PP LAB REVIEW, TRANSIT LEG 190 Prepared by Peter Blum, 3 June 2000 Shipboard contributors: Blum, Ledwon, Meissner, O'Regan, Pretorius, Prince The main purpose of the lab review on the Leg 190 transit was to perform control measurements on all measurement systems to identify potential problems with instruments or procedures. This kind of work is difficult to accomplish on coring legs. This exercise included following up on problems and development projects described in the most recent "PP task list" appended to the April LWT minutes (http://www- odp.tamu.edu/sciops/labs/physprops/min0400.txt). This report focuses on issues that immediately affect shipboard measurements, recent database problems, and active projects. 1. ISSUES OF RECENT SCIMP CONCERN 1.1. Pycnometer 1.2. NGR and core flow 1.3. Utility of AVS measurements 2. INSTRUMENT CALIBRATION ISSUES 2.1. PWL calibration 2.2. NGR tuning (and electronic problems) 2.3. NGR energy calibration 2.4. NGR background and GRA source 2.5. MS drift correction 2.6. GRA calibration 2.7. AMST-CR zero calibration 2.8. PWS calibration/measurement 3. FUTURE TRACK IMPLEMENTATIONS/UPGRADES 4. TC DATA 5. MST SENSOR PERFORMANCE 6. PP HANDBOOK UPDATE 7. MST USER INTERFACE 1. ISSUES OF RECENT SCIMP CONCERN 1.1. PYCNOMETER A third pycnometer was finally purchased by ODP to relief the maintenance stress on the ship. One pycnometer could be repaired at the end of Leg 189, the other was sent back to shore. The new backup instrument should reach the ship this week via personnel exchange boat. The placement of the pycnometer in the corner, close to the He bottle, is an improvement. However, one problem with the present MAD station is that the convection ovens heat up that corner, which will likely increase the p-T fluctuations and adversely affect pycnometer performance. The technical staff was reminded that saving the pycnometer control measurements is important for monitoring instrument performance. The use of saving the routine control measurements (calibration sphere in one of five cells on each run) was questioned because as soon as the value deviates by >0.02 cm^3 the cell is re-calibrated. However, a record of those measurements, and the date/times re-calibration was required for each cell, could significantly help troubleshooting. The "sea-sickness" could then be diagnosed in more technical terms. 1.2. NGR AND CORE FLOW NGR data quality depends directly on counting rates, which are limited by the present detector type and core flow requirements. Although this issue was not addressed specifically during the transit, the PP LWT welcomes SCIMPS interest in finding more efficient detectors. The situation is not so serious for total count profiles. Excellent records with 5 cm sample resolution were obtained on the high-recovery legs 177 and 184, for instance. Often the problem is simply defining an optimized MST sampling plan that doesn't leave the MST idle at any time, even during site transits. On carbonate- dominated sites useful records are more difficult to obtain. A calibration procedure is basically ready to obtain real-time abundance estimates for K, Th, and U from the spectral NGR data. For this kind of routine use, detector efficiency should really be at least an order of magnitude higher than at present. 1.3. UTILITY OF AVS MEASUREMENTS The AVS has not been used routinely for many years. Most scientists would agree that the semi-quantitative data collected over a limited interval of soft sediment are of little scientific use. The PP-LWT is standing by for a SCIMP decision, hoping that it will be a simple yes or no to future AVS measurements. If "no", the instrument will be moved to shore. If "yes", the AVS is best left integrated into a track system along with the PWL systems. In terms of lab space and support, it wouldn't make sense to move the AVS off-line for occasional use. 2. INSTRUMENT CALIBRATION ISSUES 2.1. PWL CALIBRATION Confirmed that calibration coefficients are calculated correctly by the MST program. Must assume that discrepancies discovered in previous legs' data are due to fudging of the electronic delay factor as part of the calibration procedure. The resulting database problems can only be resolved by taking the face values from for data from past legs, i.e., control values are not available and corrections are not possible. The present calibration procedure is not robust. Repeated calibrations produce widely varying coefficients. This is presumable mainly the result of a shaky displacement measurement device installed on Leg 188. The operator must take extreme care not to touch that device when inserting the standards for the calibration. A replacement device is planned to be installed during Leg 191. The general procedure for PWL calibration includes the following: 1. Select the appropriate standard on the screen and start data acquisition (noise will appear on the oscilloscope). 2. Move the wet standard between the transducers until a good signal is obtained on the oscilloscope. 3. Repeat this for four standards and calculate the coefficients (button says "Calibrate", but it should really read "Calculate"). 4. Confirm visually that four points form a straight line (each dot symbol should touch the regression line. 5. Confirm that the value for the velocity of the standard (inverse of m1 coefficient of time-separation regression; displayed on screen) is 2.7 (Bill, this should be given to two decimals) . 6. Follow up with a control measurements on water, which should yield 1492±5 m/s at 23 deg. C, otherwise the procedure must be continued. If calibration problems persist after Leg 191, it may be that the transducer separation and transit time ranges over which the calibration is performed are too small (7.5 mm) and one might consider to extend that range with a new set of standards. 2.2. NGR TUNING (AND ELECTRONICS PROBLEMS) The fact that we use four individual NaI crystals and photomultiplier tubes to maximize count rates requires that the sensors are "tuned", i.e., that the amplified spectral outputs from fours sensors match. This is usually done at the beginning of each leg. Exact tuning is important for spectral analysis because only the composite from all four sensors is stored in the database and peak mismatches would dramatically reduce the results. For total count profiles exact tuning is not very critical. In any case, the tuning procedure is very useful for general system monitoring and reveals miscellaneous problems with power supply and/or amplification. During the past couple of years several problems occurred that affected data quality or prevented NGR data acquisition altogether. We spend a lot of time tuning the system repeatedly during the transit and concluded that there is a serious problem with either power supply, power amplification, electronic conduits/connectors. The ETs were still trying to diagnose the problem at the end of the transit. Tuning is basically an odd, time consuming procedure and required simply because we use four independent sensors. The best mid/long- term solution would be to eliminate it altogether by installing a modern, one-crystal and multiple-sensor, self-tuning system. The technology exists and would also increase the counting rate through better counting efficiency. The price tag may be in the order of $50k. In the meantime, one practical advice for tuning is to use not only the main K and Th peaks, but 2 additional peaks of the low- energy Th spectrum. This adds confidence in the peak-matching and makes electronic drifts more obvious. A model spectrum (combination of Th and K standard counts taken over many hours), illustrated with the characteristic peaks and their radiation energies, was prepared during the transit as an aid for future tuning. Note that the energy scale shown in the Maestro program is calculated from a calibration performed some time ago with the Maestro program and is independent from the MST calibration that is stored in the database. However, it is convenient to use the Maestro calibration for peak identification and NGR tuning Make it a habit to write down the channel (marker) number for each peak and each sensor in the prepared worksheet, which allows identification of drifting sensor outputs and hence diagnosis of the problem. Note that Marker (Maestro) = 8 x channel (MST) because of our routine reduction of the 1048-channel data to 256 channels. Peaks detected with any sensor should not be offset from those obtained with the other three sensors be more than ±1 channel (±8 Markers), which is equivalent to ±12 KeV (the spectral resolution of 256 channels). If the offset is greater, the sensors must be tuned using the gain control. 2.3. NGR ENERGY CALIBRATION Once the outputs from the four sensors are matched with the tuning procedure, the composite spectrum is calibrated to relate the 256 recording channels to radiation energy windows. During the recent past this was accomplished by assigning the energies of the main K and Th peaks to the respective channels (and the most recent MST software allows the use of two points only). The linear regression based on these two points has a problem: the lowermost part of the spectrum is truncated because of the non-linear amplification of low-count/high-energy vs. high-count/low energy parts of the spectrum (intercept coefficient is a direct measure of the truncated energy window). The calibration can be improved by including additional low-energy peaks from Th. Use Th oxide flask and KCl together to acquire the calibration spectrum and identify the following peaks: Pb-212/Ra- 224 (240 KeV), Tl-208 (583 KeV), K-40 (1461 KeV), and Tl-208 (2615 KeV). Ac-228 (912 and 966 KeV) is another, optional "double-hump"; use leftmost the maximum at 912 KeV. The example spectrum mentioned above should be used as a peak-identification aid. These four control points will improve the linear regression, i.e., decrease the intercept and thus truncation of the lower-most part of the spectrum. However, we found the power fit (y=a*x^b) to be a better approach. It has the advantage to pass through the origin by definition while still being a simple 2-coefficient solution. Although there is no a priory physical truth in this, the result is better than that from linear regression simply because the spectrum is not truncated. Until the MST program is modified to allow more than two calibration points, use TL228 (584 KeV) and Tl-208 (2615 KeV) peaks in the present 2-point interface,. 2.4. NGR BACKGROUND AND GRA SOURCE It has been noted that the GRA gamma source may affect the NGR reading because of the proximity of the two instruments (50 cm). One of the control measurements taken during the transit showed an asymmetric NGR response to a 1-cm source layer. It was speculated that this could be related to the transient shielding of the NGR, letting more GRA radiation reach the NGR sensors as the core passes through the NGR. (The alternative explanation is that our electronics problem caused that asymmetry.) A simple experiment was conducted to test the direct detection of GRA radiation by the NGR sensors: background radiation was measured three times over a period of one hour each time. First, the opening of the "NGR tunnel" on the GRA side was covered with lead shields, second, both openings of the tunnel were shielded, and finally both ends were left open. The respective background measurements, normalized to counts per second, were 5.30, 5.12, 5.88. With about 20,000 count accumulated over an hour, the statistical error (z=1) is ~0.7% or ~0.04 cps. The results indicate that the GRA does have an effect on the background if no core is present, and that the magnitude of this effect is small (0.2 cps or ~4% of the background). The effect of generally imperfect shielding from cosmic radiation is larger than that (0.7 cps or ~13% of the background). Regular core measurements (50-500 counts at rates of 10-50 cps) typically yield counting errors (z=1) of 4-15% or 1-2 cps. Our present conclusion is therefore that the direct GRA effect is within the counting error of regular measurements and thus negligible. This experiment does not assess the unlikely case of the GRA radiation reaching the NGR sensors through scattering within the core. This could be tested easily by placing a core in the GRA and NGR sensors and measuring twice, one hour each time, once with the GRA source open and once with the source closed. 2.5. MSL DRIFT CORRECTION Test measurements of low-susceptibility material and several hours duration confirmed the following: 1. Electronic drift is linear with time, which is the underlying assumption of our drift correction; 2. only ~30% of the drift is corrected for, making it pretty ineffective for measurements of biogenic oozes. It was confirmed that the calculations are fine, i.e., the software bug suspected on Leg 177 and fixed on Leg 184 portcall is not the issue anymore. We are pretty confident that the problem described in the PP-LWT task "MSL Drift Correction" is responsible for the ineffectiveness of the present implementation. Our interpretation at this time is that the loop "warms up" during measurements, causing the drift, and "cools down" during the time the other measurements are being completed, and before the reference measurement is being taken. This results in an underestimate of the total drift at the time of the reference measurement, and thus an incomplete correction. As outlined in the PP task description for some time now, the reference measurement should be taken immediately after the MS core measurements are completed. It is proposed here that the core be moved past the MS loop for zeroing at the very beginning of the run. Then, the reference measurement can be taken as soon as the core is measured and past the loop, and no extra core movement is required during the run. 2.6. GRA CALIBRATION GRA control measurements on standards produced during the transit showed that negative densities (~-0.3 g/cm^3) were returned from an empty liner. This may have a number of reasons, foremost the fact that the calibration range is 1.0-2.6, which may result in a significant error for estimates at 0.0. However, we wanted to make sure the calibration "knowns" are correct and decided to redefine the calibration standard. The standard, a segmented aluminum rod mounted in a liner filled with water, was disassembled and cleaned. A redundant piece of the rod was cut off to determine the density, which apparently was estimated before but not measured. Gas pycnometry, yielded a value of 2.699 ± 0.006 g/cm^3. The Al segment diameters xi (mm) were measured and their average densities calculated (rhoi = xi*2.699 + (66-x)*1.000). The segment densities varied slightly from the previous ones. It was also confirmed visually that the GRA beam was aligned with the center plane of the core during measurement. (A disalignement caused underestimated density values on a previous leg.) 2.7. AMST-CR ZERO CALIBRATION The Minolta CM2002 instrument manual states that for the zero (black) calibration the camera must be aimed at empty space with no reflective material within at least 1 m. At present the zero calibration is apparently being performed over the track at distance of about 30 cm from reflective, metallic material. This zero calibration is therefore not valid. It would be unpractical to dismount the camera from the track to perform the calibration. As a fix in the short term, use some black felt (ask photographer) under the instrument when performing the zero calibration. In the long term, get the manufacturer's black box or device a convenient way of aiming the instrument at 1 m of empty space. 2.8. PWS3 CALIBRATION/MEASUREMENT Problems associated with the pneumatically moveable PWS3 lower transducer were discussed (see PP-LWT task list for description). A "plug" was installed to avoid relative movement of the lower transducer in the future and avoid the related errors in transducer separation measurements that affected many legs' worth of data. A protocol was defined for calibration and measurement. Any future system upgrade should measure the relative displacement of the transducers, not the displacement of one transducer relative to a fixed second one. 3. FUTURE TRACK IMPLEMENTATIONS/UPGRADES 3.1. VST/MSCL/PWS4 In April the PP-LWT discussed a generic plan to implement the split-core/full-core Geotek MSCL, to be dubbed WMST. That plan included the idea of replacing the VS-track with the MSCL, removing the AVS (pending SCIMP directive), and installing the PWS1/2 systems at the auxiliary sampling table (where MAD samples are taken). During the transit the consensus evolved that building a PWS1/2 system for the auxiliary table was not feasible. The latest plan involves the following development: 1. Build a stand-alone PWS4 system as described in the PP task list. This removes the requirement for the PWS3 to be able to measure discrete cubes/cylinders. Space for the PWS4 should not be a problem, could be anywhere. 2. Write a control program for the MSCL, which would run in two modes. First, a fully automated run (MST-style) would run the PWS3 (the Geotek system, not our present one), GRA, and potentially MSP (magnetic susceptibility point sensor, now on AMST). The optional second run would be in a semi- automated mode for the insertion sensors PWS1/2 and AVS, which stay in their present position. 3. Install the MSCL during a leg of opportunity. The present space occupied by the VS-track could accommodate all this. 3.2. AMST The automated color reflectance acquisition apparently runs fine now. Editing of the sampling plan for a run should be implemented to avoid collection of bad data or offline data editing, as described in the PP task list. Magnetic susceptibility requires a separate run which almost makes it impossible to use. In addition the sensor appears to be too close to the core and/or track for zeroing (Leg 187 comment). More vertical offset is required. Overall the AMST, which was designed to run three instruments in three separate runs (including digital imaging), appears to take a lot of space for running the Minolta CM2002 only. A smaller, compact track could do that more space-efficiently. This issue is likely to be raised when the digital imaging system will compete for that lab space. 4. TC DATA The thermal conductivity (TC) system is not integrated into "Janus". The problem was described and specifications developed some time ago.(see PP task list). At present, scientists or ODP staff need to prepare plot files (spreadsheets) of the data from the files created by the TK04 program in a brute-force manner. Up to now, the database group has received data in various spreadsheet formats and varying information content, or has not received any data at all. Archiving of TC data has therefore been rather inconsistent over the years. The data protocol was reviewed and the following decided: All data (*.DWL and *.LIST.DAT) files are to be copied on DATA1 from now on so the database group can archive them appropriately and make them available to users. The "Run".DWL files contain the raw data (t-T series) with measurement parameters, the "Run"-.LST files contain the solutions with evaluation parameters. The files are to be organized by "hole" (as is presently done), and the spreadsheets prepared for plotting and publication are also to be put in those hole directories. Several recent legs' worth of data were located on the local computer and burned on a CD from the local hard drive and handed to the database group. 5. MST SENSOR PERFORMANCE A number of standards were produced simulating 1-cm thick core layers, including aluminum, iron cuttings (high concentration), potassium chloride, and iron cuttings (low concentration). They were mounted in core liners for measurement. In addition, a 5-cm epoxy layer was poured and mounted such that a three-layer air- epoxy-water standard could be measured. MST data were acquired for all standards to produce response curves, establish background and baseline precision, and check on the calibrations. The data will be further reduced and presented in the updated PP Handbook. 6. PP HANDBOOK UPDATE The PP Handbook (ODP Technical Note 26) is being reorganized and updated using the results of the transit work, database upgrades implemented over the past three years, information in regard to older data being migrated into the database, and more practical tips for sampling and measurement. This "second edition" should be available in fall. 7. MST USER INTERFACE Several big and small MST user interface changes were suggested. They are listed here in no particular order. Allow access to all calibration interfaces through the calibration menu (instead of splitting them into different menu items). Allow plot scales to be fixed for more than one run. They switch back to automatic at the beginning of each run, which is irritating because extreme outliers define the scales. Allow saving run via enter key. When measuring a "Control" rather that "Data" (speak, "Core"), the control standard had to be re-defined each time. Subsequent runs usually presented an empty selection window, and since a selection was required, the standard had to be defined again. This probably has to do with the database interaction concept for standards and should be addressed. NGR energy calibration coefficients are given for 2048 channels while the calibration interface shows 256 channels; fix such that calibration coefficients apply to the reduced 256 channels. Allow skipping of top and bottom of cores for all sensors. Let user enter seconds for PWL sampling period rather that DAQ's, which don't mean much to the uninitiated. The program could translate the s into a number of DAQs. Allow multiple calibration points for the NGR calibration (instead of two). Use power fit to four calibration points for NGR energy calibration (see comments above). Change "Time elapsed" to display s (instead of ms) on NGR calibration interface. Allow the operator to set the time for the NGR background measurements. Allow four digits for MeV value to be entered by operator (only allows two digits at present). Provide larger default plot window for NGR peak definition in calibration interface. Plot NGR regression points and line on spectrum plot window for a quick quality check. Print not only NGR channel number, but also calculated energy value (based on current coefficients shown) next to the curser. This would allow for peak identification and quality control.