geocit= ies.com/meredithlamb/

       Have a look at 416 SS 'shoulder screws= ' 93985A205 or similar= from www.mcmaster.com.
       Alternatively, buy solid tungsten carbi= de drills and use the shank. They are sold for drilling fibreglass circuit b= oard and other hard materials. See www.DigiKey.com or www.smallparts.com. Sm= allparts also sell bearings. You can buy flat triangular Tungsten Carbide ti= ps for lathe tools quite cheaply with ~ 0.3" sides. Alternatively, you can u= se a bit of a SS knife blade glued to the end of the arm.


   
Regards,

    Chris Chapman

[ Top ] [ Back ] [ Home Page ]