PSN-L Email List Message

Subject: FW: Re: what is your advice?
From: "Meredith Lamb" meredithlamb@.............
Date: Tue, 20 Jul 2004 23:16:13 -0600



The following msg is being forwarded and was reply written by Chris Chapman:

Hi All, 

    A few more comments on seismometers....

HELPFUL HINTS FROM A FUSSY ELDERLY SEISMIC DATA LOGGER:
1. Be sure that your STATION CO-ORDINATES are correct. You can use Microsoft Streets and Trips, or online MapQuest for the purpose. Also align horizontal sensors to true North or East, or else give true direction in sensor comments box.
    OK on the alignment and reporting. Full details would often be most helpful.

     As far as I can see, the Lat/Long search option has now been removed from Mapquest. 

    The Microsoft Streets and Trips CD only gives information for the USA and parts of Canada. 

    Has anyone got a reference for a world wide map programme which you can use with Lat + Long?
2. LOCK your timing to GPS or WWV or a crystal clock slaved to WWVB.
    Fine, but it would be helpful if Larry included a programme which DECODED the WWVB minute long signals directly and could also update the clock at user defined intervals. WWVB modules are available. The receive limit is set by local radio interference / noise, but can be expected to be >2,000 miles at night. Keep the receiver well away (10 ft +) from a CRT display. The effective signal coverage is shown at http://www.boulder.nist.gov/timefreq/stations/wwvbcoverage.htm  
3. CALIBRATE YOUR SENSOR. You would not have much use for a voltmeter with no scale on it, would you? For open loop sensors, I have two methods. One uses the raw sensor output of pendulum movement between fixed stops, the other measures the force exerted by the pendulum in response to a known current. I can tell you more if you ask.
    Remembering that seismometers are used which measure displacement, or velocity or acceleration, so we need several calibration methods. 
    Remember also that the signal which you actually see can depend quite strongly on the local sub surface geological structures. If you are sitting on a thick layer of water saturated alluvial soil, it can shake much like a jelly. If your seismometer is calibrated in say volts / metre / sec, you may see some variation in signal with the reception angle and there may be large errors in the calculated amplitude / distance measurement. An alternative method is to measure the recorded amplitude of a series of 'known' quakes and then make up a rough correlation table. 
4. DO NOT submit files with high sample rates on distant teleseisms. High frequencies are attenuated with distance, and files with excessive sample rate only take up bandwidth and archival storage space. Decimate before submitting. One to five samples per second should handle most teleseisms without loss of waveform detail. Also, do not cover an overly large time span after the L wave onset.
      You will have difficulty in separating out the P and S waves if your system cannot reproduce 0.5 to 1 Hz quite accurately. 
    I would suggest that 10 sps is a more reasonable compromise. Local quakes also have P & S components of higher frequency.
    Some locations have relatively high damping down to below about 1/2 Hz. If you also have high environmental noise, you may have difficulty in resolving quake arrival times.
    
5. DO NOT use any more FILTERING than absolutely necessary. Let some microseisms come through. Leave it to the downloader of files do more filtering if they wish.
    Sure, but it is helpful if you can at least recognise the outline of the earthquake signal. I don't find a seismic signal which completely masked in either high frequency environmental or in instrument noise, very helpful.
6. If possible, adjust your sensor's NATURAL PERIOD to least 16 seconds if you record and report teleseisms. If that is not practical, I have written an application program for WinQuake files which can digitally extend the effective period of your sensor by up to a factor of five. I use it routinely on my sensors which have natural periods of 5, 8, and 14 seconds to extend their response to 24 seconds.
   This does depend on what sort of seismometer you use and on it's natural period. Local environmental noise may become serious by 10 Hz. The Ocean background will give quite large signals somewhere between 2 and 10 seconds, which need to be filtered out. Twin Tee rejection filters have been used successfully and can give over 50 dB peak rejection. The frequency and the amplitude may change with time.
    Using beam type seismometers, it is desirable to have the period between 20 and 30 seconds. A 10 second to 30 second range allows good detection of L & R waves. 
    Simple damped pendulum seismometers with a 1.4 sec period are popular in Europe and pickup P & S waves quite nicely.
7. Control your sensor DAMPING. The barest amount of overshoot on a displaced pendulum is about right.
    I agree, but this is the ONE FACTOR in seismometer construction which is likely to give MOST PROBLEMS. I found it DIFFICULT to set up an oil damped system to about 0.8 x critical. I found it NEAR IMPOSSIBLE to keep it that way without oil temperature control, or frequent measurement / re-setup. Oil is messy, attracts dust, creeps over all surfaces, drowns insects (which then give 'bug quakes') and the viscosity is strongly temperature dependant. The surface tension also varies, which can give beam drift. 
    For oil systems, you need to measure the damping for very small displacements - use your SDR recording programme to monitor and display this. If you pull the mass 1/2" to one side and then release it, you may observe a significantly higher damping due to swirl in the oil and the system may still be under-damped for normal seismic signals. 
    I honestly don't know why anyone would try to use oil damping these days. EM damping using NdFeB magnets is simple, cheap, clean, easy to set up and to adjust.
    For a Lehman, the damping required is also strongly dependant on the period you choose and hence on the suspension adjustment. You need a fair range of damping adjustment available.
    There are two ways of providing electromagnetic damping. You can either use a flat Al or Cu plate in a pole gap with a strong magnetic field, or you can use a coil of wire half in the field and put a loading resistor across it. After having set up the period, you choose a load resistor (by experiment) which gives near critical damping. This was done with the cylindrical coils and magnets in the Sprengnether seismometers. However, the sensitivity then depends on the value of the damping resistor, which complicates matters. 
    I prefer to keep the damping and sensor coil functions separate. With two N+S pairs of opposing flat magnets 1" x 1/2" x 1/4" on 1/4" baseplates, you just move the high central field over a 1/16" to 1/8" thick Al or copper tongue till you get the desired damping.
    Electro magnetic damping is far easier to adjust than oil damping and with the ready availability of strong NdFeB magnets, it is cheap and simple to implement. Suppliers are Sedona2 on Ebay, Amazing Magnets (occasional E-Bay), "Emovendo" on ebay only (perhaps the cheapest N48 supplier), K&J magnetics ( wide range) www.kjmagnetics.com  and www.wondermagnet.com

8. Use as little AMPLIFIER GAIN as possible to avoid clipping on major events. Most of the files I see on the seismicnet site have been recorded at far more gain than necessary. You may have to make component value changes in your amplifier to accomplish this.

     Agreed, but the practical choice is likely to depend on the local environmental noise and on the Ocean background. You want to be able to resolve the background signals at times of "low noise", but you don't want strong quakes to saturate the sensor. Most amateurs do not have the option of siting their seismometer in quiet rural area.
    If you use 'period extending' software, you will need more amplification to cope with the lower amplitude of the longer period waves. You are likely to get uncertain readings if you are amplifying noise. 16 bit A/D converters, with three bits of converter noise, are not too helpful in this respect.
    Perhaps we could agree on a rough amplitude for the ocean background signals? 
    The gain you can use also depends on the noise and resolution in your A/D Converter. A/D Converter boards with +/-1/2 LSB resolved internal noise are available.

>>    Subject: feasible maximum mass for a Lehman seismometer
    At the risk of ruining a good discussion going on the last couple of days -- I would like to have some input on what the best / maximum mass for Lehman horizontal seismometer.

     The best mass is the lowest which gives you clear low noise signals. This advice 'begs the question'.
    The 'kt' thermal excitation noise sets the minimum seismic mass at about an ounce, so you are better with 1/2 lb, minimum. 
    If you use a solid metal beam, as opposed to a tube or a U channel, the moment of inertia of the beam can actually reduce the 'radius of gyration k' of the combined beam + seismic mass, giving a shorter period than you would get from the seismic mass at the end of a weightless arm. The beam needs to be light but rigid (aim for a tube weight less end fittings 1/4 the weight of the seismic mass or less). I have found the light 1/2" nominal welded stainless steel water pipe to be very satisfactory. The thermal expansion coefficient matches that of a piano wire suspension quite well. It is also cheap and you can buy brass compression fittings on which to mount the suspension, the seismic mass and the damping components. This makes the construction quite easy.
    This said, the period of a simple pendulum is independent of the mass. 
    It is advisable to keep the boom length between 70 cm and 100 cm. This is because you are using the garden gate type of suspension and shorter lengths require you to set up the side to side level position with rapidly increasing precision. This can make a 12" beam not only exceptionally difficult to adjust, but very sensitive to tilt drift, either from the suspension system or from natural earth movements.  
     Using a 30/60/90 degree triangle suspension is fine. Try to keep the boom / wire angle above 20 deg, or the suspension loading will be large. You can buy nickel coated 8 thou steel wire from a music shop, for stringing mandolins. To clamp wire, I drill a 1/16" hole just under the head of a bolt. Then I 'dish' a couple of washers by putting them on a wood block and hitting the centre hole with a large centre punch + hammer. You put the two washers on the bolt with the outer cup edges touching and feed the wire between these edges and through the hole in the bolt. This gives a good 'edge clamp' on the wire.
    For adjustment screws, I use stainless steel nuts and bolts bought from a marine / boat-builder supplier. I drill out the threaded end of the bolt with a centre drill and stick a stainless ball bearing in the conical hole. I drill a plain hole in the baseplate and stick a nut onto the lower side with methacrylate or epoxy cement. It is essential that the bolt and the nut are made of the same material, or the adjustment will drift with change in temperature. I usually use a second nut + a spring washer on the lower side of the mounting nut to provide tensioning / alignment in the thread. I stick stainless steel mounting plates onto the concrete floor, either with the special concrete 'pool adhesive' or with epoxy. If you use epoxy, it is a good idea to dry out the top of the cement thoroughly with a warm air blower.

Lastly:
>> Hi gang,
  There has been some recent discussion about calibrating seismometers.    Several years ago, my article on a calibrator using a meter movement as a force transducer was put on our 'home'  www.seismic.com.  Go to "Build Your Own Seismographic Station" then to "Article by Bob Barns".

    ?? Can anyone help me find this article, please? www seismic.com seems like a huge website with lots of advertising, but only a very old article on seismology. 'Earth Science' seems to be just advertising vitamins... 

    If you want a force calibration system with much higher forces than a meter movement, you can use a small NdFeB cube magnet with a Maxwell coil. These look similar to a Helmholtz coil, but with double the winding spacing and the windings connected in opposition. This gives a constant field gradient. You can calibrate it using a pendulum of a known length and mass, by measuring the deflection / coil current.

    Regards,





 
The following msg is being forwarded and was reply written by Chris Chapman:
 
Hi All,
 
    A few more comments on seismometers....
 
HELPFUL HINTS FROM A FUSSY ELDERLY SEISMIC DATA LOGGER:
1. Be sure that your STATION CO-ORDINATES are correct. You can use Microsoft Streets and Trips, or online MapQuest for the purpose. Also align horizontal sensors to true North or East, or else give true direction in sensor comments box.
    OK on the alignment and reporting. Full details would often be most helpful.
 
     As far as I can see, the Lat/Long search option has now been removed from Mapquest. 
 
    The Microsoft Streets and Trips CD only gives information for the USA and parts of Canada.
 
    Has anyone got a reference for a world wide map programme which you can use with Lat + Long?
2. LOCK your timing to GPS or WWV or a crystal clock slaved to WWVB.
    Fine, but it would be helpful if Larry included a programme which DECODED the WWVB minute long signals directly and could also update the clock at user defined intervals. WWVB modules are available. The receive limit is set by local radio interference / noise, but can be expected to be >2,000 miles at night. Keep the receiver well away (10 ft +) from a CRT display. The effective signal coverage is shown at http://www.boulder.nist.gov/timefreq/stations/wwvbcoverage.htm  
3. CALIBRATE YOUR SENSOR. You would not have much use for a voltmeter with no scale on it, would you? For open loop sensors, I have two methods. One uses the raw sensor output of pendulum movement between fixed stops, the other measures the force exerted by the pendulum in response to a known current. I can tell you more if you ask.
    Remembering that seismometers are used which measure displacement, or velocity or acceleration, so we need several calibration methods.
    Remember also that the signal which you actually see can depend quite strongly on the local sub surface geological structures. If you are sitting on a thick layer of water saturated alluvial soil, it can shake much like a jelly. If your seismometer is calibrated in say volts / metre / sec, you may see some variation in signal with the reception angle and there may be large errors in the calculated amplitude / distance measurement. An alternative method is to measure the recorded amplitude of a series of 'known' quakes and then make up a rough correlation table. 
4. DO NOT submit files with high sample rates on distant teleseisms. High frequencies are attenuated with distance, and files with excessive sample rate only take up bandwidth and archival storage space. Decimate before submitting. One to five samples per second should handle most teleseisms without loss of waveform detail. Also, do not cover an overly large time span after the L wave onset.
      You will have difficulty in separating out the P and S waves if your system cannot reproduce 0.5 to 1 Hz quite accurately.
    I would suggest that 10 sps is a more reasonable compromise. Local quakes also have P & S components of higher frequency.
    Some locations have relatively high damping down to below about 1/2 Hz. If you also have high environmental noise, you may have difficulty in resolving quake arrival times.
    
5. DO NOT use any more FILTERING than absolutely necessary. Let some microseisms come through. Leave it to the downloader of files do more filtering if they wish.
    Sure, but it is helpful if you can at least recognise the outline of the earthquake signal. I don't find a seismic signal which completely masked in either high frequency environmental or in instrument noise, very helpful.
6. If possible, adjust your sensor's NATURAL PERIOD to least 16 seconds if you record and report teleseisms. If that is not practical, I have written an application program for WinQuake files which can digitally extend the effective period of your sensor by up to a factor of five. I use it routinely on my sensors which have natural periods of 5, 8, and 14 seconds to extend their response to 24 seconds.
   This does depend on what sort of seismometer you use and on it's natural period. Local environmental noise may become serious by 10 Hz. The Ocean background will give quite large signals somewhere between 2 and 10 seconds, which need to be filtered out. Twin Tee rejection filters have been used successfully and can give over 50 dB peak rejection. The frequency and the amplitude may change with time.
    Using beam type seismometers, it is desirable to have the period between 20 and 30 seconds. A 10 second to 30 second range allows good detection of L & R waves. 
    Simple damped pendulum seismometers with a 1.4 sec period are popular in Europe and pickup P & S waves quite nicely.
7. Control your sensor DAMPING. The barest amount of overshoot on a displaced pendulum is about right.
    I agree, but this is the ONE FACTOR in seismometer construction which is likely to give MOST PROBLEMS. I found it DIFFICULT to set up an oil damped system to about 0.8 x critical. I found it NEAR IMPOSSIBLE to keep it that way without oil temperature control, or frequent measurement / re-setup. Oil is messy, attracts dust, creeps over all surfaces, drowns insects (which then give 'bug quakes') and the viscosity is strongly temperature dependant. The surface tension also varies, which can give beam drift.
    For oil systems, you need to measure the damping for very small displacements - use your SDR recording programme to monitor and display this. If you pull the mass 1/2" to one side and then release it, you may observe a significantly higher damping due to swirl in the oil and the system may still be under-damped for normal seismic signals.
    I honestly don't know why anyone would try to use oil damping these days. EM damping using NdFeB magnets is simple, cheap, clean, easy to set up and to adjust.
    For a Lehman, the damping required is also strongly dependant on the period you choose and hence on the suspension adjustment. You need a fair range of damping adjustment available.
    There are two ways of providing electromagnetic damping. You can either use a flat Al or Cu plate in a pole gap with a strong magnetic field, or you can use a coil of wire half in the field and put a loading resistor across it. After having set up the period, you choose a load resistor (by experiment) which gives near critical damping. This was done with the cylindrical coils and magnets in the Sprengnether seismometers. However, the sensitivity then depends on the value of the damping resistor, which complicates matters.
    I prefer to keep the damping and sensor coil functions separate. With two N+S pairs of opposing flat magnets 1" x 1/2" x 1/4" on 1/4" baseplates, you just move the high central field over a 1/16" to 1/8" thick Al or copper tongue till you get the desired damping.
    Electro magnetic damping is far easier to adjust than oil damping and with the ready availability of strong NdFeB magnets, it is cheap and simple to implement. Suppliers are Sedona2 on Ebay, Amazing Magnets (occasional E-Bay), "Emovendo" on ebay only (perhaps the cheapest N48 supplier), K&J magnetics ( wide range) www.kjmagnetics.com  and www.wondermagnet.com
 
8. Use as little AMPLIFIER GAIN as possible to avoid clipping on major events. Most of the files I see on the seismicnet site have been recorded at far more gain than necessary. You may have to make component value changes in your amplifier to accomplish this.
 
     Agreed, but the practical choice is likely to depend on the local environmental noise and on the Ocean background. You want to be able to resolve the background signals at times of "low noise", but you don't want strong quakes to saturate the sensor. Most amateurs do not have the option of siting their seismometer in quiet rural area.
    If you use 'period extending' software, you will need more amplification to cope with the lower amplitude of the longer period waves. You are likely to get uncertain readings if you are amplifying noise. 16 bit A/D converters, with three bits of converter noise, are not too helpful in this respect.
    Perhaps we could agree on a rough amplitude for the ocean background signals?
    The gain you can use also depends on the noise and resolution in your A/D Converter. A/D Converter boards with +/-1/2 LSB resolved internal noise are available.
 
>>    Subject: feasible maximum mass for a Lehman seismometer
    At the risk of ruining a good discussion going on the last couple of days -- I would like to have some input on what the best / maximum mass for Lehman horizontal seismometer.

     The best mass is the lowest which gives you clear low noise signals. This advice 'begs the question'.
    The 'kt' thermal excitation noise sets the minimum seismic mass at about an ounce, so you are better with 1/2 lb, minimum. 
    If you use a solid metal beam, as opposed to a tube or a U channel, the moment of inertia of the beam can actually reduce the 'radius of gyration k' of the combined beam + seismic mass, giving a shorter period than you would get from the seismic mass at the end of a weightless arm. The beam needs to be light but rigid (aim for a tube weight less end fittings 1/4 the weight of the seismic mass or less). I have found the light 1/2" nominal welded stainless steel water pipe to be very satisfactory. The thermal expansion coefficient matches that of a piano wire suspension quite well. It is also cheap and you can buy brass compression fittings on which to mount the suspension, the seismic mass and the damping components. This makes the construction quite easy.
    This said, the period of a simple pendulum is independent of the mass.
    It is advisable to keep the boom length between 70 cm and 100 cm. This is because you are using the garden gate type of suspension and shorter lengths require you to set up the side to side level position with rapidly increasing precision. This can make a 12" beam not only exceptionally difficult to adjust, but very sensitive to tilt drift, either from the suspension system or from natural earth movements.  
     Using a 30/60/90 degree triangle suspension is fine. Try to keep the boom / wire angle above 20 deg, or the suspension loading will be large. You can buy nickel coated 8 thou steel wire from a music shop, for stringing mandolins. To clamp wire, I drill a 1/16" hole just under the head of a bolt. Then I 'dish' a couple of washers by putting them on a wood block and hitting the centre hole with a large centre punch + hammer. You put the two washers on the bolt with the outer cup edges touching and feed the wire between these edges and through the hole in the bolt. This gives a good 'edge clamp' on the wire.
    For adjustment screws, I use stainless steel nuts and bolts bought from a marine / boat-builder supplier. I drill out the threaded end of the bolt with a centre drill and stick a stainless ball bearing in the conical hole. I drill a plain hole in the baseplate and stick a nut onto the lower side with methacrylate or epoxy cement. It is essential that the bolt and the nut are made of the same material, or the adjustment will drift with change in temperature. I usually use a second nut + a spring washer on the lower side of the mounting nut to provide tensioning / alignment in the thread. I stick stainless steel mounting plates onto the concrete floor, either with the special concrete 'pool adhesive' or with epoxy. If you use epoxy, it is a good idea to dry out the top of the cement thoroughly with a warm air blower.
 
Lastly:
>> Hi gang,
  There has been some recent discussion about calibrating seismometers.    Several years ago, my article on a calibrator using a meter movement as a force transducer was put on our 'home'  www.seismic.com.  Go to "Build Your Own Seismographic Station" then to "Article by Bob Barns".
 
    ?? Can anyone help me find this article, please? www seismic.com seems like a huge website with lots of advertising, but only a very old article on seismology. 'Earth Science' seems to be just advertising vitamins...
 
    If you want a force calibration system with much higher forces than a meter movement, you can use a small NdFeB cube magnet with a Maxwell coil. These look similar to a Helmholtz coil, but with double the winding spacing and the windings connected in opposition. This gives a constant field gradient. You can calibrate it using a pendulum of a known length and mass, by measuring the deflection / coil current.
 
    Regards,
 

[ Top ] [ Back ] [ Home Page ]