`Case 6:12—cv—00799—JRG Document 71-2 Filed 09/13/13 Page 1 of 20 Page|D #: 1863
`
`EXHIBIT 2
`
`EXHIBIT 2
`
`
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 2 of 20 PageID #: 1864
`
`(12) United States Patent
`Henrot
`
`US006505131B1
`(10) Patent N0.2
`US 6,505,131 B1
`(45) Date of Patent:
`Jan. 7, 2003
`
`(54) MULTI-RATE DIGITAL SIGNAL
`PROCESSOR FOR SIGNALS FROM PICK-
`OFFS ON A VIBRATING CONDUIT
`
`.
`-
`~
`~
`(75) Inventor‘ Dems Henmt’ Loulsvlne’ CO (Us)
`
`5,555,190 A * 9/1996 Derby et a1. ................ .. 702/45
`5,583,784 A * 12/1996 Kapust et a1. .............. .. 702/77
`5,734,112 A
`3/1998 Bose et 211.
`5,741,980 A * 4/1998 Hill et a1. .............. .. 73/861.04
`5,926,096 A * 7/1999 Matter et a1.
`...... .. 340/606
`6,233,529 B1 * 5/2001 Cunningham .............. .. 702/76
`
`(73) Assignee: Micro Motion, Inc., Boulder, CO (US)
`
`FOREIGN PATENT DOCUMENTS
`
`( * ) Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(1)) by 0 days.
`
`0 866 319 A1
`EP
`1 219 887
`GB
`* Cited by examiner
`
`9/ 1998
`1/1971
`
`(21) APPL NO; 09/344 840
`’
`Jun. 28, 1999
`
`(22) Filed:
`
`Primary Examiner—Marc S. Hoff
`Assistant Examiner—Mohamed Charioui
`(74) Attorney, Agent, or Firm—Faegre & Benson LLP
`
`(51) Int. Cl.7 .............................................. .. G01F 17/00
`
`(57)
`
`ABSTRACT
`
`(52)
`(58) Field of Search
`702/70
`’
`
`(56)
`
`.. 702/54; 702/45; 73/861.35
`702/45 50 66
`357 ’ 861’ 05’
`861 06’ 86169, 324/58 5’
`'
`’
`'
`’
`'
`References Cited
`
`’
`
`’
`
`U-S- PATENT DOCUMENTS
`4 066 881 A * 1/1978 Houdard
`4:996:871 A
`3/1991 Romano
`5,321,991 A * 6/1994 Kalotay ................. .. 73/861.35
`5,361,036 A * 11/1994 White ...................... .. 329/361
`
`708/405
`
`_
`_
`_
`_
`_
`A digital signal processor for determining a property of a
`material ?owing through a conduit. The digital signal pro
`cessor of this invention receives signals from tWo pick-off
`sensor mounted at tWo different points along a How tube at
`a ?rst sample rate. The signals are converted to digital
`signals. The digital signals are decimated from a ?rst sample
`rate to a desired sample rate. The frequency of the received
`signals is then determined from the digital signals at the
`deslred Sample rate
`
`26 Claims, 7 Drawing Sheets
`
`100
`
`DRIVE SIGNAL
`
`1 1 O
`
`LEFT VELOCITY SIGNAL
`
`METER
`ELECTRONICS
`
`W .
`
`W ,
`
`101
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 3 of 20 PageID #: 1865
`Case 6:12—cv—OO799—JRG Document 71-2 Filed 09/13/13 Page 3 of 20 Page|D #: 1865
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 1 of 7
`
`US 6,505,131 B1
`
`20
`
`FIG.1
`
` IELECTRONICS
`
`
`RIGHTVELOC|TYSIGNAL
`
` D
`
`
`
`RIVESIGNAL
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 4 of 20 PageID #: 1866
`Case 6:12—cv—OO799—JRG Document 71-2 Filed 09/13/13 Page 4 of 20 Page|D #: 1866
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 2 of 7
`
`US 6,505,131 B1
`
`(0
`N
`
`N
`L5‘
`'-'~
`
`§
`
`53.N
`
`3
`:3:
`
`2 3
`
`:’
`
`§
`
`|:
`
`E5L
`
`L]
`2
`%
`
`5CU
`
`E0.1
`
`e
`N
`
`5
`N
`
`§
`
`(0
`LLI
`C.)
`
`O1 E
`
`L
`
`5
`N
`
`(\
`8
`
`9
`N
`
`to
`O
`
`:9
`N
`
`E3
`N
`
`N
`5
`
`3
`D
`
`<:
`
`<
`
`111
`
`111'
`
`110
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 5 of 20 PageID #: 1867
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 3 0f 7
`
`US 6,505,131 B1
`
`300 1
`
`FIG. 3
`
`+
`
`301 —/\
`
`GENERATE DRIVE
`SIGNAL
`
`l
`
`RECEIVE SiGNALS
`302w FROM PICK-OFF SENSORS
`
`l
`
`303$
`
`GENERATE DATA
`ABOUT SIGNAL
`
`304f
`
`CALCULATE PROPERTIES
`OF MATERIAL
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 6 of 20 PageID #: 1868
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 4 0f 7
`
`US 6,505,131 B1
`
`400 \
`
`FIG. 4
`
`401/“
`
`DECIMATE SAMPLES OF SIGNALS
`FROM FIRST SAMPLE RATE TO AN
`INTERMEDIATE SAMPLE RATE
`
`I
`
`ESTIMATE FREQUENCY OF
`SIGNALS FROM SAMPLES
`
`I
`
`DEMODU LATE
`SIGNALS
`
`402w
`
`403/“
`
`I
`
`DECIMATE SIGNALS FRoM THE
`404/‘ INTERMEDIATE SAMPLE RATE TO A
`DESIRED SAMPLE RATE
`
`V
`DETERMINE FREQUENCY
`OF SIGNALS
`
`405J‘
`
`I
`
`DETERMINE PHASE DIFFERENCE
`BETWEEN SIGNALS
`
`I
`
`DETERMINE AMPLITUDE
`OF SIGNALS
`
`406f
`
`4°7f
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 7 of 20 PageID #: 1869
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 5 0f 7
`
`US 6,505,131 B1
`
`500 1
`
`FIG. 5
`
`@
`
`RETRIEVE M SAMPLES OF
`SIGNALS TO CREATE
`STATE VECTOR
`I
`MULTIPLY STATE VECTOR
`502f
`BY VECTOR OF STATE INPUT
`TO DETERMINE A ?Ih SAMPLE
`I
`OUTPUT THE RESULT
`REPRESENTING EVERY
`Mlh SAMPLE
`
`503/“
`
`V CE)
`
`600 1
`
`601f
`
`602
`
`604f
`
`DEMULTIPLEX SIGNAL INTO
`I COMPONENT SIGNAL AND
`O COMPONENT SIGNAL
`I
`INTEGRATE EACH
`COMPONENT SIGNAL
`I
`603 f CALCULATE COMPENSATION
`FOR EACH COMPONENT SIGNAL
`I
`COMBINE COMPONENT SIGNALS
`TO GENERATE DIGITALLY
`INTEGRATED SIGNAL
`I
`CALCULATE RATIO BETWEEN
`ORIGINAL SIGNAL AND DIGITALLY
`INTEGRATED SIGNAL TO ESTIMATE
`FREQUENCY
`
`FIG. 6
`
`@
`
`GI;
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 8 of 20 PageID #: 1870
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 6 0f 7
`
`US 6,505,131 B1
`
`700 1
`
`FIG. 7
`START
`)
`
`(
`
`V
`
`702/‘
`
`f CALCULATE PULSATION OF
`701
`NORMALIZED SIGNAL
`T
`CALCULATE TWIDDLE
`FACTOR Wk
`T
`CALCULATE sATD DOT
`PRODUCT OF THE TW|DDLE
`FACTOR AND A RECEIVED SIGNAL
`
`7
`O3
`
`V
`END
`
`YES
`
`FREQUENCY
`GREATER THAN REFERENCE
`FREQUENCY
`'?
`
`NO
`
`v
`
`DETERMINED FREQUENCY :
`FREQUENCY — 250
`
`v
`
`DETERMINED
`FREQUENCY : 0
`
`802
`
`803
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 9 of 20 PageID #: 1871
`
`U.S. Patent
`
`Jan. 7, 2003
`
`Sheet 7 0f 7
`
`US 6,505,131 B1
`
`FIG. 9
`
`900 \
`
`901f CALCULATE ADAPTATION OF
`NOTCH FILTER PARAMETER
`I
`902 f‘ CALCULATE FREQUENCY
`FROM ADAPTATION
`I
`903 f PERFORM COMPLEX
`DEMODULATION OF SIGNALS
`I
`904 f PERFORM DECIMATION
`ON SIGNALS
`I
`PERFORM COMPLEX
`905/‘
`CORRELATION OF LEFT
`PICK-OFF SIGNAL WITH RIGHT
`PICK-OFF SIGNAL
`I
`DETERMINE PHASE
`DIFFERENCE BETWEEN
`SIGNALS
`
`906f
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 10 of 20 PageID #: 1872
`
`US 6,505,131 B1
`
`1
`MULTI-RATE DIGITAL SIGNAL
`PROCESSOR FOR SIGNALS FROM PICK
`OFFS ON A VIBRATING CONDUIT
`
`FIELD OF THE INVENTION
`
`This invention relates to a signal processor for an appa
`ratus that measures properties of a material ?owing through
`at least one vibrating conduit in the apparatus. More
`particularly, this invention relates to a digital signal proces
`sor for performing calculations to determine the frequencies
`of signals receive from pick-off sensors measuring the
`frequency of vibrations of the conduit.
`
`Problem
`
`It is knoWn to use Coriolis effect mass ?oWmeters to
`measure mass How and other information for materials
`?oWing through a conduit in the ?oWmeter. Exemplary
`Coriolis ?oWmeters are disclosed in US. Pat. Nos. 4,109,
`524 of Aug. 29, 1978, US. Pat. No. 4,491,025 of Jan. 1,
`1985, and US. Pat. No. Re. 31,450 of Feb. 11, 1982, all to
`J. E. Smith et al. These ?oWmeters have one or more
`conduits of a straight or a curved con?guration. Each
`conduit con?guration in a Coriolis mass ?oWmeter has a set
`of natural vibration modes, Which may be of a simple
`bending, torsional or coupled type. Each conduit is driven to
`oscillate at resonance in one of these natural modes. Material
`?oWs into the ?oWmeter from a connected pipeline on the
`inlet side of the ?oWmeter, is directed through the conduit or
`conduits, and exits the ?oWmeter through the outlet side of
`the ?oWmeter. The natural vibration modes of the vibrating,
`material ?lled system are de?ned in part by the combined
`mass of the conduits and the material ?oWing Within the
`conduits.
`When there is no How through the ?oWmeter, all points
`along the conduit oscillate due to an applied driver force
`With identical phase or small initial ?xed phase offset Which
`can be corrected. As material begins to ?oW, Coriolis forces
`cause each point along the conduit to have a different phase.
`The phase on the inlet side of the conduit lags the driver,
`While the phase on the outlet side of the conduit leads the
`driver. Pick-off sensors are placed on the conduit(s) to
`produce sinusoidal signals representative of the motion of
`the conduit(s). Signals outputted from the pick-off sensors
`are processed to determine the phase difference betWeen the
`pick-off sensors. The phase difference betWeen tWo pick-off
`sensor signals is proportional to the mass ?oW rate of
`material through the conduit(s).
`A Coriolis ?oWmeter has a transmitter Which generates a
`drive signal to operate the driver and determines a mass ?oW
`rate and other properties of a material from signals received
`from the pick-off sensors. A conventional transmitter is
`made of analog circuitry Which is designed to generate the
`drive signal and detect the signals from the pick-off sensors.
`Analog transmitters have been optimiZed over the years and
`have become relatively cheap to manufacture. It is therefore
`desirable to design Coriolis ?oWmeters that can use con
`ventional transmitters.
`It is a problem that conventional transmitters must Work
`With signals in a narroW range of operating frequencies. This
`range of operating frequencies is typically betWeen 20 HZ
`and 200 HZ. This limits the designers to this narroW range of
`operating frequencies. Furthermore, the narroW range of
`operating frequencies makes it impossible to use a conven
`tional transmitter With some ?oWmeters, such as straight
`tube ?oWmeters, Which operate in a higher frequency range
`
`2
`of 300—800 HZ. Straight tube ?oWmeters operating at
`300—800 HZ tend to exhibit smaller sensitivity to Coriolis
`effects used to measure mass ?oW rate. Therefore, a ?ner
`measurement of the phase difference betWeen sensors
`is-needed to calculate mass ?oW rate.
`In order to use one type of transmitter on several different
`designs of Coriolis ?oWmeters operating at several different
`frequencies, manufacturers of Coriolis ?oWmeter have
`found that it is desirable to use a digital signal processor to
`generate the drive signals and process the signals received
`from the pick-off sensors. A digital signal processor is
`desirable because the higher demand in measurement reso
`lution and accuracy put on analog electronic components by
`?oWmeters operating at higher frequencies, such as straight
`tube designs, are avoided by the digitaliZation of signals
`from the pick-offs as the signals are received by the trans
`mitter. Furthermore, the instructions for signaling processes
`used a digital processor may be modi?ed to operate on
`several different frequencies.
`HoWever, digital signal processors have several disadvan
`tages as compared to conventional analog circuit transmit
`ters. A ?rst problem With a digital signal processor is that
`digital processors are more expensive to produce because
`the circuitry is more complex. Secondly, digital signal
`processors require a circuit board having a greater surface
`area Which can cause problems When space is at a premium
`in a ?oWmeter design. Thirdly, digital signal processors
`require more poWer to operate than analog circuits. PoWer
`consumption is especially a problem When a processor must
`operate at a maximum clock rate in order to provide all the
`computations needed to process the signals and generate a
`material property measurement, such as mass ?oW. For all of
`these reasons, there is a need in the art for a digital signal
`processor that is adaptable across several ?oWmeter designs,
`that is inexpensive to produce and reduces the amount of
`poWer needed to perform the needed computations.
`Solution
`The above and other problems are solved and an advance
`in the art is made by the provision of a multi-rate digital
`signal processor of the present invention. The present inven
`tion is comprised of processes that are stored in a memory
`and executed by the processor in order to process the signals
`received from pick-offs on a vibrating conduit. The pro
`cesses of this invention offer many advantages that make it
`viable to use a single type of digital signal processor in many
`types of Coriolis ?oWmeters.
`A ?rst advantage of the processes of the present invention
`is that the processes do not lose accuracy in spite of using
`?nite arithmetic in lieu of ?oating point arithmetic. Asecond
`advantage of the processes of the present invention is that
`the processes can be implemented on any number of loW
`cost, loW poWer digital signal processors such the Texas
`Instruments TM3205xx, Analog Devices ADSP21xx, or
`Motorola 5306x. The instructions for the processes of the
`present invention are small enough to reside in the internal
`memory of a digital signal processor Which eliminates the
`need for fast access external memory Which increases the
`cost, poWer consumption and board space for the transmitter.
`The processes have a small number of computational con
`structs Which improves the portability of the processes
`betWeen loW cost processors.
`Athird advantage is that the computational requirement of
`the processes is minimiZed. This reduction in the computa
`tional requirement alloWs the digital signal processor to run
`at a clock rate loWer than the maximum clock rate of the
`processor Which reduces the poWer consumption of the
`processor.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`55
`
`60
`
`65
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 11 of 20 PageID #: 1873
`
`US 6,505,131 B1
`
`10
`
`15
`
`25
`
`3
`A transmitter that performs the processes of the present
`invention has the following electronic components. Analog
`signals from the pick-offs attached to the sensors are
`received by an Analog to Digital (“A/D”) converter. The
`converted digital signals are applied to a standard digital
`processor. The digital processor is a processing unit that
`executes machine readable instructions that are stored in a
`memory connected to the processor via a bus. A typical
`digital processor has a Read Only Memory (ROM) Which
`stores the instructions for performing desired processes such
`as the processes of the present invention. The processor is
`also connected to a Random Access Memory Which stores
`the instructions for a process that is currently being executed
`and the data needed to perform the process. The processor
`may also generate drive signals for the Coriolis ?oWmeter.
`In order to apply the drive signal to a drive system, a digital
`processor may be connected to a Digital to Analog (D/A)
`convertor Which receives digital signals from the processor
`and applies analog signals to the drive system.
`The processes of the present invention perform the fol
`loWing functions to determine the frequencies of the signals
`received from the pick-off sensors as Well as the phase
`difference betWeen the signals. First, the signals are received
`from the pick off sensor at a ?rst sample rate. A sample rate
`is the amount of inputs received from the pick-offs that are
`used to characteriZe the signals from the pick-offs. The
`signals are then decimated from a ?rst sample rate to a
`desired sample rate. Decimation is simply converting from
`a ?rst number of samples to a lesser number of samples.
`Decimation is performed to increase the resolution of the
`signals sampled to provide a more precise calculation of
`signal frequency for each signal. The frequency of each
`signal is then determined.
`In order to use the same processes With different ?oW
`meters having different frequencies, the folloWing steps may
`also be performed. An estimate of the oscillation frequency
`of the ?oWmeter is calculated. The estimated frequency is
`then used to demodulate the signals from each pick-off into
`an I component and a Q component. The I component and
`the Q component are then used to translate the signals to a
`center frequency if the operating frequency of the signals is
`greater than a transition frequency. After demodulation, the
`signals may be decimated a second time to improve the
`resolution of the signals a second time.
`The dominant frequency of the signals is then isolated and
`precisely measured. The translation to a Zero frequency is
`then calculated for both the I component and Q components
`of the signals. At this time, each component may decimated
`again to improve the accuracy of measurement. The fre
`quency band of each signal can be narroWed as much as
`desired by appropriate loW pass ?ltering at this time. A
`complex correlation is then performed Which determines the
`phase difference betWeen the signals.
`The above process alloWs a loW poWer, loW cost, digital
`processor to be used in a different types of Coriolis ?oW
`meter Which operate over a Wide range of operating fre
`quencies.
`
`35
`
`45
`
`55
`
`DESCRIPTION OF THE DRAWINGS
`
`The present invention can be understood from the fol
`loWing detailed description and the folloWing draWings:
`FIG. 1 illustrating a Coriolis FloWmeter having a digital
`transmitter that performs multi-rate pick-off signal processes
`of this invention;
`FIG. 2 illustrating a block diagram of a digital signal
`transmitter;
`
`65
`
`4
`FIG. 3 illustrating a How diagram of the operations
`performed by a digital transmitter;
`FIG. 4 illustrating a How diagram a process for generating
`data from signals received from pick-off sensors;
`FIG. 5 illustrating a process for performing a decimation
`of signal samples from a pick-off;
`FIG. 6 illustrating a process of calculating an estimated
`frequency of the signals received from the pick-offs;
`FIG. 7 illustrating a process for performing a high-loW
`frequency selection for the received signals;
`FIG. 8 illustrating a process for demodulating the
`received signals; and
`FIG. 9 illustrating a method for determining data about
`How tube vibration from the received signals.
`
`DETAILED DESCRIPTION
`Coriolis FloWmeter in General—FIG. 1
`FIG. 1 shoWs a Coriolis ?oWmeter 5 comprising a Corio
`lis meter assembly 10 and transmitter 20. Transmitter 20 is
`connected to meter assembly 10 via leads 100 to provide
`density, mass ?oW rate, volume ?oW rate, temperature,
`totaliZed mass ?oW, and enhanced density over path 26. A
`Coriolis ?oWmeter structure is described although it should
`be apparent to those skilled in the art that the present
`invention could be practiced in conjunction With any appa
`ratus having a vibrating conduit to measure properties of
`material. A second example of such an apparatus is a
`vibrating tube densitometer Which does not have the addi
`tional measurement capability provided by a Coriolis mass
`?oWmeter.
`Meter assembly 10 includes a pair of ?anges 101 and 101‘,
`manifold 102 and conduits 103A and 103B. Driver 104 and
`pick-off sensors 105 and 105‘ are connected to conduits
`103A—B. Brace bars 106 and 106‘ serve to de?ne the axis W
`and W‘ about Which each conduit oscillates.
`When ?oWmeter 10 is inserted into a pipeline system (not
`shoWn) Which carries the process material that is being
`measured, material enters meter assembly 10 through ?ange
`101, passes through manifold 102 Where the material is
`directed to enter conduits 103A and 103B, ?oWs through
`conduits 103A and 103B and back into manifold 102 from
`Where it exits meter assembly 10 through ?ange 101‘.
`Conduits 103A and 103B are selected and appropriately
`mounted to the manifold 102 so as to have substantially the
`same mass distribution, moments of inertia and elastic
`modules about bending axes W—W and W‘—W‘, respec
`tively. The conduits extend outWardly from the manifold in
`an essentially parallel fashion.
`Conduits 103A—103B are driven by driver 104 in opposite
`directions about their respective bending axes W and W‘ and
`at What is termed the ?rst out of phase bending mode of the
`?oWmeter. Driver 104 may comprise any one of many Well
`knoWn arrangements, such as a magnet mounted to conduit
`103A and an opposing coil mounted to conduit 103B and
`through Which an alternating current is passed for vibrating
`both conduits. A suitable drive signal is applied by meter
`electronics 20, via lead 110, to driver 104.
`Transmitter 20 receives the left and right velocity signals
`appearing on leads 111 and 111‘, respectively. Transmitter 20
`produces the drive signal appearing on lead 110 and causing
`driver 104 to vibrate tubes 103A and 103B. Transmitter 20
`processes the left and right velocity signals to compute the
`mass ?oW rate and the density of the material passing
`through meter assembly 10. This information is applied to
`path 26.
`It is knoWn to those skilled in the art that Coriolis
`?oWmeter 5 is quite similar in structure to a vibrating tube
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 12 of 20 PageID #: 1874
`
`US 6,505,131 B1
`
`5
`densitometer. Vibrating tube densitometers also utilize a
`vibrating tube through Which ?uid ?oWs or, in the case of a
`sample-type densitometer, Within Which ?uid is held. Vibrat
`ing tube densitometers also employ a drive system for
`exciting the conduit to vibrate. Vibrating tube densitometers
`typically utiliZe only single feedback signal since a density
`measurement requires only the measurement of frequency
`and a phase measurement is not necessary. The descriptions
`of the present invention herein apply equally to vibrating
`tube densitometers.
`A Digital Transmitter 20—FIG. 2.
`FIG. 2. illustrates of the components of a digital trans
`mitter 20. Paths 111 and 111‘ transmit the left and right
`velocity signals from ?oWmeter assembly 10 to transmitter
`20. The velocity signals are received by analog to digital
`(A/D) convertor 203 in meter electronic 20. A/D convertor
`203 converts the left and right velocity signals to digital
`signals usable by processing unit 201 and transmits the
`digital signals over path 210—210‘. Although shoWn as
`separate components, A/D convertor 203 may be a single
`convertor, such as an AK4516 16-bit codec chip, Which
`provides 2 convertors so that signals from both pick-offs are
`converted simultaneously. The digital signals are carried by
`paths 210—210‘ to processor 201. One skilled in the art Will
`recogniZe that any number of pick-offs and other sensors,
`such as an RTD sensor for determining the temperature of
`the ?oW tube, may be connected to processor 201.
`Driver signals are transmitted over path 212 Which applies
`the signals to digital to analog (D/A) convertor 202. D/A
`convertor 202 is a common D/A convertor and may be a
`separate D/A convertor or one that is integrated in a stereo
`CODEC chip such as a standard AKM 4516. Another
`common D/A convertor 202 is a AD7943 chip. The analog
`signals from D/A convertor 202 are transmitted to drive
`circuit 290 via path 291. Drive circuit 291 then applies the
`drive signal to driver 104 via path 110. Path 26 carries
`signals to input and output means (not shoWn) Which alloWs
`transmitter 20 to receive data from and convey data to an
`operator.
`Processing unit 201 is a micro-processor, processor, or
`group of processors that reads instructions from memory and
`eXecutes the instructions to perform the various functions of
`the ?oWmeter. In a preferred embodiment, processor 201 is
`a ADSP-2185L microprocessor manufactured by Analog
`Devices. The functions performed include but are not lim
`ited to computing mass ?oW rate of a material, computing
`volume ?oW rate of a material, and computing density of a
`material Which may be stored as instructions in a Read Only
`Memory (ROM) 220. The data as Well as instructions for
`performing the various functions are stored in a Random
`Access Memory (RAM) 230. Processor 201 performs read
`and Write operations in RAM memory 230 via path 231.
`OvervieW of Operation Performed by Digital Transmitter
`20—FIG. 3.
`FIG. 3 is an overvieW of the functions performed by
`digital transmitter 20 to operate Coriolis ?oWmeter 5. Pro
`cess 300 begins in step 301 With transmitter 20 generating a
`drive signal. The drive signal is then applied to driver 104
`via path 110. In step 302, digital transmitter 20 receives
`signals from pick-off 105 and 105‘ responsive to vibration of
`said ?oW tubes 103 A—B as material passes through ?oW
`tubes 103A—B. Data about the signals such as signal fre
`quency and phase difference betWeen signals is performed
`by digital transmitter 20 in step 303. Information about
`properties of a material ?oWing through ?oW tubes 103A
`and 103B, such as mass ?oW rate, density, and volumetric
`?oW rate, are then calculated from the data in step 304.
`
`10
`
`15
`
`25
`
`35
`
`45
`
`55
`
`65
`
`6
`Process 300 is then repeated as long as Coriolis ?oWmeter
`5 is operating Within a pipeline.
`A Process for Generating Data About the Pick-off Signals in
`Accordance With the Present Invention—FIG. 4.
`FIG. 4 illustrates process 400 Which is a process for
`generating data such as a signal’s frequency for signals
`received from pick-offs 105 and 105‘ that measure the
`oscillations of ?oW tubes 103A—B in Coriolis ?oWmeter 5.
`Process 400 offers several advantages for use in a digital
`transmitter 20. A ?rst advantage of process 400 is that there
`is no loss of accuracy despite the use of ?nite point arith
`metic instead of ?oating point arithmetic. This alloWs a loW
`cost, loW poWer processor such the TMS3205XXX manufac
`tured by TeXas Instrument, The ADSP21XX manufactured by
`Analog Devices, or the 5306X manufactured by Motorola
`Inc. Asecond advantage is that the memory requirement for
`the instructions for process 400 is small enough to reside in
`the internal memory of the processor Which eliminates the
`need for a high speed bus betWeen the processor and an
`eXternal memory. The computational requirements are
`reduced by process 400 Which alloWs the processor to
`operate at substantially less than its maXimum clock rate.
`Process 400 begins in step 401 by decimating the sample
`rate of signals received from the pick-offs from a ?rst sample
`rate to a second, lesser sample rate. In a preferred
`embodiment, the signals are decimated from a ?rst sample
`rate of 48 kHZ to a second sample rate of 4 kHZ. The
`decimation of sample rates increases the resolution of the
`signals Which increases the accuracy of the calculations as
`described beloW in FIG. 5. In the preferred embodiment, the
`reduction of the sample rate from 48 kHZ to 4 kHZ increases
`the resolution of the sample from B bits to B+1.79 bits.
`In step 402, an estimate of the signal frequency is calcu
`lated from the sampled signals. A preferred process is to
`calculate an estimated signal frequency is provided in FIG.
`6. The estimated signal frequency is then used to demodulate
`the received signals in step 403. Aprocess for demodulating
`the digital signals is given in FIGS. 7 and 8. A second
`decimation of the sampled signals is performed in step 404.
`The second decimation reduces the signal samples from a
`second sample rate to a third sample rate to increase the
`resolution of the sampled signal. In the preferred
`embodiment, the reduction is from a sample rate of 4 kHZ to
`a rate of 800 HZ Which increases the resolution to B+2.95
`bits. This reduction is performed in the same manner as the
`decimation in step 401.
`After the second decimation in step 404, calculation are
`made based upon the received signals. This is performed as
`part of a process shoWn FIG. 9. After the noise has been
`removed, the frequency of the signals from each pick-off is
`determined in step 405. In step 406, a phase difference
`betWeen the signal from a ?rst pick-off and the signal from
`a second pick-off is determined. The amplitude of each
`signal is then determined in step 407. Process 400 is then
`either repeated as long as the ?oWmeter is in operation or
`process 400 ends.
`A Process for Decimating Sample Rates of Signals from
`Pick-offs-FIG. 5
`FIG. 5 illustrates a process for decimating the rate of
`samples received from pick-offs. The same process is used
`for the decimation performed in each of steps 401, 404, and
`in the process for determining frequency of the signal. In
`each of these steps, process 500 is performed for signals
`from each pick-off separately. The difference betWeen the
`decimation performed in each step is the length of the input
`data vectors as described beloW.
`A decimation as described in process 500 is implemented
`using a block processing method With the siZe of the input
`
`
`
`Case 6:12-cv-00799-JRG Document 71-2 Filed 09/13/13 Page 13 of 20 PageID #: 1875
`
`US 6,505,131 B1
`
`7
`vector being equal to the decimation ratio. The decimation
`ratio is the amount that the sample frequency is being
`reduced by the decimation. Using this block processing
`method is an operational advantage since the process must
`only be repeated at an output data rate rather than at an input
`data rate. The principle behind recursion block ?ltering is
`that a state variable representation of the signals is:
`
`Where:
`A,B,C,D=matrices representing the state of the system;
`xk=an N+1 state vector at time k;
`uk=an input; and
`yk=an output of representing a decimated signal.
`From induction, it is clear that:
`
`xk+m
`W
`
`Am I AmilB Ami2B
`C
`D
`0
`
`B
`0
`
`Xk
`14k
`
`yk+l :
`
`CA
`
`CB
`
`D
`
`-
`
`O
`
`14,,11
`
`yk+Mil
`
`cAmil cAmi2B cAmi3B
`
`D uk+Mil
`
`When decimating a signal by a factor of M, only every M-th
`sample is going to be kept. Therefore, all but the last output
`roW of the above matrix can be eliminated to yield the
`folloWing equation:
`
`[XMM ]_[ AM IAMAB AMizB
`MM CAM ICAAHB CAME
`
`3]
`D
`
`From the above, it is obvious that the number of accumulate/
`multiply operations for one recursion of the above equation
`is:
`
`10
`
`15
`
`25
`
`35
`
`Where:
`NMAC=number of accumulate/multiply operations;
`N=order of the matrix A; and
`M=the block siZe Which is equal to the decimation rate of
`the process.
`Therefore, the computational load for performing the deci
`mation is
`
`45
`
`RsvfFout NMAC
`
`_
`
`*
`
`Where:
`Rsvd=the computational load on a processor; and
`F0m=represents the ?lter output rate.
`The memory needed to perform a decimation is as folloWs:
`memory to store each ?lter coef?cient Which may be
`read-only (ROM);
`memory to store the ?lter state vector xk Which must be
`read-Write (RAM); and
`an input block buffer memory (read-Write).
`FIG. 5 illustrates the process of decimation using the above
`block processing method. Process 500 begins in step 501 by
`receiving m samples into a buffer to create an input block.
`The input block is then multiplied by the state vector in step
`502. The results outputting every mth sample are then
`outputted in step 503 for use in other calculations. Process
`500 ends after step 503.
`
`55
`
`65
`
`8
`A Process for Estimating the Frequency of the Received
`Signals FIG. 6.
`Process 600 is a process for estimating the frequency of
`the received signals in order to demodulate the signals in a
`subsequent process step. Process 600 must be completed in
`step 403 before the signals can be demodulated. The sub
`sequent demodulation is described beloW and shoWn in FIG.
`7.
`The process 600 for estimating the frequency of the
`signals is shoWn in FIG. 6. This process 600 is performed on
`either one of the received signals. Process 600 begins in step
`601 by demultiplexing the sampled digital signal into an
`In-phase (I) and a quadrature (Q) component. A digital
`integration is then performed on the I component and the Q
`component of the signal in step 602. In step 603, a signal
`compensation is calculated on the integrated signal. The
`signal components are then multiplexed to generate a digi
`tally integrated signal in step 604. The ratio betWeen the
`original signal and the digitally integrated signal is then
`calculated in step 605. The ratio provides an estimate of the
`signal frequency Which may be used to demodulate the
`signals in process 700.
`Process 600 uses ?xed coef?cient ?lters to estimate the
`frequency. Therefore no recursive algorithm is needed.
`Since recursion is not used, process 600 alWays converges.
`Furthermore, process 600 rapidly tracks changes in the
`frequency. The estimated frequency at the end of process
`600 is given by the folloWing equation:
`
`ESE
`
`Where:
`Fest=estimatedfrequency;
`u) =normaliZedpulsation; and
`FS=frequencyofsamples.
`A Process for a High-Low Frequency Selector—FIG. 8
`Process 800 illustrated in FIG. 8 an optional process that
`may be performed betWeen the frequency estimate and
`demodulation of the signal. The high loW frequency selector
`is needed to determine the frequency of interest. Process
`1000 (SEE FIG. 9). Which accurately measures the signal
`frequency, exhibits an estimate bias and sloWer convergence
`in the normaliZed frequency range:
`
`Where:
`F0=normaliZed frequency of signal.
`From this equation, it is apparent that the process for
`determining frequency is not accurate When the sampl