I wanted to send you a photo of my new horizontal sensor, but
don't think I did because you were away. It is a close copy of my old
original design, with neater execution. The pivots and jeweled bearings
were salvaged from an old microammeter.
It is being operated as a N-S sensor, and is performing well at
a period of 11 seconds, which I extend to 24 seconds or more by digital
filtering. It is very difficult to adjust for longer natural period,
because the diamagnetism of the coil causes a decentering force on the
pendulum. At longer periods, once the pendulum drifts very far off
center, it just keeps going.
Not shown in the photo are pieces of iron, 2 inches by 1 inch by
1/2 inch, that I added to the sides of the magnet assembly. The 1/4
inch top and bottom plates of the magnet were saturating, and I needed
additional flux paths to avoid that.
The base plate was cut from a 14" by 16" by 3/4 inch surplus
aluminum plate I found on eBay for $35 plus $16 shipping. It cost
another $20 to have the plate sawed in halves to yield two 8" by 14"
pieces. I purchased the 6/32 12" threaded brass rods from
Smallparts.com, the 2" by 1" by 1/4" magnets from "ninteach" for
$24.50, the #38 copper wire from Alltronics.com, and plastic sheet from
the local glazier for a cover and the coil assembly. The aluminum angle
stock and the aluminum bar stock were items I salvaged from a drum
recorder that I had built in my youth. One might find such stock at
Home Depot, if not, you can pay a lot and get it from SmallParts.com.
I wound the 1100-turn coil dry in a demountable form, and then
let polyurethane varnish diluted with turpenture soak into it and dried
in an electric fry pan at 150F (used an oven thermometer to find the
right setting). I was able to extract the finished coil from the form
with only one turn not firmly bound to the coil.
If you want to buy an amplifier for this sensor from Larry
Cochrane, you must substitute a 100K resistor for the 10K resistor
normally used across the input stage. Also, you should add a 1.0
microfarad shunt capacitor as well to prevent parasitic oscillations in
the input stage. You must also specify a voltage gain of no more than
1000 when using the 12-bit DI-154, as the sensor has an output of 0.86
v-s/cm. (You will probably have to use more gain for the 10-bit
DI-194.). Damping of the sensor is accomplished by adding more shunt
resistance across the coil. I use about 94K, not including the 100K
Suitable magnets are not always easy to find on eBay. I think
the best size is 50mm by 18mm by 6mm, which GaussBoys.com sells as
blk18. That is what I used on the E-W sensor. That size seems to
produce less diamagnetic decentering effects.