| 
 
            New high precision reference oscillatorHigh current transformer driverImproved low temperature dependence drive
                transformerMagnetic modulator input for measuring frequency
                responseElectrostatic modulator input for measuring
                frequency responseGGP low pass filterImproved layout with 4 layer PCB and shielded
                input stageOn-board temperature sensorsImproved componentsGWR5 Low Pass Filter  
            GGP1 filter intended for 1 Hz sampling rate -
                (See Figure 1)
                    8 pole Bessel filterCorner frequency at 61.5 mHz (16.3 sec
                        period)Constant time delay of 8.2 seconds (Phase
                        lag 0.034 deg/cpd)100 dB attenuation at 0.5 Hz (fnyq
                        for 1 Hz sampling)Attenuation < 1% (-.086dB) below 0.01
                        Hz (100 sec period)Attenuation < 4% (-.341dB) below 0.02
                        Hz (50 sec period). 
GGP2 filter intended for 0.5 Hz sampling rate
                (optional)
                    8 pole Bessel filterCorner frequency at 30.8 mHz (32.6 sec
                        period)Constant time delay of 16.4 seconds
                        (Phase lag 0.068 deg/cpd)100 dB attenuation at 0.25 Hz (fnyq
                        for 0.5 Hz sampling)Attenuation < 1% (-.086dB) below 0.005
                        Hz (200 sec period)Attenuation < 4% (-.341dB) below 0.01
                        Hz (100 sec period).  
            Magnetic modulator
                    Adder allows injection of current into
                        the feedback circuit, to measure closed
                        loop response.Jumper selection removes adder
                        eliminating unnecessary components.Jumper selection allows both open and
                        closed loop characterization. Electrostatic modulator
                    Allows similar measurements by using
                        electrostatic forceResponse depends only on geometry of
                        sphere & plates and is independent of
                        magnetic levitation.Allows measurement of charge.  
            Improved oscillator - (See Figure 3)
                    Improved temperature stabilityReduced harmonic distortion Improved drive transformer - (See Figure 4)
                    Toroid design replaces bobbin design
                        decreasing TC by order of magnitude. High current transformer driver
                    Lowers distortion and impedance of drive
                        circuit 4 layer PCB, improved grounding and shielded
                input stage
                    Reduced cross-talk between drive and
                        sense circuits - (See Figure 5)Reduced overall broad band noise - (See Figure 6) On-board temperature sensors
                    Simplifies monitoring of electronics
                        temperatureImproved components
                    Hermetically sealed ultra stable passive
                        components Selected grade IC's for low noise,
                        thermal drift, and long term stability Conformal coating improves resistance to
                        humidity and surface contamination  
            Old gravity card had a
                temperature dependence between 0.1 to 1 mgal/°C. (See Figure
                7) New gravity cards have
                temperature dependence less than 0.01 mgal /
                °C (See Figure 8)  
            Intended for future acquisition system capable of
                sampling at approximately 1 plc (50Hz or 60Hz).
                Data system should implement real-time digital
                filter decimating output to 1 Hz. 2 pole Bessel filter Corner frequency at 200 mHz (5 sec. period).  
 
            Check feedback
                characteristics Uses should measure the response of gravity,
                temperature power and temperature balance to a
                change in temperature control null position. In Figure
                9, an offset from
                ``off" to +000 (=0.21 reset) equals about
                .084 mK and produces a gravity offset of 1.25
                mgal.
                When the feedback is working well, both the
                temperature power and balance should return to
                equilibrium with only 1 to 2 excursions.
 
            Expected temperature
                balance noise (and what gravity noise level this
                corresponds to) Temperature balance noise is about ± 0.05
                mV
                => ± 1.5 mK => 15 ngal
 
            What are the implications
                of spikes on heater power or temperature balance
                signals? Could these produce offsets in data?
                Spikes in the temperature power or balance could
                indicate that an offset in temperature has
                produced an offset in gravity. In concept, one
                could measure an offset in power associated with
                the temperature offset. However, noise limits
                measurement of power offsets to 10 mV. Therefore,
                since the power sensitivity is only 0.4 mV/mgal, one
                can only resolve offsets larger than 25 mgal.
 
            Can the spikes be recorded
                to indicate a problem? Yes, one must correlate gravity offsets to power
                spikes to prove cause and effect. However, this
                requires at least a 2 second sampling interval
                since the power spikes are only 10-15 seconds
                wide.
 
 
            Check feedback
                characteristics Users should measure the response of tile power
                and tilt balance to a change in both X and Y
                reset. In Figure 10 the X Reset has been changed from
                +521 to +526 which corresponds to about 30 mradian.
                Check that both the power and balance return to
                equilibrium with only 1 or 2 oscillations.
 
            Expected tilt balance noise
                The tilt noise will depend on how quiet the
                user's site is. At GWR the tilt balance noise is
                about ± 10 mV which corresponds to a tilt noise
                of about ± 0.1 mradian. Users are encourage to
                measure the relationship between micrometer
                ``mils", tilt balance volts (BD=7), reset
                units and mradian. In the instrument tested at
                GWR: 1 mil = 2.9 V = 5.8 reset = 32 mradian.
                These relationships will depend on the length of
                the tilt arms, the electronic gain, and the
                tiltmeter sensitivity.
 
            What are the implications
                of spikes on X or Y tilt power or balance
                signals? Could these produce offsets in data?
                As with temperature, these spikes could indicate
                that changes in the tilt control position are
                producing offsets. However, such a conclusion
                cannot be reached without using secondary
                tiltmeters or by correlation of spikes with
                gravity offsets. The width of tilt balance spikes
                is 2 to 4 minutes and can easily be observed by
                sampling with a 20 second interval.
 
            Tilt geometry & manual
                tilt desensitizing (FB & LR) All compact Dewars and many other instruments now
                use equilateral leveling frames versus the older
                isosceles triangular frames. The geometry of the
                two systems is illustrated in Figure
                11. The advantage
                of the isosceles support frames is that the left
                and right axes are orthogonal and that the two
                tiltmeters can be aligned with these axes. In
                SET-UP, this means that reading of the X (or Y)
                micrometer does not affect the null position
                measured with the other micrometer. With the
                equilateral frame, the user must simultaneously
                use both left and right micrometers to define two
                new tilt axes labeled Left-Right (LR) and
                Forward-Back (FB). LR is defined by moving both X
                and Y micrometers in the same direction, e.g.:
                DELTA X = +5 mils and DELTA Y = +5 mils; while FB
                is defined by moving in opposite directions,
                e.g.: DX = +5 mils and DY = -5
                mils.
 
            Tilt desensitizing in
                feedback using X or Y Resets With the leveling platform operating in feedback
                (RUN), both equilateral and isosceles support
                frames can be tilt minimized in an orthogonal
                fashion using the Left (X) Reset and the Right
                (Y) Reset. The X & Y reset functions produce
                orthogonal tilts because they produce electronic
                offsets in the tiltmeters themselves. However,
                since the thermal levelers do not produce
                orthogonal tilts, they both must respond to
                either a X or Y Reset change. This is shown in Figure
                12, where the X
                tilt axis is being tilt minimized using only the
                X-Reset. As shown, the X Power response is larger
                than the Y power. From the geometry of the
                equilateral frame and tiltmeter alignment, the
                ratio of the Y tilt power should be 0.27. In the
                example, the power ratio was 0.36 which is most
                likely due to errors in the electronic square
                root function in the feedback network and to
                different leveler response to heat.
 
            Expected
                tilt noise close to minimum The slope of the tilt curve Dg/( D Reset)2
                can be used to calculate the expected gravity
                noise produced by tilts. For example, in the data
                show, the gravity noise is about:
 DgN @ 4.6 x 10-4 (mgal/(mrad)2)(Q-Q0) DQ                           (1) 
            Therefore, the
            tilt noise depends on the tilt slope, how close the
            instrument is adjusted to the tilt null Q0 and
            the noise level DQ at the site. In the example shown, if (Q-Q0 <
            1 Reset (=5mrad) and the noise DQ = 0.1mrad,
            then the tilt induced gravity noise DgN @ 2.5 ngal.
            At a site such as Membach, the tilt noise is about 5
            times quieter than at GWR. Therefore, the tilt
            induced gravity noise will be DgN
            (Membach) @ 0.5 ngal 
            Need for an annual tilt
                check GWR recommends that users perform annually tilt
                checks for both X-Reset and Y-Reset. When the
                system is operating properly, neither the X or Y
                Reset values will change in time. This means that
                the tilt minimum of the gravity meter and the
                null points of the tiltmeters have remained
                stationary with time.
 
 - (Membach,
        Belgium data as an example)  
            Instrumental channels to record gravity and
                for SG maintenance Data channels can be classified into three
                categories. Recording the main signals of gravity
                and air pressure is already fully discussed in
                previous GGP meetings. The auxiliary signals are
                used to check and monitor the subsystems of the
                gravimeter to make sure they are operating
                correctly. It is important to establish a
                baseline of operation on the subsystems for
                regular comparison to make sure performance does
                not degrade in time. This will be especially
                helpful when problems arise that must be
                discussed with GWR. Comparison of data with the
                system working correctly versus improper
                operating is essential for diagnosing failures
                rapidly. The geophysical signals are of prime
                importance for correlating with observed gravity
                changes. This correlation allows further
                reduction of secular or short term signals from
                the data. For example, groundwater will produce
                long term secular signals; while rain fall may
                produce 1 to 3 day spikes in the data. Finally,
                regular absolute gravity measurements allow
                either confirmation of long term trends in
                superconducting data or correction of
                instrumental drift.
 
            Main signals:Gravity
 Air pressure
Temperature balanceHeater power
 X & Y Tilt balance
 X & Y Tilt power
 Electronics temperature
 Vault temperature
 Neck temperatures
 Helium flow
 (Compressor water coolant temperatures)
 (mode data?)
Geophysical signals:Permanent GGP measurement of elevation changes
 Ground water variations
 Soil moisture
 Rain and snowfall
 Other??
Periodic Absolute Gravity measurements  
            Notes of caution: 1) User's data systems should use
                differential and isolated inputs. Each
                signal from the GEP-2 electronics has a
                corresponding return (common) which is referenced
                to a specific location on the board where the
                signal is generated. Connecting commons together
                at the data system will disrupt the electronic
                design and will produce ground currents and
                noise. It is for this reason that GJ1 through GJ4
                and the front panel connector are isolated BNC
                connectors. The commons of these BNCs should not
                be grounded.
 2) Use caution on recording Heater Power.
                We have observed that connecting a long cable to
                the heater power (GJ5 pins 7/15) can cause noise
                on the gravity signal. Pin 7 connects directly to
                the output of the temperature feedback integrator
                (U7). The increased capacitance to ground present
                in the cable connected to pin 7 may cause U7 to
                oscillate at a high frequency. This oscillation
                will shift the DC level of the output and produce
                a shift in the temperature control position. If
                this happens rapidly the result looks like noise.
                If it happens infrequently it will look like
                random offsets on the gravity meter. Users with
                short leads between GEP-2 and their data systems
                are probably safe from oscillations. However, for
                users with long leads we recommend that they
                discontinue recording the heater power by
                disconnecting pins 7/15 in the cable connected to
                GJ5.
 
            Daily & weekly analysis of data
 Daily - It is important to analyze and
                monitor the main and auxiliary signals frequently
                in order to minimize data interruptions and gaps.
                The best way to guarantee the highest quality
                gravity data is to generate a tidal residual
                signal by subtracting a tide model based on the
                analysis of previous data. Ideally this can be
                done on a daily basis as in the data shown in Figure 13 (air
                pressure, tidal residual, & theory from
                Membach). Once the residual is generated, the
                user can determine:
 1) Has the instrumental noise level remained low?
 2) Are there any offsets or spikes (besides
                earthquakes)?
 3) Are there any abrupt changes in slope or
                drift?
 4) Is the data system, timing and data storage
                working properly?
 
 Weekly - The same type of analysis can
                be done on a weekly basis as in the data shown in
                Figure 14 (air
                pressure, tidal residual & theory from
                Membach). This residual data should be compared
                to the X & Y Tilt balance and power (Figure 15), Heater
                power and temperature balance (Figure 16),
                electronics temperature, and vault temperature.
                For example on the X tilt power there is a large
                spike. However, by comparison of this event to
                the tidal residual one can determine that this
                event did not produce a corresponding spike or
                noise on the residual data. What caused this
                spike then? If it was a user entering the vault,
                it should cause a change in vault temperature or
                produce an entry into the log book.
 
 Clearly, the more often the tidal residual is
                generated and checked the shorter gaps in the
                data will be. However, weekly checks of the
                temperature, tiltmeters, and refrigeration
                systems are most likely adequate. Monitoring the
                neck thermometers to observe an increase in
                temperature is the most rapid way to determine
                when the cold head is beginning to age. This
                allows plenty of time to replace the cold head
                since they degrade slowly.
 
            Is the Mode filter useful anymore? Recording the mode bandpass filter on a strip
                chart recorder (See Figure
                17) is a quick and simple way to examine the
                high frequency noise present on the gravity
                signal. Therefore, it may be useful for users who
                are not generating and examining daily tide
                residuals. However, it can only be used for
                checking for changes in instrumental noise and
                for offsets and spikes. It cannot be used for
                changes in drift (since DC signals are filtered
                out) or for operation of the data system or
                storage medium.
 
 The mode filter could be recorded as auxiliary
                data at 20 second intervals and be used to scan
                the data for offsets. As shown in Figure 17, a 40 mV
                step function into the mode filter produces a
                spike response of about 1.22 V (peak to peak).
                Therefore, the magnitude of offsets producing
                such spikes can be (practically) read of the mode
                filter data if the noise level is low. For high
                noise regions it is more difficult to remove such
                offsets. Possibly, the mode filter data would be
                useful to other GGP participants to quickly
                determine data quality before they commit to the
                process of ``cleaning" the data for further
                analysis.
 
            Importance of Absolute Gravity Measurements
                Figure 18 shows
                the gravity residuals of the superconducting
                gravimeter (SG) compared to the measurements from
                the absolute gravimeter (AG) at Membach, Belgium.
                In this case there was a data gap and offset that
                occurred in the SG near the end of May-96. This
                offset was estimated and corrected by comparison
                of SG to AG data. From these data sets it appears
                that the SG at Membach has very little
                instrumental drift. Generally, however, some
                drift may be present on the SG at other sites.
                Such drifts are always monotonic and usually
                decrease in time. These drifts can be measured by
                comparison of the SG residual data to AG data if
                AG is taken at regular periodic intervals. The SG
                can also be used to check proper operation of the
                AG. As can be seen from the data, there are two
                AG data points at 18000 hours that disagree
                significantly with the rest of the data.
 
            Importance of measuring other geophysical data
                The agreement of SG and AG data markedly
                increases confidence in both data sets and proves
                that the observed gravity variations are of
                geophysical origin and not of instrumental
                origin. The geophysicist's job is now to identify
                the cause of such variations. One common method
                of doing this is by establishing a correlation
                between the gravity residual and other
                geophysical signals. However, this powerful
                technique can only be used if the geophysical
                signals have been recorded over the same time
                period as the gravity data itself! Therefore,
                users must establish a list of geophysical
                signals that are most likely to influence data at
                their sites and implement methods to measure and
                record them as soon as possible.
 
 Acknowledgment: The author sincerely thanks
                Olivier Francis and Marc Hendrickx for supplying
                the data from Membach, Station, Belgium which is
                used as examples in this section.
 
 
            GOP recommends accuracy better than 0.1 hPa.
Admittance of air pressure to gravity is about 3
                nms2 / hPa.
                    GOP recommends measuring to accuracy of
                        0.3nms2
Best transducers are stable to no better than 0.1
                bPa / year.
To maintain GOP specification, calibration on
                yearly basis is required.
                    Calibration by factory with dead weight
                        tester.Calibration in the field with transfer
                        meter. 
            Quartz Bourdon Tube
                    Mainly for Laboratory standardVery Expensive.
Vibrating Cylindrical oscillator (Weston or
                Schlumberger. sensor)
                    Long history of stable measurementsSensitive to changes in media density
                        (humidity).
Silicone Resonant Pressure Transducer (e.g. RPT
                by Druck)
                    Newest TechnologyStable, accurate, insensitive to media
Capacitance bridge (e.g. MKS Baratron)
                    Best suited for low pressure (vacuum) 
            Vibrating Cylindrical Oscillator and RPT sensors
                are intrinsically digital
                    Changes in Pressure effect the frequency
                        of a resonating structureConversion to an analog signal is an
                        unnecessary step and degrades the system
                        stabilityData should be collected digitally which
                        may require modification of data
                        acquisition software. Comparison of various types of
        pressure transducers
            
                |  | Temp.
                Controlled Capacitive Sensor (MKS Baratron) | Silicone
                Resonant Pressure Transducer (RPT Druck) | Vibrating
                Cylinder (Weston or Schiumberger sensor) |  
                | Accuracy | 1.5 hPa | 0.1 hPa | 0.1 hPa |  
                | Drift | not
                specified | 0.1 hPa /
                year | 0.1 hPa /
                year |  
                | Operating
                temp. range | +15°C
                / +40°C | -20°C
                / +60°C | -40°C
                / +70°C |  
                | Temperature induced
                errors | 0.3 hPa / °C | 0.2 hPa over full
                temperature range | better than stated
                accuracy |  
                | Humidity sensitive | NO | NO | YES |  
                | output format | analog (DCV) | digital (R5232/R5422
                selectable) | digital (R5232 OR
                R5422) |  
                | DC! Gain calibration | Evacuate! dead weight
                test | dead weight test | dead weight test |  
                | Typical usage | low pressure | general barometric | barometric |  
                | History |  | Recently developed | Long History, industry
                standard | 
            All GGP participants should upgrade pressure
                transducers to either the RPT or Vibrating
                Cylinder type
Data should be collected digitally via a serial
                port rather than through an analog channel
GWR now recommends a Silicone Resonant Pressure
                Transducer manufactured by Druck Instruments
GWR (or other?) could maintain a calibrated
                transfer meter for use by GGP participants if the
                community desires it 
 
            Refrigeration system is the most maintenance
                intensive part of the SG system!
Compressor adsorber replacement every 10,000
                hours (13 months)
Annual coldhead clearance check
Coldhead maintenance approximately every 20,000
                hours
Compressor cooling must be maintained within
                specified limits
                    Evaluate heat management criteria
                        carefUlly when setting up the systemChoose proper cooling system for ambient
                        conditionsMonitor cooling water temperatures 
            Failure to provide adequate cooling will
                eventually result in premature compressor failureIn some cases inadequate cooling can result in
                oil contamination to the entire cooling system
                including the gas hoses and coldhead!APD HC2 Compressor specifies cooling water is
                maintained within certain limits!
                    2.3 liters / minute minimum flow rate27 "C maximum inlet temperatureApproximately 1.8 kW of heat rejection is
                        required. 
            Dry type coolers
                    Operate by passing water to be cooled
                        through tubes attached to cooling fins.
                        Fans force ambient air over fins to
                        remove heatThese coolers will never cool to
                        the ambient air temperature, cooling to
                        within 5 "C to 10 "C of ambient
                        is typicalInexpensive, simple, reliable
Compressor driven chillers
                    Operate with freon or other compressed
                        gas (like a home refrigerator or air
                        conditioner)Require additional electrical powerCan maintain cooling water at temperature
                        far below ambient air temperature 
            APD Cool-Pack 4
                    Dry type coolerMaintains cooling water inlet to HC2
                        approximately 5°C
                        above ambientAcceptable when ambient air temperature
                        does not exceed 22"C
GWR AW75 Cooler
                    Dry type coolerMaintains cooling water inlet to HC2
                        approximately 4 °C
                        above ambientAcceptable when ambient air temperature
                        does not exceed 23°CIncreased pump capacity decreases
                        compressor outlet temperature maintaining
                        increased safety marginHeavy duty pump, fan, and heat exchanger
                        for increased reliability
Schreiber 1 Ton Compressor Driven Water
                Chiller
                    Compressed freon refrigerator maintains
                        low temperatures in extreme ambient
                        conditions (up to 45°C)Rated for outdoor useLow ambient controls allows operation in
                        sub zero conditionsLarge water reservoir reduces
                        refrigeration cycling improving
                        reliability
RECOMMENDATION -
                Purchase suitable water chiller from your local
                refrigeration expert! |