In a message dated 27/02/2005, jpopelish@........ writes:

I am an  electrical engineer residing in the Shenandoah valley in central  
Virginia.  Analog circuit design is my forte. I have been interested  in 
geology, volcanism and earthquakes for a long time.  After discussing  seismometer 
amplifiers with someone, recently, in sci.electronics.design, I  started 
searching the web for home built seismometer and sensor designs.   

One thing that struck me about many of the sensor designs is their  lack of 
optimisation and sophistication.  Either this means that the  sensor is not the 
limiting part of most designs or else it means that  considerable improvement 
is possible.
Hi John,
 
    Welcome! I can't comment on your discussions - I  can't find the website 
you quote. Seismometer amplifiers do need quite  specialised design with low 
noise, low drift, high gain and good filters with a  low pulse overshoot.  
 
    I don't know where you are finding the 'many'  amplifier designs? There 
is one on Larry's website which uses LT1007s and is  optimised. You need to 
optimise both current and voltage noise sources. LT1007s,  OP07s and OP27s can 
all give satisfactory performance. If you wish to use  very long periods where 
1/f noise is a limitation, MAX432 and chopper amp  circuits are available. 
Don't confuse apparently simple with unsophisticated!  You are trying to get the 
amplifier noise a factor of 10 lower than the seismic  noise. The ready 
availability of inexpensive but powerful NdFeB magnets has  allowed the use of 
smaller sensor coils and magnets with increased  sensitivity. 

After  puzzling a bit over how I might design an inductive sensor that would 
improve  upon the simple solenoidal coil and horse shoe magnet approach, I 
think I have  come up with a more sensitive design that also has noise cancelling 
capability  that will help it reject line generated fields (AC hum), 
variations in the  Earth's magnetic field caused by the solar wind and lightning 
magnetic fields.  This is based on making two similar coils that produce equal and 
opposite  signals when exposed to large, common, external fields, but produce 
equal and  aiding signals when exposed to the relative movement between the 
coils and  magnet structure.  The magnet structure also has no net external 
field to  interact with the geo field that might interact with the seismometer  
boom.

I have purchased a batch of NeFeB magnets on EBAY and am awaiting  a quote 
for construction of the 6 iron pole pieces to make one of these fist  sized 
sensors.  I will make the coil forms and wind the coils,  myself.  I will also 
make the signal amplifier and filter.  If that  all comes together, I will take a 
shot at building a Lehman  type
horizontal, long period pendulum. 
    I suggest that you consider 1" square NdFeB  magnets in a quad formation, 
NS opposing SN, in between two 1/4" thick  rectangular mild steel plates, say 
3.5" long by 2" wide. You wind a flat  rectangular coil to half cover the 
magnet poles, say ~1" square. The coil is  completely screened by the soft iron 
backing plates, which should be earthed to  the same point as the seismometer 
frame and the amplifier inputs. You can use  the same layout, but with thicker 
rectangular 1" x 1/2" magnets for an Al  or Cu inductive damping plate. This 
design gives very low stray external  magnetic fields. Try it - you will like 
it!
    AC hum is fairly low and is strongly filtered by  your 3 to 10 Hz 
amplifier filters. You should use woven screen  connecting cable. The main problem in 
domestic situations is in limiting  interference coming in through the 
utility supply and from various domestic  sources. Fridges, cookers and electrical 
heating systems can produce  large surges. You may benefit from a line filter. 
It is preferable to make the  seismometer arm and weight using non magnetic 
materials. Stainless steel water  pipe is quite useful for the arm and you can 
buy brass screw clamp fittings  quite easily, to fit.
    Don't use a knife blade or a point suspension.  Ball on a flat, crossed 
cylinder, crossed wire and crossed foil suspensions are  all OK. Single wire 
and single foil (Cardan hinge) may also be OK. See 
_http://pages.prodigy.net/fxc/_ (http://pages.prodigy.net/fxc/)     _http://pages.prodigy.net/fxc/JC.html_ 
(http://pages.prodigy.net/fxc/JC.html)    
_http://www.jclahr.com/science/psn/gldn_psn.html_ (http://www.jclahr.com/science/psn/gldn_psn.html)  &  
_http://physics.mercer.edu/petepag/MKXVII.pdf_ 
(http://physics.mercer.edu/petepag/MKXVII.pdf) 

I also  have a DATAQ DI-194-RS to hook it up to a computer but no software 
other than  what came with that unit.
    You won't be very happy with a 194 for very long,  but it is a start. You 
do really need 16 bits resolution for this type of work.  Remember that you 
also need good triggering, recording, display and data  analysis software. And 
you do need 0.1 sec timing accuracy. Calculate how much  storage space you 
would need at 20 sps for a single day?

Eventually I want to add an optical beam sensor that will make the unit  act 
as a tilt meter (true DC operation, similar to the differential capacitive  
bridge type pickup, but with much simpler support circuits) and allow  
experiments with feedback using the
original inductive sensor as a linear  motor.  This should keep me busy for a 
year or more.
    You might find some information to interest you at  
_http://jclahr.com/science/psn/index.html_ (http://jclahr.com/science/psn/index.html)  You  can 
make OK optical sensors using large area photodiode pairs (7sq mm) and a  
tungsten filament lamp with either a resistance or a voltage stabilisation  circuit. 
Infra red LEDS change their output by about a factor of 5 at constant  current 
between 0 and 100 C, so you would need to use additional photodiode  
stabilisation if you used one of them. I can get down to about +/- 15 nano  metres of 
noise, or less if I reduce the bandwidth below 10 Hz.

 
    You can also use NdFeB magnet quads and an A3515  Hall Effect sensor. See 
_http://www.geocities.com/meredithlamb/page003.html_ 
(http://www.geocities.com/meredithlamb/page003.html)  Two  pairs of rectangular magnets, one SN and 
the other NS, are mounted on  parallel soft iron backing plates connected by 
mild steel bolts. The sensor is  suspended in the central field join.
 
    There are also differential capacitor designs  available - if you need 
sub nanometre resolution. These are a subject in  themselves.
 
    Anyway, here is some "food for thought". 
 
    Can I suggest that you visit 
_http://psn.quake.net/maillist.html#archives_ (http://psn.quake.net/maillist.html#archives)  and  download the last few 
years' letters? There is a great deal of good information  and experience 
detailed therein.
 
    Regards,
 
    Chris Chapman





In a message dated 27/02/2005, jpopelish@........ writes:
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>I am an=20
  electrical engineer residing in the Shenandoah valley in central=20
  Virginia.  Analog circuit design is my forte. I have been intere=
sted=20
  in geology, volcanism and earthquakes for a long time.  After discuss=
ing=20
  seismometer amplifiers with someone, recently, in sci.electronics.design,=20=
I=20
  started searching the web for home built seismometer and sensor designs.&n=
bsp;=20
  

One thing that struck me about many of the sensor designs is their=
=20
  lack of optimisation and sophistication.  Either this means that the=20
  sensor is not the limiting part of most designs or else it means that=20
  considerable improvement is possible.
Hi John,
 
    Welcome! I can't comment on your discussions -=20=
I=20
can't find the website you quote. Seismometer amplifiers do need quite=20
specialised design with low noise, low drift, high gain and good filters wit=
h a=20
low pulse overshoot.  
 
    I don't know where you are finding the 'many'=20
amplifier designs? There is one on Larry's website which uses LT1007s and is=
=20
optimised. You need to optimise both current and voltage noise sources. LT10=
07s,=20
OP07s and OP27s can all give satisfactory performance. If you wish to u=
se=20
very long periods where 1/f noise is a limitation, MAX432 and chopper amp=20
circuits are available. Don't confuse apparently simple with unsophisticated=
!=20
You are trying to get the amplifier noise a factor of 10 lower than the seis=
mic=20
noise. The ready availability of inexpensive but powerful NdFeB magnets=
 has=20
allowed the use of smaller sensor coils and magnets with increased=20
sensitivity. 
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>After=20
  puzzling a bit over how I might design an inductive sensor that would impr=
ove=20
  upon the simple solenoidal coil and horse shoe magnet approach, I think I=20=
have=20
  come up with a more sensitive design that also has noise cancelling capabi=
lity=20
  that will help it reject line generated fields (AC hum), variations in the=
=20
  Earth's magnetic field caused by the solar wind and lightning magnetic fie=
lds.=20
  This is based on making two similar coils that produce equal and opposite=20
  signals when exposed to large, common, external fields, but produce equal=20=
and=20
  aiding signals when exposed to the relative movement between the coils and=
=20
  magnet structure.  The magnet structure also has no net external fiel=
d to=20
  interact with the geo field that might interact with the seismometer=20
  boom.

I have purchased a batch of NeFeB magnets on EBAY and am awai=
ting=20
  a quote for construction of the 6 iron pole pieces to make one of these fi=
st=20
  sized sensors.  I will make the coil forms and wind the coils,=20
  myself.  I will also make the signal amplifier and filter.  If t=
hat=20
  all comes together, I will take a shot at building a Lehman=20
  type
horizontal, long period pendulum. 
    I suggest that you consider 1" square NdFe=
B=20
magnets in a quad formation, NS opposing SN, in between two 1/4" thick=20
rectangular mild steel plates, say 3.5" long by 2" wide. You wind a flat=20
rectangular coil to half cover the magnet poles, say ~1" square. The coil is=
=20
completely screened by the soft iron backing plates, which should be earthed=
 to=20
the same point as the seismometer frame and the amplifier inputs. You can us=
e=20
the same layout, but with thicker rectangular 1" x 1/2" magnets for an=20=
Al=20
or Cu inductive damping plate. This design gives very low stray external=20
magnetic fields. Try it - you will like it!
    AC hum is fairly low and is strongly filtered b=
y=20
your 3 to 10 Hz amplifier filters. You should use woven screen=20
connecting cable. The main problem in domestic situations is in limiting=20
interference coming in through the utility supply and from various domestic=20
sources. Fridges, cookers and electrical heating systems can produ=
ce=20
large surges. You may benefit from a line filter. It is preferable to make t=
he=20
seismometer arm and weight using non magnetic materials. Stainless steel wat=
er=20
pipe is quite useful for the arm and you can buy brass screw clamp fittings=20
quite easily, to fit.
    Don't use a knife blade or a point suspens=
ion.=20
Ball on a flat, crossed cylinder, crossed wire and crossed foil suspensions=20=
are=20
all OK. Single wire and single foil (Cardan hinge) may also be OK. See http://pages.prodigy.net/fxc/&nbs=
p;   http://pages.prodigy.net/fxc/J=
C.html =20
http://www.jclahr.c=
om/science/psn/gldn_psn.html &=20
http://physics.mercer.=
edu/petepag/MKXVII.pdf
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>I also=20
  have a DATAQ DI-194-RS to hook it up to a computer but no software other t=
han=20
  what came with that unit.
    You won't be very happy with a 194 for very lon=
g,=20
but it is a start. You do really need 16 bits resolution for this type of wo=
rk.=20
Remember that you also need good triggering, recording, display and data=20
analysis software. And you do need 0.1 sec timing accuracy. Calculate how mu=
ch=20
storage space you would need at 20 sps for a single day?
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000=20
  size=3D2>Eventually I want to add an optical beam sensor that will make th=
e unit=20
  act as a tilt meter (true DC operation, similar to the differential capaci=
tive=20
  bridge type pickup, but with much simpler support circuits) and allow=20
  experiments with feedback using the
original inductive sensor as a line=
ar=20
  motor.  This should keep me busy for a year or more.
    You might find some information to interest you=
 at=20
http://jclahr.com/science/=
psn/index.html You=20
can make OK optical sensors using large area photodiode pairs (7sq mm) and a=
=20
tungsten filament lamp with either a resistance or a voltage stabilisat=
ion=20
circuit. Infra red LEDS change their output by about a factor of 5 at consta=
nt=20
current between 0 and 100 C, so you would need to use additional photod=
iode=20
stabilisation if you used one of them. I can get down to about +/- 15 nano=20
metres of noise, or less if I reduce the bandwidth below 10 Hz.

 
    You can also use NdFeB magnet quads and an A351=
5=20
Hall Effect sensor. See http://www.geoci=
ties.com/meredithlamb/page003.html Two=20
pairs of rectangular magnets, one SN and the other NS, are mounted on=20
parallel soft iron backing plates connected by mild steel bolts. The sensor=20=
is=20
suspended in the central field join.
 
    There are also differential capacitor designs=20
available - if you need sub nanometre resolution. These are a subject in=20
themselves.
 
    Anyway, here is some "food for thought". 
 
    Can I suggest that you visit http://psn.quake.net/ma=
illist.html#archives and=20
download the last few years' letters? There is a great deal of good informat=
ion=20
and experience detailed therein.
 
    Regards,
 
    Chris Chapman