TN 295 .U4 I H ■ ^H m*j< ■ ■ ■ I ■ ( Ihiflft'W !,->.' rt»/ :■> ■ > .-ft; ■ • ■ I ■ ■ %/ • v.^ :. V^ X °°*^V * *> v »!,•"* c\. *?* ..i OB* <^ .., ^•y v^y.. • "^ :'JS •o »4q V • : ^1 ^ / L^^. v^ .-ate- %,♦* .-^fefe\/-.-ate> %^ ^ vV^, Y^ 1 vV^, 0^ o«^».^o. .°^ i A* *^J^^* A «. *.£.,£..• «P^ A* >«4RW^" t ^ «. *»^^.« ^ A* ^^ • * ^ *° ^ y ^ -Mm* .#"%. • «5°a a Hq, '•' <** 'V A <* 'o.l W ... <«> o* ^. *?Kv\a <^ *•.»* ^ ho ^ »bv* V%^*\/ °o*^- / y^ & .^V>jlV -% >' ••' -^. ^ ^- 4Va\ V./ •' q,^ * . , 1 • aP J i»v W > _**V : .1B§V l ^W: _**%, \mg: J>% °»' ***■ & ^ VSR** > v ^> •: v« .V . o » ■ I ^ - ■q. */,,•' a?' «5 * V %>** V o 1 ^ .& •vs^r- ^ ** * iO v .*«il 0* ^ *• ^ & *A ' w q. •„, °- /♦*% ' %ifc o - X^\> * s4Mk>* /**m ^ / V ** ... v v /r %. A J v v ,•* r " « " V lV^ V w :• ° 4 .♦* ^ » iL ^^ v »!,•»*. q, ? v .* c) ♦/TvT» A V *♦ * %> d ••^fl£- ° ^^ fgiisk* ** c *"^a£- ° 4 *£(!%?>■> S. A ^^J^L^V. .v l- >, .^ .**«* -*iSBS-: ^^ ;^i%^: ^« v^ 4 ^ * *\ % * .0* .o^^^b, -V % m ^ S V y i? **o« ■bv* ^«fc..%. .y.^°:.X y..ittfe,^. ,/.;i-i:-X ^,^fc-^ Vv^ q, "•'TVi* A <* *••»* ^ G N ^. *?TX* .A •V > ^ 4 0, ? ^ ^ .A' .f -, '•*. r>^ ,..' J£J 8960 Bureau of Mines Information Circular/1983 Microseismic Instrumentation Developments A Tape-Triggering System and Energy Analyzer By Bernard J. Steblay UNITED STATES DEPARTMENT OF THE INTERIOR (MLM_&&*. fr ^fM^ g) Information Circular 8960 Microseismic Instrumentation Developments A Tape-Triggering System and Energy Analyzer By Bernard J. Steblay UNITED STATES DEPARTMENT OF THE INTERIOR William P. Clark, Secretary BUREAU OF MINES Robert C. Horton, Director T f\0 J Library of Congress Cataloging in Publication Data: Steblay, Bernard J Microseismic instrumentation developments. (Information circular / United States. Bureau of Mines ; 8960) Bibliography: p. 12. Supt. of Docs, no.: I 19.4/2:8960. 1. Mine safety— Equipment and supplies. 2. Seismology— Instru- ments. 3. Digital-to-analog converters. 4. Analog-to-digital convert- ers. I. Title. II. Series: Information circular (United States. Bureau of Mines) ; 8960. TN295.U4 622s [622'.8] 83-600282 oL contents Page Abs tract 1 Introduction 2 Tape-triggering system 2 Recording section 2 Reproducing section 4 \. Energy analyzer 6 Analog-to-digital conversion 6 Arithmetic processing. 8 Application. 9 Conclusions 11 Tape-triggering system 11 Energy analyzer 12 References 12 ILLUSTRATIONS 1 . Tape-triggering system and analog reproducer system 3 2. NRZ recording signal format 4 3. Timing channel format 5 4. Time of day information as presented on output record 5 5. Energy analyzer system 6 6 . Waveform digitization 7 7. Energy analyzer arithmetic processing 10 8 . Energy system circuit modifications 11 TABLE 1 . Internal energy processing 8 A UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT dB decibel pF microfarad ft foot )JS microsecond h hour ns nanosecond Hz hertz pet percent ips inch per second pF picofarad kHz kilohertz s second min minute V volt MICROSEISMIC INSTRUMENTATION DEVELOPMENTS A Tape-Triggering System and Energy Analyzer By Bernard J. Steblay ] ABSTRACT Two instruments have been constructed that extend microseismic data collection and processing capability for Bureau of Mines research in rock burst, coal bounce and outburst, and roof fall monitoring. The first instrument is a 13-channel tape-triggering system. This system allows better utilization of instrumentation tape recorders by collect- ing data digitally, turning the tape on only when valid data are pres- ent, and increasing the dynamic range of the recorder by using digital data. A digital-to-analog converter is used to provide analog output of the recorded digital data. The instrument records a much larger time slice of transient events than do current commercial devices. The sec- ond instrument is a microseismic energy measurement device. This device uses recent developments in integrated circuits to overcome the dynamic range and accuracy limitations of previous instruments. The analog sig- nal is converted to a digital one, and then high-speed multiplication techniques are used to square the amplitude value in real time and inte- grate it. 1 Mechanical engineer, Denver Research Center, Bureau of Mines, Denver, CO. INTRODUCTION The first instrument addressed in this paper deals with the frequently en- countered problem in data collection of recording analog data that consist of randomly occurring transient events. If an analog tape recorder is run continu- ously, it places time constrictions on the data collection and also requires laborious scanning of a long tape to find a few pieces of significant data. This problem occurs for the Bureau of Mines in collecting microseismic data. The tape- triggering system is a solution to these problems. It triggers the tape only when valid data are occurring, but because it first records the data in its solid state memory, it does not lose any pretrigger data. It can operate unattended for long periods since it requires servicing only when the tape fills up. It operates with ordinary frequency modulation (FM) analog recorders but greatly increases their dynamic range because of its digital en- coding. Using 24,000-byte memory per channel allows long transients to be recorded. The second instrument addressed deals with the data collection problem of determining the energy of an incoming analog signal. This energy estimation is a valuable piece of data for rock stabil- ity determination. Analog or quasi-analog approaches such as voltage-to-frequency conversion for acoustic emission measure- ment have serious limitations in dynamic range and accuracy (l). 2 The energy mea- surement system discussed in this report overcomes such limitations by using a fully digital processing approach. TAPE-TRIGGERING SYSTEM The tape-triggering device was con- ceived and specified by the Bureau of Mines Denver Research Center and designed and constructed under a Bureau contract (_2 ) . Its two major sections are the event recorder and control unit and the event reproducer unit (fig. 1). The units are designed to work with an FM instrumentation tape recorder with a min- imum bandwidth of 40 kHz (3). RECORDING SECTION The event recorder and control unit first filters the incoming data using a low-pass filter set at either 1 or 10 kHz. The filtered analog signal is then converted to a 12-bit digital signal at an 83.3-kHz sample rate. The sample rate was chosen to provide good analog resolu- tion at the output of the event repro- ducer even at the upper bandwidth of 10 kHz. The 12-bit resolution provides a total dynamic range for the bipolar input of greater than 70 dB , which exceeds the typical performance of an analog FM recorder by more than 16 dB. The digitized amplitude values are then stored in a 294,912-bit (24,576-word) per channel memory. This large per channel memory allows the storage of nearly 0.3 s of signal in memory at any given time. This overcomes the time limitation caused by the short memories typically found in commercial transient recorders. The event that the author felt made such a device commercially practical was the development of 64,000-bit charge- coupled-device (CCD) shift register mem- ories, but these could not be supplied in time for construction. Substituting 64K dynamic access memory proved satisfactory. The memory of the event recorder is divided into pretrigger and posttrigger sections. Data are continually being recorded by the memory. As each new sam- ple is collected, the previous sample is shifted one memory cell down the chain. The trigger criteria are controlled by a microprocessor. If a valid trigger occurs, the 2,500 samples that were col- lected just prior to the trigger are saved. The remaining 22,076 words of the 2 Underlined numbers in parentheses re- fer to items in the list of references at the end of this report. FIGURE 1. - Tape-triggering system and analog reproducer system. memory chain are then filled with post- trigger data. Thus about 0.03 s of data prior to the trigger is saved for analy- sis. This information is often crucial for making first-arrival analyses. The microprocessor controls both the threshold criteria and the valid event criteria. The threshold level is se- lected by front panel switches and ranges from 0.4 V to 2 V in 0.2-V increments. If the threshold level on any channel is exceeded, the event recorder begins re- cording all of the other channels simul- taneously. When the memory chain is filled, the microprocessor determines whether enough channels exceeded their thresholds to constitute a valid event. This channel requirement number, which is settable from 2 to 5, insures that a spurious single event on one or more channels will not trigger the system. In the case of a mining application, a typi- cal spurious event would be drilling near a geophone. Once it has been determined that sufficient channels have valid data, the recorder is remotely triggered. When the recorder tachometer indicates proper speed, or a 5-s default time is reached, insuring that proper tape speed has been reached, the 0.3 s of digital data for all channels are recorded on the tape. The digital data are encoded for the FM tape using a digital standard nonreturn- to-zero (NRZ) signal clocked out at 40,000 Hz (fig. 2). Since each channel was sampled using a 12-bit-wide word 24,576 times, 294,912 bits must be written to the tape. The write time is Sample NRZ data stream r NOTE : NRZ format requires the dc levels of FM record. Bi L format would have to be used with direct record FIGURE 2. - NRZ recording signal format. then about 7.5 s. Each channel is written to its respective tape channel simultaneously. If the tape is initially stopped, the total time to accelerate to speed, write data, and decelerate to a stop is about 12 s. If a second tran- sient or group of transients occurs be- fore the tape is stopped, the transfer of data will, of course, occur in a much shorter time. The deadtime for an event lasting longer than the 0.3 s of memory is essentially equal to the 7.5-s write time. The 14th channel is used to record the time of day and for providing clock pulses for later use by the output digital-to-analog converter. The time of day is written on the tape in a pulse- coded magnetic (PCM) format. The accu- racy of this clock is ±1/2 s. The over- all format is custom-designed for the needs of this system. The general opera- tion of the recording side of the tape- triggering system is to digitize and record 13 channels of analog data. Fre- quency may be limited to 10,000 or 1,000 Hz at the user's option. The worst case situation occurs with a 10,000-Hz signal. Even here eight samples are taken per cycle, insuring good analog resolution. The system is designed to record tran- sients, which appear on two to five chan- nels minumum, as selected by the user, during the memory time window so that spurious recording is eliminated. The system is designed to record only valid data and pack them on the tape as densely as possible by operating the tape recorder only when good data have been received. A pretrigger history is re- tained so that signals may be studied prior to the point that they exceed the threshold. The time of day is also recorded with each valid event so that the time history of the tape is pre- served. About 100 events can be stored on a 7,200-ft reel of tape. REPRODUCING SECTION Once the tape has been filled, or the desired period of monitoring is com- pleted, the output section of the tape triggering and recording system may be used to reproduce the digitally recorded signals in analog form. The basic func- tion of the output section is to read the digital signals from the tape and convert them to an analog output. The timing signals on channel 14 are used to control the digital-to-analog conversion rate so that the reproduced data look exactly like the recorded data. Fluctuations in tape speed or head skew will thus not degrade the data. The flexibility of the reproduction speeds on the tape recorder can be used to scale the analog output to match the speed of the recording device. One must, of course, keep track of these scaling factors so the proper time scale for the output records is known. The reproducing section uses compar- ators to read the serial bit stream for each channel and shift registers to assemble these bits into 12-bit words. As each 12-bit word is assembled, it is coupled to a digital-to-analog converter. After appropriate conditioning the analog signal for each channel is sent to its respective output connector, where it may be used to drive recording pens or oscillograph galvanometers. The output levels are ±10 V maximum. The present calibration factor is 1 V into the re- cording section equals 1.41 V out of the reproducing section. The pulse amplitude modulation type PCM format of the 14th channel contains a bit clock to operate the serial shift registers in the data channels, a word clock at one-twelfth the bit clock rate to operate the output latches, and the time of day information (fig. 3). The time of day is decoded and presented in a manner that is easy for the operator to read from graphic rec- ords. The output amplifier on the clock channel has only three possible states: -2 V, V, or +2 V. Positive excursions are read as binary coded data (BCD) ones, and negative excursions as BCD zeros. The overall time of day output as seen on a graphic recorder displays the hours, minutes, and seconds of the occurrence (fig. 4). The capacity of the tape system is being extended by implementing a biphase (Bi(j)L) encoding system as an option. This allows the use of direct recording, which means the tape can be run at much lower speeds for a given bandwidth. The NRZ data require the direct current (dc) levels provided by FM recording. By con- verting to biphase, the slower tape speed (15 ips) will at least quadruple the num- ber of events that can be recorded per tape. The specific parameters for the device discussed in this paper were chosen with a specific microseismic recording problem in mind (2^). Most of these parameters are readily variable to suit other appli- cations. The parameters were also re- stricted by the fact that the Bureau's recorders were already equipped with FM 40-kHz reproduce clock Example clock_ digit = 9 I0 PCM channel ;o n p__m ra r V The PCM signal goes negative every 12 th clock pulse FIGURE 3. - Timing channel format. [o I i oj (o Seconds Minutes Hours Time = 17 h 34min 26s FIGURE 4. - Time of day information as presented on output record. electronics. There are obvious trade- offs between bandwidth and recording time. Total memory and prehistory memory length are obviously variable. By using direct-record electronics the record bandwidth could be greatly increased, though the coding would have to be changed to biphase since dc levels would be lost. Using direct record it is con- ceivable that a system with no dead time between events could be designed. The current system was designed to take ad- vantage of recorders the Bureau already possessed and to maintain compatibility with the normal analog use of these re- corders. The concept could be extended to use magnetic cartridge recorders or disc storage, though the system would have to be significantly modified. ENERGY ANALYZER The energy analyzer (fig. 5) was con- ceived and specified by the Denver Re- search Center of the Bureau of Mines. Design and construction were performed by the Rockwell International Energy Systems Group under a Bureau contract. Testing and final modification were performed by the Denver Research Center. The analyzer processes an externally supplied analog signal. On external command the unit measures the energy in a six-segment fre- quency distribution of the incoming sig- nal. The processed data are then made available to an external minicomputer. ANALOG-TO-DIGITAL CONVERSION The analyzer uses six passive low-pass analog filters to separate the signal into bands of frequency of interest and insures that no signal aliasing occurs. The analog signal is then converted to a digital signal, squared, scaled, and integrated for a selected time period. The external minicomputer-based microseismic system is then used to trigger, read status and data, and reset the analyzer. The unit has a built-in self test. Data are also readable from a front panel light-emitting diode (LED) display. The first stage of signal processing is to filter the analog signal. Filter bandwidths are each from near dc to 10 kHz, 1.6 kHz, 0.8 kHz, 0.4 kHz, 0.2 kHz, and 0.1 kHz. The 10-kHz section is a full-bandwidth filter for the Bureau's FIGURE 5. - Energy analyzer system. microseismic application. The other sec- tions are designed to show possible energy distribution shifts as failure processes progress. The filters roll off at about 24 dB per octave with ultimate attenuation in excess of 36 dB. The 100- Hz filter section will thus attenuate any 200-Hz components in the signal by 24 dB. These attenuations are obviously not infinite, but they are sufficient to sep- arate significant energy and for fre- quency shifts for the microseismic appli- cations for which they were built. The outputs of the analog filters are alternating-current-coupled to an analog multiplexer with sample and hold. Each of the six analog filter outputs is then converted to a digital signal using a 10- bit analog-to-digital converter. The sample rates of the highest to lowest frequency filter output are 2 13 , 2 12 , 2 11 , 2 1 ", 2 9 , and 2 8 samples per 0.1 s. Considering the filter bandwidth, it can then be seen that the frequencies in each filter section are sampled at the rate of 25.6 samples per cycle. The only excep- tion to this is the full bandwidth sec- tion, which is sampled at 8.2 samples per cycle at its upper end. This timing, which at first may seem awkward, has a very logical basis. The basic system clock is 163,840 Hz so each of the sample rates can be obtained by a simple divi- sion by 2. The highest frequency channel gets sampled every 2d clock pulse (81,920 Hz), the next highest every 4th clock pulse, and this continues to form the previously mentioned sample rate. Simi- larly the integration segments can be normalized for direct comparison by sim- ply shifting left. The rates are chosen this high so that the sampling interval is very short, making integrals ob- tained by simple addition very good representations of exact integrals. The sample and hold can acquire a 10-V signal change in 350 ns to 0.01-pct accuracy. The analog-to-digital converter used in this system is a 12-bit unit, although only the 10 most significant bits are used in the calculation section. The signals from the filters are prescaled using voltage dividers so that the front panel input range is ±12 V. The maximum total throughput time of the analog- to-digital converter is 4 ys so it has no difficulty in maintaining the proper con- version rate for all six signal sections. The 10 bits that are used give a unipolar dynamic range of about 54 dB (9 data bits plus sign) . This dynamic range and the real-time conversion speed are maintained throughout the subsequent signal process- ing. If extending the dynamic range were desirable, an obvious way to accomplish this would be to precision-rectify the analog signal and use the converter in a unipolar fashion. The sampling scheme maintains a constant sample time to sig- nal period (AT/T) ratio (fig. 6) at the band limit. The highest frequency in each filter section is sampled at the same rate. For this system this sampling method offers no real advantage other than ensuring that the higher frequency sections get sampled adequately. A dis- advantage is that a signal whose fre- quency is low enough to allow it to pass through two or more filter sections will be sampled at diffferent rates in each section. This might yield slightly dif- ferent energy results for the same sig- nal. This problem is minimized because of the very high sample rates used in this system. If a set of bandpass fil- ters had been used to break the spectrum into nonoverlapping segments that were FIGURE 6. - Waveform digitization. not continuous, the constant AT/T sam- pling would be necessary to ensure equal accuracy for all segments. ARITHMETIC PROCESSING The next step in the energy processing sequence is to square each digitized val- ue. The original concept of the Bureau was to use one of the high-speed multi- plier integrated circuits that have recently become commercially available. Rockwell's approach was to use a program- able read only memory (PROM) look-up table. Both approaches are capable of the speed and dynamic range required. The Bureau's approach, however, offered much more flexibility since changes in word length and other system parameters could be easily accommodated. Using the Bureau's approach, the overall system operation would be simplified and accu- racy improved as well. The Rockwell scheme is complex and can perhaps best be understood by referring to table 1. Rockwell uses the 10-bit analog- to-digital output to form the address selector for the PROM's. The PROM's serve as a look-up table for the squared values. Each unique 10-bit analog-to- digital output has a corresponding squared value stored in PROM. The table uses minus voltages for illustration since the plus voltages are represented by their binary complement because of the sign bit. Rockwell chose to use the squared value rounded to the 16 highest bits in the PROM look-up table. The largest squared value that can be obtained with a 9-bit number is 18 bits, so some accuracy is sacrificed by rounding the 2 lowest bits. The Bureau's initial application calls for using only the 24 most significant bits of the 32-bit integrator output downstream at the squaring PROM's. The 2 least significant bits of the square then have no contribution to these 24 highest bits. The actual error in accuracy as a percent of full scale for typical signal levels is also insignificant in terms of a percentage of true value. A 0.4-V root root mean square (RMS) signal would be in error about 2 pet over a typical inte- gration. It should, however, be clearly understood that this error is not inher- ent in this energy-measuring concept. It is an artifact of design decisions. If the Bureau's original concept of a 20 or more bit high-speed multiplier were used, or another 1,024.x 4-bit PROM were added, this error would be eliminated. After being processed by the squaring PROM's, the digital value is shifted left to normalize it to compensate for the different sample rates. Each sample interval-amplitude level pair defines a rectangular box, which approximates the signal increment for each AT (fig. 6). Shifting the signals makes the rectangu- lar box width equal for all filter sec- tions. Mathematically each shift is a division by 2. For example, the 1,600-Hz section (AT = 24.4 ys) values are shifted once to provide an effective AT of 12.2 ys. Likewise, the 100-Hz section (AT = 390.6 ys) is shifted five times to provide an effective AT of 12.2 ys. Another way of looking at this is to realize that the 1,600-Hz section is sampled four times as often as the 100-Hz section so each of its samples is weighted at one-fourth the value of a 100-Hz sample as it is added to form the integral approximation of its respective section. This makes the energy numbers from the various sections directly com- parable. For example, the energy numbers TABLE 1. - Internal energy processing J Input voltage Analog-to-digital (sign bit removed) Actual squared value PROM output 12 ■8.49.. ■2.34.. -.234.. 11111111(511 10 ) 101101001(361 10 ) 001100100(100 10 ) 000001010(10 10 ) 11111111000000000(261,121 10 ) 11111110100010001(130,321 1(J ) 10011100010000(10,000) 0001100100(100 10 ) 1111111100000000(65,280) 111111101000100(32,580) 100111000100(2,500) 00011001(25 10 ) Table is not exact because of rounding, from each section for a signal of low enough frequency to pass through all sec- tions will all be nearly equal, though the signal was actually sampled at dif- ferent rates in each section. The resultant 21-bit numbers from the shifter are then simply added for each respective filter section to provide the approximate integral of the squared sig- nal (fig. 7). The maximum integration period selectable is 1.5 s. The adder is 32 bits wide to insure that no overflow will occur. Assume that the input was a sine wave at 6,000 Hz, 12 V peak-to-peak, and the integration period was 1.5 s. A sine wave has the property that the mean squared value is one half the peak value. The analog-to-digital converter would be full scale at the peaks. The average output of the squaring PROM would be 216/2. The total number of samples of this average size would be 122,880. The output of the adder would then be 4,026,531,840, which is less than the full-scale output of the adder (2 32 ). The integration period, which can range from 0.1 to 1.5 s in 0.1-s increments, is selectable from front panel switches. The Bureau uses only the 24 most sig- nificant bits of the adder output in the present microseismic application, as was mentioned in the squaring PROM dis- cussion. Again it should be noted that this is not a limitation of the accuracy of the concept, only a design decision. APPLICATION The Bureau's Denver Research Center uses PDP 8/e data processors 3 in its rock burst-coal bump monitors. Since these processors have a 12-bit-wide bus, the upper 24 energy number bits are con- veniently read in 2 words. The rock burst monitor processes data from 24 channels, each of which is de- rived from a geophone at a unique loca- tion. One of these geophones is selected ■^Reference to specific products does not imply endorsement by the Bureau of Mines. to be the energy channel. Its output is split between the rock burst monitor and the energy analyzer. Any active channel triggers the rock burst monitor. This trigger signal is then used to trigger the energy analyzer. A complication is introduced into this scheme since the rock burst monitor does not interrupt the computer unless certain valid event cri- teria are met within its logic. This by itself would leave the energy analyzer triggered on an invalid event and even- tually ready to output invalid data. The Bureau modified the energy analyzer's original trigger circuits so that when- ever a trigger occurs the energy analyzer is first reset and then triggered. This ensures that the energy analyzer and the rock burst monitor will both be process- ing the same event. Since the reset operation takes about 40 us, a delay circuit was introduced (fig. 8) so that the energy unit first has time to reset before it is triggered into a processing mode. The Bureau also found it necessary to introduce hysteresis into the energy analyzer trigger circuit to avoid spuri- ous noise triggers. Initial fieldwork using the energy analyzer has begun. The original filters were changed to 6.4, 3.2, 1.6, 0.8, 0.4, and 0.2 kHz for the rock burst monitoring system, since this provided more informa- tion on the spectral distribution of these particular signals. The motivation for this energy analysis stemmed from the promising application to ground failure prediction of the energy analysis done in the development of a roof fall warning system ( _4-_5 ) . Neither the energy analyzer nor the cruder techniques that preceded it have yet been successfully applied to rock burst pre- diction. The rock burst energy data are complicated by distance scaling effects, mine geometry effects, and rapid natural events mixed with manmade events during blasting. The scaling problem is being approached by taking ratios between the outputs of the various sections of the energy analyzer. The event-overlapping problem may be solvable by running the energy analyzer for fixed time intervals, 10 40-Hz clock 40-Hz filter Amplifier 6 V RMS Calibrate Channel Filter.Hz 10,000 I ,600 800 400 200 100 Signal input ± 12V peak Start pulse (§>_ Run 6-channel low- pass filter 10 kHz Sampling, rate, Hz 81,920 40,960 20,480 10,240 5, I 20 2,560 1.6 kHz .8kHz ■4kHz .2kHz .1 kHz Seguencer B Multiplexer address 2 Data 3 acguisition 4 system 5 Sample strobe I63,840-Hz clock 9 bits ■+ sigr Start switch Step switch — ' Reset switch Status out (reset) : from CPU Data out accepted : by CPU Timer selector switches Timer and control logic Sguaring PROMs Seguencer A Shift code LED panel ready display 16 bits (rounded) 0-to 5-bit parallel shifter \ Arithmetic section 21 bits 32-bit adder LED panel address display 32 bits k- 32 bits 6-by-32-bit register TrJ state out Write [Bits 21 to 32 (most significant word) Bits 9 to 20 Tristate drivers t->!2 bits data out Data status to CPU Word address code 4 BCD time code 4 -> Status in LED panel data display Status selector multiplexer Shift I 2 3 4 5 Status 12-15 to CPU (central processor unit) FIGURE 7. - Energy analyzer arithmetic processing. 11 DATA (37)_ LT |5 ACCEPTEDOUTL STATUS OUTL ~i — rso^! FIGURE 8. - Energy system circuit modifications. similar to the approach used in the roof higher quality data than were previously warning system, instead of trying to available, but research must be done to establish a separate energy for each maximize its benefit, event. The energy analyzer provides CONCLUSIONS TAPE-TRIGGERING SYSTEM overall performance are exceeded by other transient recording devices, the device A tape triggering and recording device is a significant advance in the state of has been designed, constructed, and the art in terms of the number of data tested. While individual parts of the channels, the data bandwidth, memory 12 length, transient capture, and digital record-reproduce technology as grouped in one instrument (_b) . Indeed, before re- cent advances in performance and reduc- tions in cost of electronic circuits such as solid state memories, this device would have been impractical for indus- trial use (6). Since the trend in the electronics industry is toward a continu- ation of cost-performance improvement, the cost of similar units should drop while performance increases. Since the taped data are digitized, digital pro- cessing could be used as an alternate to, or in conjunction with, the present ana- log output. This device should have application in a wide variety of commercial recording problems. ENERGY ANALYZER The energy approximation produced by the energy analyzer is superior to those produced by any other devices the Bureau has used or examined. The hybrid digital technique of voltage-to-frequency conver- sion, for example, is limited because its frequency output varies with the ampli- tude of the voltage. Since the frequency output also indirectly sets an effective sample rate, each amplitude level is integrated with a differing degree of accuracy. At low amplitude, high signal frequency, the output can be a poor representation of the input integral. Analog techniques suffer from drift, lim- ited accuracy, and limited dynamic range. The all-digital approach used in the energy analyzer is a significant step forward in real-time energy processing. It is, however, recognized that at fre- quencies less than about 40,000 Hz, fast Fourier transform analyzers would provide a more detailed real-time energy analy- sis. The concept of the energy analyzer was to solve, in principle, the problem of real-time energy estimation throughout the range of frequencies used by acoustic emission researchers at a reasonable cost. With the present availability of video frequency analog-to-digital con- verters and very high speed digital mul- tipliers, it is felt that this has been accomplished. The motivation for a con- cise, few-number energy estimate stems from the Bureau's research in microseis- mic roof fall warning systems ^4-_5 ) • It has been found that having an approximate energy figure may be a key to being able to assess the stability of the rock mass. Often, for the Bureau's real-time stabil- ity assessments, a complete spectrum would be too much information. It is desirable to have a simple, compact means of presenting energy information to the on-site engineer. The energy analyzer and the principles it embodies provide a means for obtaining a good energy esti- mate that satisfies these criteria. REFERENCES 1. Harris, D. D. , and R. L. Bell. The Measurement and Significance of Energy in Acoustic Emission Testing. Exp. Mech. , Sept. 1977, pp. 347-352. 2. Sites, K. R. , and L. A. Millonzi. Tape Recording and Triggering Systems (contract HO282026, Science Applications, Inc.). BuMines OFR 116-81, 1980, 14 pp.; NTIS PB 81-243248. 3. Blake, W. , F. Leighton, and W. I. Duvall. Microseismic Techniques for Mon- itoring the Behavior of Rock Structures. BuMines B 665, 1974, 65 pp. 4. Fisher, C, Jr. Microseismic Roof Fall Warning System Development. Field -U.S. GOVERNMENT PRINTING OFFICE: 1983-705-020/96 Trials and Commercial Prototype Fabrication. Volume II. Appendix C: Mine Data Collection Summary. Final Report (contract H0272029, Integrated Sciences Inc.). BuMines OFR 163(2)-81, 1980, 228 pp.; NTIS PB 82-137845. 5. Steblay, B. J. Progress in the Development of a Microseismic Roof Fall Warning System. Paper in Tenth Annual Institute on Coal Mining Health Safety and Research (Blacksburg, VA, Sept. 5-8, 1979). Virginia Polytechnic Institute, 1979, pp. 177-195. 6. Nelson, R. Storage Oscilloscopes. Electronic Design News, June 10, 1981, pp. 76-88. INT.-BU.OF IES,PGH.,PA. 27222 t.% y.^fe«x y.-afc^ yx&X. *^xy.^% *>• ■>". .•->, ^\ %/X»\ \/ •'£&'' \X.'«^ \„/ :]»•. V ,/%. & V"X '. . X cF t o"'* "*b .A -*' ^ k^°* ^IIP/ X°X, W$^i &*°* - v •:,*»- X . V • »••» X a? >'..!•: .«!. *<< \ V<^ A *o.* a «G V \9, *'.<,Y S A X&- % *•;&*.* X>^iX ^° -^<.°- ,A X*^>V %x„. . % <^. .A L> ^ ^ / A^X o, ♦'TTi* A «^ > %, *^f** A J -.y. v T4 *y ^ * <> *••»• aG ;\ ^-. %># t« "oV * ^ » ^ -■!!?•./• %.*?z&\& v^^v./ "V'Js^v __ v< \y :jflK-. ^,# ft -^^ ^ # -^Kx* + Wws J* \ 'SK* ^ ^ °w # ^ - -*8Kx** ^ °Ww ,^ V *^:t* a <^ *•.** v ^ ^t* «g* V *??7?« A ^ '*•* a° • ^ j ^ * O H ►*x •v » bJ^ * - : ; ^ ^ -if/ . ^ ^ . mm : y\ j -mes .^%n # ^ v X.^ * rt 4y . * *> c°*..555S:^o >*"vJ •- X. ^ ;*. 'bV 3s? VV x^g^j^X - ""- • .^^ r X.^ •* dBlMiUtaialii ■ ■ I