I think that you might find this interesting. I will be following it up with some experimental notes on modifications / improvements.
Hi all seis builders and modifiers!
I have built, tested and now use the Type LX1358 Linear Variable Differential Transformer Seismometer Sensor sold as a Kit by Nuova Elettronica in Italy. Their Website is at http://www.nuovaelettronica.it/ The price of the Kit is E 51.65, ~US $52 + Carriage, which looks like quite good value. The net weight is about 12 oz. The sensor is described in their Magazine No195 June-July 1998, which is on sale but it is in Italian. If you call up their Website, just type 1358 in slot labelled 'Ricerca'. If, like me, you have difficulty reading Italian, call up http://babelfish.altavista.com/tr, enter Italian ---> English translation and type in the website, but I have experienced a slow response. A picture of the NE 1.4 sec short period pendulum seismometer is shown at http://www.nuovaelettronica.it/it/pop/index.cfm?fb=scheda_kit&w.kit_id=1849 NE sell the complete pendulum seismometer Kit, with the LX1358 detector included, but it weighs about 22 lb. They also sell a microprocessor monitoring system with a 12 bit A/D converter, a paper printout and have a new 16 bit PC computer interface kit.
The LX1358 sensor board can be seen in the bottom of the vertical steel case with the ferrite rod positioned horizontally between the two white coils. The rod is suspended from the pendulum bob on two strips of glass circuit board, which also dip into the oil tub beneath the board to provide damping. The LX1358 Kit includes the circuit board, four integrated circuits, resistors, capacitors, diodes, an analogue meter, a plastic tub for oil damping, two sensor coils and the ferrite sensing rod to make up a full working sensor. It will work with simple pendulums and also with Lehmans. Basically, the sensor does just what it is supposed to do!
The principle of operation is quite simple. A ferrite rod is threaded through two identical transformer windings, with it's ends about half way into each former. The formers are of the type used for double insulated mains transformers with two equal width coils, side by side. The two magnetising windings are nearest the centre of the rod and the secondary sense windings are on the outside. A precisely controlled sinusoidal voltage is applied to the primary windings, which are connected in series so that their magnetic fields add and magnetise the ferrite rod. The secondary windings are connected in opposition so that their signals subtract. With the rod central, there is no net secondary output. If the rod is moved axially, a bit further into one coil and out of the other; the induced voltage is increased in one secondary coil and reduced in the other, giving a net output. This is passed to a direct / invert switch running in phase with the oscillator, which rectifies the signal. The chopped output is fed into a two pole low pass filter to smooth it.
This intermediate level signal is a frequency limited DC voltage level giving about 0.3 V out per mm of rod movement. This sensitivity allows for the full 15 mm travel of the ferrite rod without reaching the saturation levels of the amplifier. For setup and monitoring purposes the signal is displayed on a small centre zero volt meter. The AC component passes through a high pass filter and it is then amplified by a factor of x213 to give the final output. This allows the rod to slowly drift in position over +/-12 mm, while maintaining a very high sensitivity for the much faster seismic signals.
The double sided NE circuit board is 14.5 cm long by 12.5 cm wide by 1.8 mm thick and is of excellent quality. The component positions / number references are printed on the board. The sensor and electronics form a single unit and both are mounted on the board with the electronics on the front long side. The two sensor coils are mounted at either end of the rear long side of the board. The ferrite rod is threaded through the coils and it should be set up to move along an axis roughly parallel to the long axis of the circuit board. A small plastic tub can be attached underneath the circuit board in between the coils, to hold oil for a damping vane, allowing +/-15mm travel.
An 'on board' 12 V DC Regulator is provided which needs an absolute minimum supply input of 14.5 V DC at about 50 mA (42.5 mA measured). NE recommend a 24 V DC unregulated supply. I use a 15 V 100 mA 'double insulated' miniature transformer, two diodes and a 1000 micro farad capacitor, which gives me ~20 V DC unregulated and allows for occasional voltage reductions in the Mains Supply. I also fitted components to protect against mains and lightning surges. These components, a fuse, a LED and a switch should be mounted in a separate screened case away from the seismometer rig. You need some transient protection and a good earth to reject switching surges from domestic appliances like refrigerators, etc.
The circuit is designed around the Phillips NE5521 LVDT chip. The electronics consists of the LVDT section which produces a smoothed linear response calibrated at approximately 300 mV per mm of travel. This is followed by a high pass filter and an amplifier with a gain of x213, giving a final sensitivity of about 65 mV per micron. The output noise level of my final circuit is below +/- 0.5 mV, which corresponds to about +/- 7 nanometres. The zero reference level is at 1/2 the regulated supply voltage, about 6 V. For more information on Philips LVDT circuits, go to http//www.semiconductors.philips.com/search/ and type NE5521. This should bring up the download links for the data sheet and for applications note AN1182.
To obtain the large linear range, the sensor itself needs to be physically quite large. A 99 mm (4") long by 9.6 mm dia (3/8") ferrite rod and two square section transformer type windings, each ~3.5 cm cube, are used. The rod weighs approx. 35 gm. Quite chunky stuff! But it allows a comfortable +/- 3 mm gap inside the formers for variations in the lateral and vertical position of the ferrite rod, which I found to be very useful indeed! This makes it easy to set up and align and allows the rod to swing in an arc.
The output voltage is linear for sensor movements of up to +/- 6 mm and the
output is only 10% down at +/- 12 mm. It is 'usable' to beyond +/- 15 mm. This large range allows the sensor to also be used with seismometers of the 'garden-gate' Lehman type, but if your Lehman is set up with a long natural time constant giving it a high intrinsic sensitivity, you might wish to reduce the gain. Putting it another way, your seismometer arm can drift 1/2", but you still get 90% of the central position seismic response.
The high pass output filter has an RC time constant of 1 second (0.15 Hz), which is fine for a short period pendulum, but may be rather low for a 'garden-gate' seismometer. I found that by changing the coupling capacitor and one resistor, I could extend the time constant up to 47 sec without any problems. The existing output circuit can ALSO be configured to differentiate the displacement output to give a velocity plot. This may be the preferred option for longer period Lehman systems.
To assist in setting up the balance / zero point and for checking for drift, a small 36 mm moving coil 'tuning' meter is provided. This is driven directly off the LVDT output. The meter is nominally +/- 200 micro A at 740 Ohm. As designed, the calibration marks on the meter correspond to about +/- 0.2 mm and +/- 1.2 mm off centre. If you are planning to use this sensor with a Lehman, you might find it helpful to increase the full scale range to +/- 12 mm or more. This will give you a remote readout of the sensor position. It can be most useful to have a visible check which does not disturb the seismometer, for any tilt / drift problems.
Since I live in a place which has intermittent high traffic / environmental noise and this gave serious interference on the seismic recordings, I decided to investigate the characteristics of the LVDT filter. The LVDT detection circuit uses a two pole Sallen and Key low pass filter design with gain, which gave a 3 dB point at 33 Hz and a 20 dB point at 250 Hz on my board. It seems to be a copy of the circuit in the Philips data sheet and is better suited to commercial LVDT applications. I changed three resistors and a capacitor to improve the noise rejection above 10 Hz and to give a three pole response with a much sharper cut-off. I measured the new 3 dB point at 10 Hz and the 20 dB point at a bit under 20 Hz. The cut-off frequency may be lowered still further if desired, to below 3 Hz, but two additional capacitor changes are needed. Miniature polyester capacitors are used.
My sensor is housed inside a steel container and is well screened from external magnetic fields. However, ferrite rods were designed for the efficient reception of radio signals and the oscillator frequency is approx. 15 KHz, so interactions are possible between an unshielded sensor and the 15 KHz scanning fields produced by TVs and computer monitors and also with some VLF radio signals. I checked the sensor output for interference with low magnetic fields of ~15 KHz and it is quite sensitive, but the bandwidth is very narrow. If you use an unshielded sensor, I suggest that you do check for any pickup problems. The oscillator frequency is set by a capacitor and a resistor, so it may easily be changed. The actual frequency in not critical to within a KHz or so, but the frequency stability needs to be good. The frequency calculations are given on the Philips data sheet. The magnetic field from the coils extends to about seven cm beyond the end of the board.
The commercial free armature LVDTs that I have used had only 1/2 mm difference between the outer diameter of the armature and the inner diameter of the coil. These were seriously difficult to set up and maintain in alignment. However, Schaevitz do produce LVDTs with a 1/16" clearance. An introductory article is given at http://www.msiusa.com/schaevitz/pdf/lvdt/LVDT_Intro.pdf The 050 HR has a free armature with a +/- 1.27 mm stroke is 28.7 mm long, the 100 HR has a free armature with a +/-2.54 mm stroke and is 46 mm long - see http://www.msiusa.com/schaevitz/pdf/lvdt/HR-Series.pdf
Schaevitz have a technical article including some details of the design of LVDT sensors and systems at http://www.msiusa.com/schaevitz/products/LVDT/signal.pdf You may have to 'sign on' to download from their Website.
Another description of LVDTs complete with a Java Applet showing the operation of AC transducers is at http://www.rdpe.com/displacement/lvdt/lvdt-principles.htm
I will be happy to pass on information on the filter and other minor design changes that I found by experiment to significantly lower the noise and on the modifications to increase the time constant of the output of my LX1358, for periods of up to 47 seconds. This information, which I believe to be correct, will be supplied in good faith but without any warranty or liability, either direct or implied. The suggestions are the result of my personal experiments and I have not discussed the circuit changes with Nuova Elettronica. I have no connection with Nuova Elettronica other than being a satisfied customer. The suggestions should not in any way be interpreted as a criticism of Nuova Elettronica equipment or designers. The original Kit worked as supplied - but it worked even better with a few component changes.