CA2111698C - A technique for determining a mechanical zero value for a coriolis meter - Google Patents

Abstract

Apparatus and ac-companying methods for inclusion in a Coriolis me-ter (5) that substantially eli-minate temperature in-duced measurement errors which might otherwise be produced by performance differences existing be-tween the separate input channels contained in the meter. Specifically, two pairs of input channels (44, 54, 64) are used in the me-ter. In operation, the meter repetitively measures the internal phase delay of each of these pairs and then subtracts the delay associated with each pair from actual flow based measurement data subsequently obtained therefrom. While one channel pair is measuring actual flow, the other channel pair is measuring its internal phase delay, with the channels being continuously cycled between these functions. Because both channel pairs are cycled at a sufficiently rapid rate, the current value of the internal phase delay for each of the pairs accurately reflects any temperature induced changes then occurring in the performance of that pair thereby eliminating substantially all temperature induced error components from the flow measurement subsequently obtained therefrom. In addition, the meter measures flow tube temperature in a manner that re-moves temperature induced errors therefrom. Furthermore, the meter also measures and updates its mechanical zero value using only zero flow .DELTA.t measurements that have a sufficiently low noise content; this mechanical zero value is subsequently subtracted from the flow based measurement data to eliminate errors therein that would otherwise occur.

Description

'0 93/01472 PCI/US92/05583 -1- 2~11698 A IL.~ FOR o~. A MT~rT~NTr~T
ZERO VALUE FOR A CO~TOT TR liETER
BACKGROUND OF TT~ INVF:NTION
1. Field of the Invention The present invention relates to apparatus and methods for inclusion in, illustratively, a Coriolis mass f low rate meter that substantially eliminate t~ LuL e induced mea2,u~ L errors which might otherwise be l0 produced by performance differences existing between two separate input channel circuits contained in the meter.

2. Description of the Prior Art Currently, Coriolis meters are finding increasing use in a wide variety of commercial applications as an accurate way to measure the mass flow rate of various process f luids .
- Generally c:p~k;n~, a Coriolis mass flow rate meter, such as that described in United States patent 4,491,025 (issued to J. E. Smith et al on January 1, 1985 and owned by the present ;~:si~n~ hereof -- hereinafter referred to as the ' 025 Smith patent), contains one or two parallel conduits, each typically being a U-shaped flow conduit or tube. As stated in the '025 Smith patent, each flow conduit is driven to oscillate about an axis to create a rotational frame of reference. For a U-shaped flow conduit, this axis can be termed the bending axis . As process f luid f lows through each oscillating f low conduit, movement of the f luid produces reactionary Coriolis forces that are orthogonal to both the velocity of the f luid and the angular velocity of the conduit. These reactionary Coriolis forces, though quite 3s small when compared to a force at which the conduits are y - -r _ _ _ _ 21~g98 ~
WQnl/n~ 2-- PCI/US92/OSS83 --driven, nevertheless cause each conduit to twist about a torsional axis that, for a U-shaped flow conduit, is normal to its bending axis. The amount of twist imparted to each conduit is related to the mass flow rate of the S process fluid flowing therethrough. This twist is frequently measured using velocity signals obtained from r~ Pt ~ ,- velocity sensors that are mounted to one or both of the flow conduits in order to provide a complete velocity profile of the ~ of each flow conduit 10 with respect to either the other conduit or a f ixed reference. In dual conduit Coriolis meters, both flow conduits are oppositely driven such that each conduit oscillates (vibrates) as a separate tine of a tuning fork. This "tuning fork" operation advantageously 15 cancels substantially all undesirable vibrations that might otherwise mask the Coriolis force.
In such a Coriolis meter, the mass f low rate of a fluid that moves through the meter is generally ~0 proportional to the time interval (the so-called "~t"
value) that elapses between the instant one point situated on a side leg of a flow conduit crosses a pre-de~-rm;nP~ location, e.g. a respective mid-plane of oscillation, until the instant a corr~p~n~lin~ point 25 situated on the opposite side leg of the same f low conduit, crosses its co~-7l,v~;ng location, e.g. its respective mid-plane of oscillation. For parallel dual conduit Coriolis mass flow rate meters, this interval is generally equal to the phase difference between the 30 velocity signals generated f or both f low conduits at the fllrltl2 Lal (resonant) frequency at which these conduits are driven. In addition, the resonant frequency at which each flow conduit oscillates depends upon the total mass of that conduit, i.e. the mass of the conduit itself, 35 when empty, plus the mass of any fluid flowing therethrough. Tn2~l ~h as the total mass varies as the density of the fluid flowing through the conduit varies, the resonant frequency likewise varies with any changes -2~11698 ~093/01472 PCI/US92/05583 in fluid density and, as such, can be used to track changes in f luid density .
For some time, the art has taught that both S velocity signals are processed through at least some analog circuitry in an effort to generate output signals that are proportional to the mass f low rate of the process fluid. In particular, the output signal associated with each velocity sensor is ordinarily lO applied through analog circuitry, e.g. an integrator followed by a zero crossing detector (comparator), contained within a separate c oLL~ullding input channel.
In this regard, see illustratively United States patents 4,879,911 (issued to N. J. Zolock on November 14, 1989), lS 4,872,351 (issued to J. R. Ruesch on October 10, 1989), 4,843,890 (issued to A. L. Samson et al on July 4, 1989) and 4,422,338 (issued to J. E. Smith on Dec. 27, 1983) -- all of which are also owned by the present assignee hereof. While the various approaches taught in these 20 patents provide suf f iciently accurate results in a wide array of applications, the meters ~ lo~c~d in these references, as well as similar Coriolis meters known in the art, nevertheless suffer from a common drawback which complicates their use.
Specifically, Coriolis mass flow meters operate by detecting what amounts to be a very small inter-channel phase difference between the signals pl o~uced by both velocity sensors, i . e . the ~t value, and 30 transforming this difference into a signal proportional to mass flow rate. While, at its face, a l~t value is obtained through a time difference mea~uL~ , this value, in actuality, is also a pha6e measurement. Using such a time dif f erence measurement conveniently provides 35 a way to accurately measure a manifestation of a phase difference appearing between the velocity sensor signals.
In Coriolis meters currently manuf actured by the present assignee, this difference tends to ~mount to 2~ 1 169~
WO 93/0147~ PCr/US92/05583 approximately 13 Olsec at maximum f low . Each input channel in a Coriolis meter imparts some internal phase delay to its input signal. While the amount of this delay is generally quite small, it i5 often significant when 5 compared to the small inter-channel phase difference, i . e . 13 olsec or less, that is being detected . Currently available Coriolis meters have relied on ACC--rn; ng that each input channel imparts a f inite and f ixed amount of phase delay to its uc,~ o~ ~lin~ velocity signal. As lO such, these Coriolis meters generally rely on first measuring, at a true zero flow condition occurring during meter calibration, either the inter-channel phase difference (~t) or the indicated mass flow rate.
Subsequently, while metering actual flow, these meters 15 will then subtract the resulting value, in some fashion, from either the measured ~t or mass flow rate value, as appropriate, in order to generate an ostensibly accurate mass flow rate value for the process fluid then flowing therethrough .
~0 Unfortunately, in practice, this assumption has proven to be inaccurate. First, not only does each input channel often produce a different amount of internal phase delay with respect to the other, but also the phase5 dQlay that is pLuduce-l by each channel i5 t~ LL~Le t and varies differently from one channel to the other with corrP~pn~l; ng changec in temperature. This temperature variability results in a temperature induced inter-channel phase dif f erence . Because the measured 30 phase difference (~t~ that results from actual flow through the meter is relatively small, then an error in the measured phase dif ference between the velocity signals and attributable to the t ~ItU~ ~ induced inter-channel phase difference can, in certain instances, 35 be signif icant . This error is generally not taken into account in currently available Coriolis mass f low rate meters. In certain situations, this error can impart a noticeable t~ .ILur~ ~PpPnt1Pn~ err--r ~nto masG flow 93/01472 2 ~ i 9 ~3 PCI`/US92/05583 'O --5--rate mea~uL~ ts, thereby corrupting the measurements somewhat .
In an effort to avoid this error, one well 5 known solution in the art is to shroud an installed piped - Coriolis meter, inrlllAin~ its electronics, with a t~ -~ CS~UL~: controlled enclosure. This approach, which prevents the meter from being exposed to external t~ clLuLe variations and maintains the meter at a 10 relatively constant t~ uLe while it is in operation, greatly increases the installed cost of the meter and is thus not suited for every application. Hence, in those applications where installed cost is a concern, this approach is generally not taken. Specifically, in those 15 applications and particularly where the meter is to be sited indoors and not exposed to wide tl CltUL è
variations, then the mea2,uL~ ~ error which results from the temperature induced inter-channel phase difference, while generally expected, tends to remain quite small and 20 relatively constant. As such, this error is usually tolerated by a user. Unfortunately, in other applications where the meter is not housed in a temperature controlled enclosure, such as outdoor installations where the meter is expected to experience 25 wide f luctuations in operating temperature, the error generally varies and can become significant, and thus needs to be taken into account.
Apart from errors arising from temperature 30 induced inter-channel phase differences, many currently available Coriolis mass flow rate meters also disadvant~ge~11cly exhibit an additional source of mea~L ~ L inaccuracy related to ~ UL ~: . In particular, Coriolis meters generally measure the 3s t~ è~ uLè of the flow conduit and, owing to changes in flow conduit elasticity with t~ .~UL~:, accordingly modify a meter factor value based upon the current temperature of the conduit. ~his meter factor, as , ~:

2~ 11698 WO 93/01472 PCI`/US92/05583 modified, is then subsequently used to proportionally relate the inter-channel phase difference (~t) value to mass f low rate . Flow conduit temperature is measured by digitizing an output of a suitable analog temperature 5 sensor, such as a platinum RTD (resistive temperature device), that is mounted to an external surface of a flow conduit. The digitized output usually takes the form of a frequency signal, oftentimes produced by a voltage-to-frequency (V/F) converter, that is totalized 10 (counted) over a given timing interval to yield an A~,_ lAted digital value that is proportional to flow conduit temperature. Unfortunately, in practice, V/F
converters usually exhibit some temperature drift which, based upon the magnitude of a change in ambient 15 temperature, could lead to an error, amounting to as much as several degrees, in the mea:,uL L of f low conduit temperature. This error will, in turn, corrupt the mass f low rate.
A solution proposed in the art to ostensibly deal with temperature depDn~Dnt variations in the performance of the input ~ hAnnDl ~ of Coriolis meters is taught in United States patent 4,817,448 (issued to J. W.
Hargarten et al on April 4, 1989 and also owned by the present assignee hereof -- hereinafter referred to as the ' 448 Hargarten et al patent) . This patent discloses a two channel switching input circuit for use in a Coriolis meter. In particular, this circuit includes a two-pole two-throw FET (field effect transistor) switch located between the outputs of the velocity sensors and the inputs to both of the rhAnnDl~:. In one position, the FET
switch connDctc the outputs of the left and right velocity sensors to C~I~L~1J~nA;ng inputs of the left and right channels, respectively; while in the opposite position, these connections are reversed. The switch is operated to change its position at every successive cycle of f low conduit movement . In this manner, the output of each velocity sensor is alternately applied to both 2 ~ 8 '0 93/01472 PCI`/US92/05S83 rhAnnPlc in succession. Over a two cycle interval, appropriate time intervals are measured with respect to the velocity waveforms applied to both rhAnnPl ~ and then averaged together to yield a single time interval value 5 from which errors attributable to each individual channel have been r~nrDl Pd . This resulting time interval value is then used in dptprm;ninj mass flow rate through the meter .
While this solution does indeed substantially eliminate temperature induced inter-channel phase differences, it possessPC a drawback which limits its utility somewhat. Specifically, this input circuits in the apparatus taught in ' 448 Hargarten et al patent do 15 not include integrators. owing to the lack of any low pass f iltering that would have been provided by integrators, these input circuits are therefore susceptible to noise. Unfortunately, the switching scheme taught in this patent does not permit integrators 20 to be included in the switched portion of the input circuitry, hence requiring that, to provide noise immunity, an integrator must be located after the FET
switch. Unfortunately, in this location, the phase delay inherent in the integrator can not be readily -ncated, if at all. TnA ch as the integrator disadvantageously tends to provide the largest source of phase delay in the input circuitry, inclusion of such an integrator would add an error .r-^t, i . e. an lln~ ^^cated phase delay, to the mea-cured ~t values.
Moreover, this phase delay would also vary with temperature changes. Consequently, the resulting measured f low rate values would contain an error component. Thus, it became apparent that the solution posed in the ' 448 Hargarten et al patent has limited applicability to relatively noise-free environments.
Therefore, a need exists in the art for a Coriolis meter that provides accurate flow and flow rate ~ill69~
WO 93/01472 ~ PCr/US92/05583 output values that are substantially insensitive to ambient temperature variations and hence does not appreciably exhibit adverse temperature af f ects an could provide appreciable noise immunity. Such a meter should 5 possess negligible, if any, t elLUL~ induced measurement inaccuracies over relatively wide variations in ambient temperature thereby permitting the meter to be used to provide highly accurate f low mea~uL~ Ls in a wide variety of applications and particularly without a 10 need to house the meter in a t~ ~LUL~ controlled enclosure. Advantageously, the increased mea-uL- L
accuracy provided by such a meter and the attendant installed cost savings associated therewith would likely broaden the range of applications over which such a meter 15 could be used.

~0 93/01472 PCI/US92/05583 SllMMARY OF THE INVEN~rION
An object of the present invention is to provide a Coriolis meter that provides accurate output 5 meat,uL~ ts that are substantially insensitive to variations in ambient temperature.
A specific object is to provide such a meter that substantially, if not totally, eliminates the need 10 for a temperature controlled enclosure.
Another specific object is to provide a Coriolis meter in which the measured flow and flow rate values do not contain appreciable error, if any at all, 15 that would otherwise resul~ from switching transients appearing in the input rh:~nn~ c .
These and other objects are accomplished in accordance with the tD~rhin~C of my invention by cycling 20 the operation of each channel, particularly using a relatively short period, between: (a) measuring the internal phase delay of that channel and (b) mea~uring raw flow based ~t value(s). The raw value(s) are then -- ted, typically by subtracting, the measured phase 2s delay value therefrom in order to yield a corrected ~t value . A current value mass f low rate is then det~nminDcl using the corrected rather than, as occurs in the art, the raw ~t value(s).
Specifically, the two identical input rhi~nn~lc (i.e. left and right), as commonly used in prior art Coriolis flow meters, are replaced with two pairs of input rh~nnDlc (i.e. pairs A-C and B-C) that permit the current internal phase delay exhibited by each channel 3s pair to be measured. Each of the channel pairs is operated to cycle between measuring its own internal phase delay, i.e. a "zeroing" mode, and measuring ~t values for actual flow conditions, i.e. a "mea:,ur~ ~"

2~116~
WO 93/01472 -lO- PCI/I)S92/0~583 mode. Given the short cycle time, the current phase delay value accurately reflects any temperature induced changes then occurring in the perf ormance of each channel pair. Once the current internal pha6e delay value is 5 known for each pair, that value i6 then used to correct f low based ~t values s1lhcPqnpntly produced by that pair during its next mea~uL- ~ mode. Because the ~t flow based mea,,uL ~ I.s provided by each channel pair are corrected for the current internal phase delay associated 10 with that particular pair, these ~t values do not contain any appreciable t~ l u-~ induced error ~ ts regardless of the ambient temperature of the meter and its variation. As such, a Coriolis meter constructed in accordance with my invention, can advantageously be used 15 in environments with widely varying temperatures with essentially no diminution in accuracy owing to temperature changes.
In accordance with the teAch;n~C of a preferred 20 Pmho~l i t of my invention, my inventive f low mea~
circuit utilizes three separate similar input rh~nn~P1 s (i.e. -h~nnPlc A, B and C) through which inter-channel phase difference mea:,uL~ ~s are successiYely and alternately taken for each of two pairs, i.e. pairs A-C
25 and B-C, of the three r~h~nnPl c. Channel C serves as a reference channel and is continuously supplied with one of the two velocity waveform sensor signals, and specifically for purposes of the preferred Pmho~;r ~ the left velocity sensor signal, as its input signal. The 30 input to 'h lnnPl C A and B is either the left or right velocity sensor signals. While both the zero and mea~ul ~ modes involve measuring the inter-channel phase difference in a pair of r~hAnnPlC, the principal distinction between the modes is that in the zero mode, 35 the same velocity sensor signal is applied to both rhilnnPl c in that pair so that the resulting inter-channel phase difference measurement provides a measurement of the internal phase delay for that pair; wh~ le, i~ the 2~1~6~g '0 93/01472 PCr/US92/05583 measurement mode, the left and right velocity signals are applied to different iULL~ in~ ~!h~nnPlc in that pair so as to provide a mea~u-- - , though uncorrected, of the current f low based ~t value for subsequent use in S detPrminin~ current mass flow and flow rate values.
Though inter-channel phase difference (~t) mea,u.c ~D
are taken during both modes, to simplify matters and avoid confusion, I will distinguish between these values in terms of their ~c~ u-.e~.c;e. I will henceforth refer to 10 those phase measurements which occur during the zero mode as being intêr-channel phase difference meaau-~ Ls and those which occur during the meaDuL - L mode as being ~t values .
Specifically, for any channel pair operating in the zero mode, such as pair A-C, the same, i.e. left, velocity sensor signal is appliêd to the inputs of both rh~nnPlc in that pair. Inter-channel phase difference meaDu-~ ts are then successivêly and repêtitively taken 20 during a so-called "zeroing" interval with the results being averaged during this interval. Ideally, if both of the rhRnn~1 c in this pair exhibit the same internal phase delay, i . e . the phase delay through channel A equals that of reference channel C, then the resulting inter--channel 25 phase difference meaDu-~ ts will all equal zêro.
However, in actuality, at any instant, all three rh;lnnl~l c usually possess different internal phase dêlays.
Neverthelêss, since the phase delay for each pair is measured with respect to the same ref erênce channel, i . e .
30 channel C, any differences in the phase delay between the two pairs is caused by differences in the internal phase delay occurring between rh InnPl c A and B. Once the "zeroing" interval has terminated, the input to the non-reference channel in that pair is switched to the 35 other velocity sensor signal, i. e. the right velocity sênsor signal. A finite period of time, i.e. including a so-called "switching" interval, is then allowed to expire before that channel pair is operated in the "measurement"

WO 93/01472~ -12- PCI/US92/05583 mode during which f low based l~t values are measured. The switching interval is sufficiently long to enable all resulting switching transients to settle out.
While one pair of ~ h~nn~lc, e.g. A-C, is operating in its zero mode, the other pair, e.g. B-C, is operating in its mea~iuL~ mode in order to provide continuous flow metering. For any channel pair, each successive current flow based ~t value obtained during its mea~uL~ L mode is -- ~ted by, typically subtracting, the most recent value of the internal phase delay that has been measured for this channel pair during its preceding zero mode.
The time during which one channel pair operates in the measurement mode, i.e. the measuring interval, equals the entire time that the other pair operates in the zero mode. This latter time includes the time during which the latter channel switches its non-reference channel input from the right to the left velocity sensor signal, then performs zeroing, and finally switches its non-reference channel input from the left back to the right velocity sensor signal.
At the conclusion of the measurement interval, the channel pairs simply switch modes, with illustratively channel pair B-C initially switching its non-reference channel input from the right to the left velocity sensor signal, and channel pair A-C _ --in~
flow based ~t mea:,u~ ~ ~s. Once this input switching is complete, channel pair B-C then undertakes zeroing followed by channel switching in the opposite direction -- while channel pair A-C remains in the mea,,uL~ L
mode, and so on for successive cycles of operation.
Furthermore, in accordance with my inventive teat~hin~c, tl, ~Lur-: induced errors in the temperature mea_uL~ L of the flow cond~t provided through the RTD, 2~ 11638 and specifically associated with temperature drift in the V/F converter, are also advantageously eliminated.
Specifically, to eliminate these errors, two reference voltages, in addition to the RTD voltage, are selectively 5 and successively converted through the V/F converter into - frequency values, in terms of counts, and are then used to define a linear relatir~nchir~ specifically a proportionality factor, that relates the counted frequency value to measured flow conduit temperature.
10 Then, by simply multiplying the counted frequency value for the RTD voltage by this factor, a value for the COL.~.~O~ ;n~ measured flow conduit temperature results.
Tn;~ ch as the reference voltages do not appreciably change, if at all, with temperature variations and are 15 each repetitively converted through the V/F converter at a relatively short periodicity, on the order of illustratively .8 seconds, any temperature drift produced by the V/F is accurately ref lected in the resulting counted frequency values for the reference voltages 20 themselves. Since temperature drift equally affects the counted values for both reference voltages and the RTD
voltage, but does not change the relati~nchirs thereamong, the proportionality factor when multiplied by the counted frequency value for the RTD voltage produces 25 a true temperature value that is substantially i n~erc-nrl~t of any temperature drift produced by the V/F
converter. By eliminating temperature induced errors in the measured temperature, the meter factor will be appropriately modif ied in a manner that accurately 30 ref lects changes in f low conduit temperature .
Furthermore, while my inventive meter determines a current mechanical zero value (i.e. the zero flow offset value of the meter) based upon a number of no 35 flow ~t meaauL~ Ls taken during meter calibration, a feature of my inventive meter is to use that value in subsequently c~mr~nc~ting actual flow meaauL- ~s only if the noise content of these no flow Qt meaauL~ t s is WO 93/01472 -14- PCrlUS92/05S83 sufficiently low, otherwise that value is ignored. The number of no flow ~t mea~.uL- - Ls is guv~:Llled by any of three factors: (a) whenever the standard deviation of these mea~urO --Ls falls below a convergence limit, (b) 5 whenever a user manually terminates the mechanical zero process, or (c) if a pre-defined maximum number of such measurements has been taken.
BRIEF DESt~RTPTION OF ~ RAWINGS
The t~rh;n~C of the present invention may be clearly understood by considering the following detailed description in conjunction with the A~ nying 15 drawings, in which:
FIG. l is an overall diagram of-Coriolis mass f low rate metering system 5;
FIG. 2 depicts a high level block diagram of 20 well known meter electronics 20 shown in FIG. l;
FIG. 3 shows the correct alignment of the drawing sheets for FIGs. 3A and 3B;
~5 FIGs. 3A and 3B collectively depict a high level block diagram of a preferred ~ho~ t of flow mea~-lL. L circuit 30 according to my present invention;
FIG. 4 shows the correct alignment of the 30 drawing sheets for FIGs. 4A and 4B;
FIGs. 4A and 4B collectively depict a timing diagram of the operations performed by channel pairs A-C
and B-C in flow measurement circuit 30 shown in FIGs. 3A
35 and 3B;

21 l~S9~ `
~/0 93tO1472 -15- PCI'/US92/05583 FIG. 5 depicts a state table of circuit 70 that is contained within flow mea,,ur ~ ~ circuit 30 shown in FIGs. 3A and 3B;
FIG. 6 shows the correct alignment of the drawing sheets for FIGs. 6A and 6B;
FIGs. 6A and 6B collectively depict a simplified flowchart of Flow Measurement Basic Nain Loop 600 that is executed by microprocessor 80 that is contained within flow measurement circuit 30 shown in FIGs. 3A and 3B;
FIG. 7 shows the correct alignment of the drawing sheets f or FIGs . 7A and 7B;
FIGs. 7A and 7B collectively depict a flowchart of Zero Determination Routine 700 that is executed as part of Main Loop 600 shown in FIGs. 6A and 6B;
FIG. 8 shows the correct alignment of the drawing sheets for FIGs. 8A and 8B;
FIGs. 8A and 8B collectively depict a flo-~rchart o~ Mechanical Zero Routine 800 that is executed as part of Zero Determination Routine 700 shown in FIGs. 7A and 7B;
FIG. 9 diagrammatically shows the zeroing operations that occur for each corrPCpnn~l;n~ range in the standard deviation, i.e. ~t, of the measured Qt values that are obtained during a mechanical zero process;
FIG. 10 diagrammatically shows the ranges of acceptable and non-acceptable mechanical zero values; and FIG. ll shows a flowchart of RTD T~ La-UL~
Processing Routine 1100 which is executed on a periodic ~, ~169~

interrupt basis by microprocessor 80 that is contained within inventive flow mea- u.~ ~ circuit 30 shown in FIGs. 3A and 38.
To facilitate understanding, identical reference numerals have been used, where appropriate, to designate identical elements that are common to the f igures .
DETATr~r`n DESCRIPTION
After reading the following description, those skilled in the art will readily appreciate that my inventive technigue can be incu-~uL~,ted within a wide variety of circuitry that measures multiple inputs using multiple analog input rhi~nnpl c. Advantageously, use o~
my invention substantially, if not totally, eliminates errors that might otherwise arise from performance differences occurring among the individual rh innPl c and attributable to, for example, temperature, aging and/or other 1' - that differently affect the analog circuitry contained therein. Of course, such usage would include any Coriolis meter regardless of whether that meter is measuring f low, f low rate, dens ity or other parameter(s) of a process fluid. Nevertheless, for purposes of brevity, my inventive input circuit will be ~licr11c8Pfl in the context of a dual conduit (tube) Coriolis meter that specifically measures mass flow rate and totalized mass f low.
FIG. l shows an overall diagram of Coriolis mass f low metering system 5 .
As shown, system 5 consists of two basic ents: Coriolis meter assembly l0 and meter electronics 20. ~qeter assemhly lO measures the mass flow rate of a desired process fluid. Meter electronics 20, rnn"Pcted to meter assembly l0 ~i~ ~eads lOO, ~ro 93/01472 2 ~ 1~ 6 ~
illustratively provides mass f low rate and totalized mass flow information. ~ass flow rate information is provided over leads 26 in frequency form and in scaled pulse form.
In addition, mass flow rate information is also provided in analog 4-20 mA form over leads 26 for easy connection - to downstream process control and/or mea~u.~ L
L.
Coriolis meter assembly lO, as shown, includes lo a pair of manifolds 110 and 110'; tubular member 150; a pair of parallel flow conduits (tubes) 130 and 130 ';
drive ~n;Fm 180; a pair of velocity sensing coils 160L and 160R; and a pair of porr~nollt magnets 170L and 170R. Conduits 130 and 130' are substantially U-shaped and have their ends attached to conduit mounting blocks 120 and 120 ', which are, in turn, secured to respective manifolds 110 and 110 ' . Both flow conduits are free of pressure sensitive joints.
With the side legs of conduits 130 and 130 ' fixedly attached to conduit mounting blocks 120 and 120' and these blocks, in turn, fixedly attached to manifolds 110 and 110 ', as shown in FIG. 1, a continuous closed fluid path is provided through Coriolis meter assembly 10. Specifically, when meter 10 is connectod~ via inlet end 101 and outlet end 101 ', into a conduit system (not shown) which carries the process f luid that is being measured, fluid enters the meter through an orifice in inlet end 101 of manifold 110 and is conducted through a r~C~se~-aY therein having a gradually changing cross-section to conduit mounting block 120. There, the fluid is divided and routed through flow conduits 130 and 130 ' .
Upon exiting flow conduits 130 and 130', the process f luid is recombined in a single stream within conduit mounting block 120 ' and is thereafter routed to manifold 110'. Within manifold 110', the fluid flows through a passageway having a similar gradually changing cross-section to that of manifold ~10 -- as shown by 2~l16~
O 93/01472 PCr/llS92/05583 dotted lines 105 -- to an orif ice in outlet end lol ' . At end 101', the fluid reenters the conduit system. Tubular member 150 does not conduct any fluid. Instead, this member serves to axially align manifolds 110 and 110 ' and 5 maintain the spacing therebetween by a pre-determined amount 80 that these manifolds will readily receive mounting blocks 120 and 120 ' and flow conduits 130 and 13 0 ' .
U-shaped flow conduits 130 and 130 ' are selected and appropriately mounted to the conduit mounting blocks so as to have substantially the same moments of inertia and spring constants about bending axes W-W and W ' -W ', respectively . These bending axes are 15 perpendicularly oriented to the side legs of the U-shaped f low conduits and are located near respective conduit mounting blocks 120 and 120 ' . The U-shaped flow conduits extend outwardly from the mounting blocks in an essentially parallel fashion and have substantially equal 20 moments of inertia and equal spring constants about their respective bending axes. TnA ~!h as the spring constant of the conduits changes with t ~u~e, resistive temperature detector (RTD) 190 (typically a platinum RTD
device) is mounted to one of the flow conduits, here 25 conduit 130 ', to continuously measure the temperature of the conduit. The temperature of the conduit and hence the voltage appearing across the RTD, for a given current passing therethrough, will be governed by the temperature of the f luid passing through the f low conduit . The 30 temperature d~r~n~nt voltage appearing across the RTD is used, in a well known method, by meter electronics 20 to appropriately compensate the value of the spring constant for any changes in conduit temperature. The RTD is connected to meter electronics 20 by lead 195.
3s Both of these flow conduits are driven, typically sinusoidally, in opposite directions about their respective bending axes and at essentially their 211~8 ~0 93/01472 PCr/US92/05583 common resonant frequency. In this manner, both flow conduits will vibrate in the same manner as do the tines of a tuning fork. Drive ---h~n;cm 180 supplies the oscillatory driving forces to conduits 130 and 130 ' .
5 This drive r--hAnO:m can consist of any one of many well - known arrAn~ ~s, such as a magnet mounted to illustratively flow conduit 130 ' and an opposing coil mounted to illustratively flow conduit 130 and through which an alternating current is passed, for sinusoidally vibrating both flow conduits at a common frequency. A
suitable drive signal is applied by meter electronics 20, via lead 185, to drive r--hAnie~m 180.
With f luid f lowing through both conduits while lS these conduits are driven in opposing directions, Coriolis forces will be generated along adjacent side legs of each of flow conduits 130 and 130 ' but in opposite directions, i.e. the Coriolis force generated in side leg 131 will oppose that generated in side leg 131 ' .
This rh~r -nnn occurs because although the f luid f lows through the flow conduits in essentially the same parallel direction, the angular velocity vectors for the oscillating (vibrating) flow conduits are situated in opposite though essentially parallel directions.
2s Accordingly and as a result of the Coriolis forces, during one-half of the oscillation cycle of both flow conduits, side legs 131 and 131 ' will be twisted closer together than the minimum distance occurring between these legs produced by just the oscillatory - v~ ~5~ of the conduits generated by drive r~ ni cm 180. During the next half-cycle, the generated Coriolis forces will twist side legs 131 and 131 ' further apart than the maximum distance occurring between these legs produced by just the oscillatory movement of the conduits generated by drive ----hAni~m 180.
During oscillation of the flow conduits, the adjacent side legs, which are forced closer together than 2~ g8l O 93/01472 PCr/US92/05583 their counterpart side legs, will reach the end point of their travel, where their velocity crosses zero, before their counterparts do. The time interval (also referred to herein as the inter-channel phase difference, or time 5 difference or simply "~t" value) which elapses from the instant one pair sf adjacent side legs reaches their end point of travel to the instant the counterpart pair of side legs, i.e. those forced further apart, reach their respective end point is substantially proportional to the lO mass f low rate of the f luid f lowing through meter assembly 10. The reader is referred to United States Patent 4,491,025 (issued to J. E. Smith et al on January 1, 1985) for a more detailed r~ c~lc~ion of the principles of operation of parallel path Coriolis flow meters than 15 that j ust presented .
To measure the time interval, Vt, coils 160L
and 160R are attached to either one of conduits 130 and 130 ' near their free ends and permanent magnets 170L and 20 170R are also attached near the free ends of the other one of the conduits. Magnets 170L and 170R are ~ pQ5 so as to have coils 160L and 160R located in the volume of space that ~ur, uu~lds the respective p-~rr-n~nt magnets and in which the magnetic flux fields are essentially 25 uniform. With this configuration, the electrical signal outputs generated by coils 160L and 160R provide a velocity profile of the complete travel of the conduits and can be processed, through any one of a number of well known methods, to determine the time interval and, in 30 turn, the mass flow rate of the fluid passing through the meter. In particular, coils 160L and 160R produce the left and right velocity signals that appear on leads 165L
and 165R, respectively. As such, coils 160L and 160R and corL~ lin~ magnets 170L and 170R respectively form the 35 left and right velocity sensors. While at its face Vt is obtained through a time difference mea~u~e:.ue-.L, Vt is in actuality a phase measurement. Using a time difference mea2,ul t here provides an accurate way to 2111 6~8 ~'0 93/01472 PCr/US92/05583 measure a manifestation of the phase difference that occurs between the left and right velocity sensor signals .
As noted, meter electronics 20 accepts as input the RTD signal appearing on lead 195, and the left and right velocity signals appearing on leads 165L and 165R, respectively. Meter electronics 20 also produces, as noted, the drive signal appearing on lead 185. Leads 165L, 165R, 185 and 195 are collectively referred to as leads 100. The meter electronics processes both the left and right velocity signals and the RTD signal to determine the mass flow rate and totalized mass flow of the fluid passing through meter assembly 10. This mass flow rate is provided by meter electronics 20 on associated lines within leads 26 in analog 4-20 mA form.
Nass flow rate information is also provided in frequency form (typically with a maximum range of 0-10 KHz) over an appropriate line within leads 26 for connection to d.. ~LLeam eSrl; L.
A block diagram of meter electronics 20, as known in the art, is depicted in FIG. 2. Here, as shown, meter electronics 20 consists of flow mea:,uL~ - L circuit 2s 23, flow tube drive circuit 27 and display 29.
Flow tube drive circuit 27, depicted in FIG. 2, provides an appropriate repetitive alternating or pulsed drive signal, via lead 185, to drive -~^h~n;f~ 180. This 30 circuit synchronizes the drive signal to the left velocity signal which appears on leads 165L and 25. In operation, circuit 27 maintains both flow tubes in opposing sinusoidal vibratory motion at a fllnl~ ~al resonant frequency. As is known in the art, this 35 frequency is y~v~LI-ed by a number of factors, including various characteristics of the tubes themselves and the density of the process f luid f lowing therethrough . Since circuit 27 is very well known in the art and its specific , 211~698 WO 93/01472 PCr/US92/05583 implementation does not form any part of the present invention, this circuit will not be ~9iCCllCCP-q in any further detail herein. In this regard, the reader i5 illustratively ref erred to United States patents 5,009,109 (issued to P. Kalotay et al on April 23, 1991);
4,934,196 (issued to P. Romano on June 19, 1990) and 4,876,879 (is6ued to J. Ruesch on October 31, 1989) --all of which are owned by the present assignee hereof and describe different r-~o~l; rts for the flow tube drive circuit.
Plow measurement circuit 23 processes the left and right velocity signals appearing over leads 165L and 165R, respectively, along with the RTD signal appearing on lead 195, in a well known manner, to determine the mass flow rate and totalized mass flow of the process fluid passing through meter assembly 10. The resulting mass flow rate information is provided as a 4-20 mA
output signal over lead 263, for easy connection to additional d. l-~LL~alll process control equipment (not shown), and as a scaled frequency signal over lead 262 for easy connection to a remote totalizer (also not shown). The signals appearing on leads 262 and 263 form part of the process signals that collectively appear on leads 26 shown in FIG. 1. Other leads (not specifically shown) within leads 26 provide totalized flow information, as well as other process parameters, in digital form for connection to suitable display, telemetry and/or downstream processing equipment.
TnA~ ~h as the method through which flow mea~uL L circuit 23 generates mass flow and totalized flow rate information is well known to those skilled in the art, only that portion of its constituent electronics that are germane to the present invention will be ~9;ccllcsecl hereinafter. In this regard, mea~uL. t circuit 23 contains two separate input rh~nnPlc: left channel 202 and right channel 212. Each channel contains 2~1~6~
~0 93/01472 PCI/US92/0 an integrator and two zero crossing detectors. Within both rhAnn~l ~, the left and right velocity signals are applied to respective integrators 206 and 216, each of which effectively forms a low pass filter. The resulting 5 outputs of these integrators are applied to zero crossing detectors (effectively comparators) 208 and 218, each of which generates a level changes whenever the CCILL _~I,ol ~li nq integrated velocity signal exceeds a voltage window defined by a small predefined positive and 10 negative voltage level, e. g . +v. The outputs of both zero crossing detectors 208 and 218 are fed as control signals to counter 220 in order to measure a timing interval, in terms of clock pulse counts, that occurs between COL L -`1JOI~1 i nq changes in these outputs . This l5 interval is the well known At value and varies with the mass f low rate of the process f luid . The resulting ~t value, in counts, is applied, in parallel, as input data to processing circuitry 235. In addition, RTD 190 is connected to an input of RTD input circuit 224 which 20 supplies a constant drive current to the RTD, linearizes the voltage that appears across the RTD and converts this voltage using voltage/frequency (V/F) converter 226 into a stream of pulses that has a 6caled frequency which varies proportionally with any changes in RTD voltage.
25 The resulting pulse stream produced by circuit 224 is applied as an input to counter 228 which periodically counts the stream and produces a value, in counts, that is proportional to the measured te o ~UL ~2 . The contents of counter 228 are also applied in parallel as 30 input data to processing circuit 235. Processing circuit 235, which is typically a mi- L~,~L~essor based system, det~nm;n~ the current mass flow rate from the digitized ~t and t~ CltUL~ values applied thereto. In this regard, the digitized t elLULt: value is used to 35 modify a meter factor value based upon the current temperature of the f low tubes and, by doing so, account for changes in flow tube elasticity with t~ C~tULt:.
The meter factor, as modified, (i.e. a temperature W093/01472 21~ 8 -24- PCI`/US92/05583 ~
-ncated meter factor -- RF) is then subsequently used to proportionally determine the mass flow rate from the current measured ~t value. Having determined the mass flow rate, circuitry 235 then updates totalized mass flow 5 and ~lso provides, for example, suitable mass flow rate output signals over leads 26 for connection to local display 29 and/or to downstream process control e~uipment.
It is now become apparent that the analog circuitry contained within the left and right rh InnPl c disadvant~ cly injects some error into the resulting mass flow and mass flow rate values E~L~-luced by processing circuitry 235. Specifically, not only does 15 each input channel often possess a different amount of internal phase delay with respect to the other, as measured f rom the input of an integrator to an output of its zero crossing detectors, but also the phase delay that is internally E.Lo.luced by each channel is 20 t~ ~ltUL~: 19Prpn~lpnt and often varies differently from one channel to the other with cuL L -r ~Jo~ ; n~ changes in temperature. As such, left channel 202 may, for example, exhibit phase delay that has a dif f erent temperature (~PrPnrlPnt variation than that exhibited by right 25 channel 212. This variability results in a temperature induced inter-channel phase difference that manifests itself as an error 1 in the measured ~t value.
Because the ~t value that results from actual f low itself through the meter is relatively small, this error 30 component can, in certain instances, be significant.
This error is generally not taken into account in currently available Coriolis mass flow rate meters. In certain situations, particularly where the meter is situated in an outdoors environment and subjected to wide 35 temperature f luctuations, this error can impart a noticeable t~ ~tUL~ ~PrPn~lpnt error into mass flow rate measurements, thereby corrupting these measurements somewhat .

0 93/01472 ~ 6 ~ 8 Pcr/US92,05583 Now, apart from temperature d~pPnfl-~nt errors in the measured ~t value, the temperature measurement circuitry itself imparts an additional source of S temperature induced mea~,ù- ~ L error into the mass flow and flow rate values ~I~.luced by processing circuitry 235. In this regard, V/F converter 226 contained within RTD input circuit 224 exhibits, as do nearly all such converters, measurable t~ ~tuLe drift. This drift, 10 based upon the magnitude of a change in ambient temperature, may lead to an error, amounting to as much as several degrees, in the mea~uL~ L of the flow conduit temperature. This error will, in turn, lead to errors in the modified meter factor which, in turn, will 15 also corrupt the mass f low rate and totalized mass f low values .
To eliminate the defiri-~nrie~ as60ciated with Coriolis meters known in the art and particularly those 20 containing circuitry typified by flow mea~uL~ -t circuit 23, I have developed a technique for use in the f low mea2,uL~ L circuit of a Coriolis meter that advantageously renders the mass f low and mass f low rate values produced by the meter substantially insensitive to 2s temperature changes thereby improving their overall accuracy .
Specifically, in accordance with the tc-~rhin~c of my present invention, the two identical input rhAnn~
30 (i.e. left and right), as commonly used in prior art flow mea~uL~ ~ circuits, are replaced with two pairs of input rh~nn~.~c (i.e. pairs A-C and B-C) that permit the phase delay exhibited by each channel pair to be measured. Once the current value of the phase delay is 35 known for each channel pair, that value is subsequently used to correct f low based ~t values subsequently measured by that channel pair. Since each o~ the channel pairs is operated to cycle, on a relatively short period, 21 1 ~98 WOg3/01472 -26- PCI/US92/05583 between measuring its own internal pha6e delay, i.e. a "zeroing" mode, and measuring ~t values for actual flow conditions, i.e. a "measurement" mode, the current phase delay value accurately ref lects any temperature induced 5 changes then occurring in the perf ormance of each channel pair . Because the ~t f low based mea~uL ~ Ls provided by each channel pair are corrected for the current internal phase delay associated with that particular pair, these ~t values do not contain any appreciable temperature 10 induced error - -ts regardless of the ambient t~ -tuLa of the meter and its variation. AB such, a Coriolis meter constructed in accordance with my invention, can advantageously be used in environments with widely varying t~ ~LUL~S with essentially no 15 diminution in accuracy owing to temperature changes.
In particular, my inventive flow mea2.uL.
circuit utilizes three separate similar input ch;~nnr~
(i.e. rh~nnF~l c A, B and C) through which inter-channel 20 phase difference meaDuL. ts are successively and alternately taken for each of two pairs, i.e. pairs A-C
and B-C, of the three r hAnn"lc. Channel pair A-C
contains çh~nnPl c A and C; whi le ~ nnPl pair B-C
contains r~h InnPl c B and C. Channel C serves as a 25 reference channel and is continv~ cly supplied with one of the two velocity waveform sensor signals, and specifically for ~uL~oses of the preferred ~ L the left velocity sensor signal, as its input signal. The input to ~h~nnPl c A and B is either the left or right 30 velocity sensor siqnals. While both the zero and measurement modes involve measuring the inter-channel phase difference in a pair of r hAnnDl c, the principal distinction between the modes is that in the zero mode, the same, i.e. left, velocity sensor signal is applied to 35 both ~h;~nn~l c in that pair so that the resulting inter-channel phase difference mea~.uL~ L provides a meaDuLl L of the internal phase delay for that pair;
while, in the me~DuL~ t mode, the left and right 2~11698 0 93/01472 PCr/US92/05583 velocity signals are applied to different corr~c:po~l;n~
~hAnnel c in that pair so as to provide a mea~,uL~
though uncorrected, of the current flow based ~t value for subseyùt-.~ use in ~l~t~rm;nin~ current mass flow and 5 flow rate values. Though inter-channel phase difference (~t) mea~uL~ t6 are taken during both modes, to simplify matters and avoid confusion, I will di6tinguish between these values in terms of their OC~;UL L ~,.ce . Thus, I will henceforth refer to those phase meatiuL~ Ls which 10 occur during the zero mode as being inter-channel phase difference mea~ UL~ Ls and those which occur during the zero mode as being ~t values. Also, both the inter-channel phase difference mea,,uL~ --Ls and the ~t values for any channel pair will be collectively and 15 hereinafter referred to as timing measurements.
Specifically, for any channel pair operating in the zero mode, such as pair A-C, the same, i.e. left, velocity sensor signal is applied to the inputs of both 20 rhAnn~lc in that pair. Inter-channel phase difference meaDuL~ Ls are then successively and repetitively taken during a so-called "zeroing" interval with the results being averaged during this interval. Ideally, if both of the rhAnn~.l R in this pair exhibit the same internal phase 25 delay, i . e. the phase delay through channel A equals that of reference channel C, then the resulting inter-channel phase difference mea~uL~ Ls will all equal zero.
However, in actuality, at any instant, all three rhAnn~1 c usually possess different internal phase delays.
30 Nevertheless, since the phase delay for each pair is measured with respect to the same reference channel, i.e.
channel C, any differences in the phase delay between the two pairs is caused by differences in the internal phase delay occurring between chAnn~1 ~ A and B. Once the 35 "zeroing" interval has terminated, the input to the non-reference channel in that pair is switched to the other velocity sensor signal, i . e. the right velocity sensor signal. A finite period of time, i.e. including a 116~8O 93/014~2~ PCr/US92/05583 60-called "switching" interval, is then allowed to expire before that channel pair is operated in the "mea~uL~ "
mode during which f low based ~t values are mea6ured. The switching interval is sufficiently long to enable all S resulting switching transients to settle out, e.g. for their amplitude to decay below a pre-defined level.
While one pair of rh~nnPl~, e.g. A-C, is operating in its zero mode, the other pair, e.g. B-C, is 10 operating in its mea~uL ~ mode . For any channel pair, each successive measured f low based ~t value that is obtained during its mea2~uL~ t mode is compensated by, typically subtracting, the most recent value of the internal phase delay that has been measured f or this 15 channel pair during its preceding zero mode.
The time during which one channel pair operates in the meaDuL~ L mode, i.e. the measuring interval, eguals the entire time that the other pair operates in 20 the zero mode. This latter time (i.e. the "zero"
interval) includes the time (i.e. the "switching"
interval) during which the latter channel switches its non-reference channel input from the right to the left velocity sensor signal, then performs zeroing (during a 25 so-called "zeroing" interval), and finally switches its non-ref erence channel input f rom the lef t back to the right velocity sensor signal. Note that the zero interval includes both two switching intervals and a zeroing interval.
At the conclusion of the mea:,uL~ ~ interval, the channel pairs simply switch modes, with illustratively channel pair B-C initially switching its non-reference channel input from the right to the left 35 velocity sensor signal, and channel pair A-C commencing f low based l~t mea~-uL ts . Once this input switching is complete, channel pair B-C then undertakes zeroing followed ~y ch ~nne~ switching in the opposite direction 2 ~ 8 *!0 93/01472 -29- PCI`/US92/05583 -- while channel pair A-C remains in the measurement mode, and so on for succèssive cycles of operation.
After a channel pair has completed the latter switching operation but before commencing its operation in the 5 measurement mode, that channel can, if desired, undertake mea~uL~ Ls of flow based l~t values for a finite period of time, hereinafter referred to as the "active"
interval, which, to simplify implementation, has a duration equal to the "zeroinq" interval. Since both 10 rhAnnPl c can simultaneously provide flow based ~t values during the "active" interval from both velocity sensor signals, then, ideally, in the absence of any noise, isolated perturbations or differences between the internal phase delays associated with the pairs of 15 rhAnnPlc, the same ~t values should be ~Lu-luced by both r~hAnnr~l ~. Hence, as an added check, one or more of the measured flow based ~t values obtained from each channel pair during the "active" interval can be ~ ted by the most recent value of the measured phase delay for 20 that pair to yield ~WL L ~ O~ i n~ pairs of corrected ~t values. The two values in each such pair could then be compared against each other. A sufficient discrepancy between the values in any of these pairs would generally signify an error condition.
Tn ~ h as channel switching only occurs on the channel pair opposite from that which is being used to provide flow based mea:~uL~ ~s, any switching transients (and noise associated therewith) are 30 effectively isolated from and advantageously do not corrupt the flow and flow rate mea~uL~ ts. Moreover, by allowing an appropriately long switching interval to expire even before zeroing begins, the switching transients advantageously do not affect the internal 35 phase delay mea:,uL~ --Ls for the channel pair being zeroed. As such, the performance of a Coriolis meter that utilizes my invention is substantially, if not `8 O 93/01472 PCr/US92/05583 totally, immune from input switching transients and the like .
The specif ic length of time of the switching 5 and zeroing intervals is not critical. However, since switching transients die out rather quickly and additional averaging generally provides increased accuracy for the internal phàse delay mea_uL~ t6, the switching interval is typically set to be much shorter 10 than the zeroing interval. In this regard, the switching interval, as measured in tube cycles, may last for illustratively 16-32 such cycles, while the zeroing interval may be set to consume upwards of illustratively 2048 such cycles.
Furth, ~-~, in accordance with my inventive t~;~rhin~5~ t. ~.tuLe induced errors in the t~ ClLULt:
mea~UL. ~-L of the flow tube provided through the RTD, and specifically associated with t aLULt: drift in the 20 V/F converter, are also advantageously eliminated.
Specifically, to eliminate these errors, two reference voltages in addition to the RTD voltage are selectively and successively converted through the V/F converter into frequency values, in terms of counts, and are then used 25 to define a line2r relation~hirl specifically a proportionality factor, that relates the counted frequency value to measured flow tube temperature. Then, by simply multiplying the counted frequency value for the RTD voltage by this factor, a value for the cuLL~ 1;n~
30 measured flow tube temperature results. Tn~ I`h as the reference voltages do not appreciably change, if at all, with temperature variations and are each repetitively converted through the V/F converter at a relatively short periodicity, on the order of illustratively .8 second6, 35 any temperature drift produced by the V/F is accurately reflected in the resulting counted frequency values for the ref erence voltages themselves . Since temperature drift equally affects the counted values for both ; .

2~ g8 1O 93/01472 PCr/US92/05583 reference voltages and the RTD voltage, but does not change the relationships thereamong, the proportionality factor when multiplied by the counted frequency value for the RTD voltage produces a true t~, ~ q~UL~ value that is 5 substantially; nrl~ppnrlpnt of any temperature drift produced by the V/F converter. By eliminating temperature induced errors in the measured temperature, the meter factor will be appropriately modified in a manner that accurately reflects change6 in flow tube r' , I~UL~.
A. Hardware Description With this description in mind, a high level 15 block diagram of a preferred ~ t of inventive flow meaDuL~ ~ circuit 30 is collectively depicted in FIG6.
3A and 3B, for which the correct alignment of the drawing sheets f or these f igures is shown in FIG . 3 .
In es6ence, flow meabuL. - ~ circuit 30 contains an input multiplexor and three similar input channels -- one of which is reference channel C, a finite state machine with associated timing counters, and a microcomputer system. The inputs to the two non-reference rh~nn~1~ A and B are s~1 ~rted, through the multiplexor, by the finite state machine, as it cycles through its various 6tates. The outputs from the three rh;-nn~ are applied to the counters in order to generate the timing mea~uL, - ~s, i . e. the inter-channel pha6e difference mea~uL~ s and the ~t value6, for each of the two channel pairs A-C and B-C. The timing meaauL~ ~s provided by these counters along with the state information from the finite state machine are supplied to the mi~;Lo. _~er which, in turn, determines current COL L r-~y~ rl; n~ values of mass f low rate . In addition, the RTD output and two reference voltages are sequentially converted into corresponding frequency values, through an appropriate input switch, V/F
, . .

211169~8 W093/01472 -32- PCI/US92/0~83 --converter and associated circuitry, and counted through a timing counter associated with the f inite state machine .
The resulting counts therefor are then supplied by this counter to the mi~:L.- _Ler for its use in properly s modifying the meter factor.
Specifically, as depicted, flow mea~uL~ L
circuit 30 contains three similar input rhAnnPlq 44, 54 and 64, also respectively referred to herein as l~hAnnPl q 10 A, C and B. in addition, this flow mea-~-L~ ~ L circuit also contains multiplexor 31, circuit 70, analog switch 35, reference voltage generator 39, RTD input circuit 42, microcomputer 80, output circuitry 90 and input circuitry 95.
RTD input circuit 42, shown in FIGs. 3A and 3B, performs the same functions and contains essentially the same circuitry as RTD input circuit 224 6hown in FIGs. 2A
and 2B and ~1 i ccllqcp~ above.
Each of f hAnnPl c A and B, of which channel A is illustrative, contains input analog circuitry, which is simply represented as an amplif ier connected to a level detector. With respect to channel A, amplifier 46 25 provides appropriate input filtering of the left velocity sensor signal, level shifting and amplif ication of the resulting shifted signal. Level detectors 48, ef f ectively a windowing comparator, provides a level change on its output signal whenever the output signal 30 produced by amplifier 46 increases above or decreases below a small fixed positive and negative voltage. In this regard, each of these f~hAnnPl ~ provides essentially the same functions as uuLL.~r.JJ~lfl;~fJ circuitry in flow measurement circuit 23 shown in FIG. 2. Channel C shown 35 in FIGs. 3A and 3B contains circuitry represented by amplifier 56 and level detector 58. Reference channel C
is quite similar to f hAnnel q A and B with the exception that level detector 58 contains a single level detector, 2~698 0 93/01472 PCr/US92/OSS83 rather than a windowing comparator, to detect whenever tne output signal from amplifier 56 exceeds a s~all positive voltage level. Multiplexor 31, which is illustratively formed of three separate 2-to-l S multiplexors selectively routes either the left velocity sensor signal appearing on lead l65L or the right velocity sensor signal appearing on lead l65R to the input of each of the three h InnP1~. In this regard, the left and right velocity sensor signals are applied to the 10 first (Ao~ Bo and CO) and second (Al, Bl and Bl) inputs, respectively, of multiplexor 3 l . The status of select signals SO, Sl and S2 specifies whether the right or left velocity sensor signal is applied to the three separate (OA~ OB, and C) outputs of the multiplexor. Select 15 signals 33, formed of signals RPO_A and RPO B connected to select inputs SO and Sl, cause the multiplexor to separately route either the left or right velocity sensor signals as the inputs to -hi~nnP1 ~ A and B, respectively;
while grounded select signal S2 causes multiplexor 31 to 20 continuously route the left velocity sensor signal appearing on lead l65L to the input of reference channel C. Select signals 33 are set by control logic 72 in circuit 70 to perform appropriate input switching.
Circuit 70 contains control logic 72 and timing counters 74, 76 and 78. Circuit 70, preferably formed of a single application specific integrated circuit, is essentially a finite state machine that defines a periodic and repetitively occurring sequence of timing 30 intervals and accompanying states. During each such timing interval, externally applied input signals can start and stop an appropriate timing counter. At the conclusion of that interval, the contents of that timing counter can be read in parallel form for subsequent use.
35 As this circuit applies to flow measurement circuit 30, timing counters 74 and 76, grouped together as counters 75, are used to determine the timing mea~uL~ ~ ~s for channel pairs A-C and B-C, respectively. Timing counter 6~8 WO 93/01472 _34_ PCI/US92/05583 --78 is used to count the frequency value produced by RTD
input circuit 42 for a 6elected analog input signal applied thereto through switch 35. This counter is reset by control logic 72 prior to each conversion interval by 5 applying an appropriate signal being applied to lead 79.
Control logic 72 is formed of well known combinatorial and other logic. After having been initialized with the duration, in tube cycles, of the zeroing and switching intervals, the control logic generates select signals lO over leads 33 to operate multiplexor 31 to select and route the proper waveform sensor signals to the inputs of either channel A or B, as appropriate, such that the channel pairs are repetitively and oppositely cycled through their zero and measurement modes. In addition, lS control logic 72 also generates appropriate control signals which, when applied via leads 77 and 79, properly reset counters 76 and 74 for each timing interval. In addition, the control logic generates, on leads 34, appropriate select signals to the control input (C) of 20 analog switch 35. These select signals cause the switch to route a particular one of its input voltages, namely the RTD voltage appearing on lead 195 or one of two reference voltages (Vref1 or vref2 which are illustratively 1.9 and zero volts, respectively) to an 25 input of RTD input circuit 41 for subsequent conversion by V/F converter 41 situated therein. Reference voltage vref1 is supplied, via lead 38, from reference voltage generator 39 which itself contains a well known highly stable voltage source that exhibits negligible drift with 30 temperature variations. As will be ~; Rcl~cRed hereinbelow and particularly with ref erence to RTD Temperature Processing Routine llOO (discussed in con~unction with FIG. 11), the V/F converter is operated to perform a conversion every .1 seconds with each of eight analog 35 voltages (of which only those three that are relevant to the present invention being specifically shown and discussed herein) applied to the inputs (Io~ I1 and I2 for the three voltages shown) of analog switch 35 being 2 ~ 8 ` i. ' /0 93/01472 _35_ PCr/US92/05583 selected, on a time staggered basis, once every . 8 seconds for conversion into a ~!oL-~ onding frequency value. Control logic 72 specifies which one of the input voltages to analog switch 35 is to be selected at any one 5 time. The states of circuit 70 are described in considerable detail below in conjunction with state table 400 and timing diagram 500 which are respectively shown in FIGs. 4 and 5.
As circuit 70 cycles through its different states -- of which there are eight in total, this circuit writes the value o~ its current state into an internal register (not shown) which, when accessed by microcomputer 80, applies this value onto leads 85. The 15 miuLo-- _Ler then reads this value which, in turn, permits it to appropriately process the counted values provided by counters 75 and 78, via CuLL~ ;n~
internal registers (not shown) and leads 87 and 88.
Leads 87 supply raw timing measurements, designated 20 RAN_RATE_A and RAN_RATE_B, to miuL.: _Ler 80 for channel pairs A-C and B-C, respectively . Dprpnfl i ng upon the mode in which each channel pair is operating, RAW_RATE_A and RAN_RATE_B will each provide, in terms of counts, a single inter-channel phase difference 25 measurement or a single ~t value for each channel pair.
Leads 88 provide the miuL~ _Ler with the counted frequency mea~;uLI - ~ data for the RTD and reference voltages. In addition, logic 72 also writes a value into another internal register (not specif ically shown) which 30 specifies which analog voltage is then being selected by z~nalog switch 35 for conversion by RTD input circuit 42.
This value is also read, via leads 85, by microcomputer 80 .
Fur~hP _e, the miuLuc _Ler applies c~yLu~Iiate signals onto leads 84 to control the overall operation of circuit 70. The mi-;Lu~ uLer also provides appropriate address signals, via leads 82, to designate 2~ 11698 WO 93/01472 -36- PCI`/US92/05583 --to control logic 72 a specif ic internal register from which the microcomputer is to read data or into which it will write data s The mi~ er is also connected, via leads 9l and 93, to respectively well known output circuitry 90 which provides a number of standard outputs (such as illustratively a display interface(s), communication ports, 4-20mA output lead 263 and scaled frequency output lead 262) over leads 26, and to well known input circuitry 95 which provides the meter with interfaces to a number of well known input devices (such as switches, user keypads, communication ports and the like) Nicrocomputer 80 utilizes any one of many well known commercially available microprocessors (not specifically shown) along with sufficient random access memory (RAN) 83 for data storage and sufficient read only memory (ROM) 86 for program and constant storage Tn;-l rh as this program utilizes an event-driven task architecture, a database is implemented within the mi~ er to facilitate easy transfer and sharing of measured and calculated data among the various tasks Based upon its input information, specifically the timing 2s measurements, containing the inter-channel pha6e difference mea,.u. Ls and ~t values for each pair of rh;lnn~lq, and the counted frequency data along with the state information -- all of which are supplied by circuit 70, microcomputer 80 appropriately corrects the measured ~t values produced by each channel pair to account for the measured internal phase delay therefor, determines an accurate temperature compensated meter f actor and thereafter, using the corrected ~t values and this factor, determines the current mass flow and mass flow rate values -- all of which i5 ~ CI~q5F'~ in greater detail below in conjunction with Flow Mea~uL~ L BaGic Nain Loop 600 shown in FIGs 6A and 6B, Zero Determination Routine 700 shown in FIGs 7A and 7B, 2 ~ 8 10 93/01472 PCr~US92/05~83 Mechanical Zero Routine 800 shown in FIGs. 8A and 8B, and RTD Temperature Processing Routine llO0 shown in FIG. lI.
To provide a thorough understanding of the S interactions between circuit 70 and miuLùC _Ler 80, this discussion will now address timing diagram 400 and state table 500 shown in FIGs. 4A, 4B and 5 which collectively detail the functions provided by circuit 70 and their temporal relationship. To facilitate 10 understanding, the reader should simultaneously refer to FIGs. 4A, 4B and 5 tll~uuy}lu~lL the following discussion.
Timing diagram 400 shown in FIGs. 4A and 4B
defines the normal sequential modal operations for each 15 of the channel pairs and the temporal relationships therebetween .
As described above, each of the channel pairs, A-C and B-C, operates in either a r~~' ~ ~ mode or a 20 zero mode. While one channel pair operates in the measurement mode, the other operates in the zero mode with these operations reversing at the end of these modes. The duration of each of these modes (the "modal"
interval) is always the same, i.e. time "t". In this 25 regard, zero mode 410 for channel pair A-C and measurement mode 420 for channel pair B-C simultaneously operate, as do mea,,,ll~ t. mode 440 and zero mode 450, zero mode 470 and mea~.L- L mode 480 for channel pairs A-C and B-C, respectively. Arrows 430, 460 and 490 30 signify mode reversal between the channel pairs at the conclusion of three successive modal intervals.
Channel C is continuously supplied with the left tL) velocity sensor signal and serves as the 35 reference channel with respect to which the internal phase delay of each of the other two ~h;~nn~l E; is continually measured. However, the input signals applied to non-reference ch~nnel~ A and B are switched, ~Pp~n~l;n~
-21~1S~8 upon the mode of corrPcrnnrlin~ channel pair A-C and B-C, between the left and right (R) velocity sensor signals with phase difference mea~uL~ s being taken for each di~ferent input configuration to yield inter-channel S phase difference mea2iuL~ Ls or ~t values for each pair.
In particular, while a channel pair operates in the mea:,uL~ ~ mode, the non-reference channel in that pair, e.g. channel A for pair A-C, is supplied with the 10 right velocity sensor signal and meaDuL~ Ls are made of the inter-channel phase difference occurring for that pair. These meaauL, Ls provide raw flow based ~t values . These mea~uL Ls occur throughout the entire time "t" that the channel exists in the meaDuL - L mode.
15 During this time, these mea~uL ~ Ls are provided to the mi~L. _Ler for subsequent processing into corrPcpr~nrl; n~ mass f low rate values .
By contrast, four separate f~nrti~nC are 20 performed in the following se~uence for any channel pair, e.g. pair B-C, during its zero mode: (a) switching the input for the non-reference channel in that pair from the right to the left velocity sensor signal during the switching interval, (b) providing meaDuL~ -rLs of the 25 internal phase delay for that channel pair (i.e.
"zeroing") during the zeroing interval, (c) switching the non-reference channel input back to the right velocity sensor signal again during a switching interval, and (d) permitting that pair to be "active" for a zeroing 30 interval during which mea:,uLt -- ~s of flow based ~t values can be made. Since the opposite channel pair, e.g. pair A-C, will be actively measuring flow based ~t values during its measurement interval while channel pair B-C is active, both rhAnnPlc are able to ~UI1~ ULLellLly 35 provide flow based ~t values for the same velocity sensor ~ignals during this "active" interval. If additional error rhPrking is needed, the mi~:L.- _Ler can process the measurements provided by the "active" channel pair ~093/01472 _39_ 2~ PCI/US92/05583 and compare the resulting corrected ~t values against those being provided using the other channel pair. A
sufficient discrepancy therebetween would generally indicate an error condition.
As illustratively shown in FIGs. 4A and 4B, each switching interval is 16 tube cycles in duration, while each zeroing interval occurs over 2048 successive tube cycles. Accordingly, time "t" ~ormed of two interleaved switching and zeroing intervals occurs for 4128 tube cycles. During meter initialization, microcomputer 80, shown in FIGs. 3A and 3B, loads the durations, in terms of tube cycles, of the switching and zeroing intervals into circuit 70 and specifically control logic 72 therein.
As shown in state table 500 depicted in FIG. 5 for circuit 70, this circuit, in normal operation, continuously cycles through eight states in sequence, illustratively designated as states 26, 46, 26, 66, 6A, 6C, 6A and 6E -- of which two states, i.e. states 26 and 6A, are repeated.
Each of these states exists for a f ixed duration, either the switching interval or the zeroing interval. During all eight states, the left velocity sensor signal is continuously applied to the input of reference channel C.
For the first four states (states 26, 46, 26 and 66), channel pair A-C operates in the measurement mode (hereinafter referred to as the channel A
measurement mode) while channel pair B-C concurrently operates in its zero mode (hereinafter referred to as the Channel B zero mode). Throughout the channel A
mea~uL~ ~ mode, circuit 70 generates a low level on multiplexor select signal RP0_A such that the right velocity sensor signal is continuously applied to the .

2 ~ g W093/01472 PCr/US92/05583--input of channel A. During this mode, as indicated by the letter "X", channel pair A-C provides flow based ~t values and hence serves as the measuring channel pair.
In addition, at the beginning of state 26, circuit 70 s - the beginning of the channel B zero mode by initially applying a high level to multiplexor select signal RPO_B in order to f irst switch the channel B input from the right to the left velocity sensor signal. ~his ----~ Channel B Switching state 26 during which channel pair B-C undertakes no mea,,uL ~ Ls but merely affords an adequate period of time, i.e. switching interval t8W, for all switching transients and similar perturbations on channel B to settle out. Once this state is completed, circuit 70 invokes Channel Pair B-C
Zeroing state 46. During state 46, which lasts for zeroing interval tZERo~ inter-channel pha6e difference meaDuL~ ~s are continually made by circuit 70 for channel pair B-C. These meaDuL ~ ~s are read and averaged by the mi~;L~L~CessoI to yield a mea~uL~ L, in counts, of the internal phase delay for that channel pair. At the conclusion of the zeroing interval, Channel B Switching state 26 rc O~ ULI~ to switch the input of channel B from the left velocity sensor signal back to the right velocity sensor signal. To do so, circuit 70 generates a low level on multiplexor select signal RPO_B.
Again, this state, during which no meaDuL. ~s are made on channel pair B-C, remains in existence for the switching interval in order to allow all switching transients and the like on channel B to settle out. At the conclusion of state 26, Both rhAnn~1~ Active state 66 occurs for a zeroing interval during which both ~hAnn~
are "active" and flow based ~t measurements can be made, if desired, through channel pair B-C in addition to those meat,uL Ls simultaneously occurring through channel pair A-C. At the conclusion of state 66, states 6A, 6C, 6A and 6E occur in sequence which merely provide the same operations but on the opposite channel pairs. All the states then repeat Im seri~tim, and so on.

93/01 472 ~ ~ ~ 16 ~ 8 PCJ/US92/OSS83 ~0 -4 1-B. Software Description With the above understanding in mind, the S ~li Cc~lcsion Will now address various aspects of the software executed by microcomputer 80 shown in FIGs. 3A
and 3B. Inasmuch as the mi-L, _Lel performs a number of well known administrative and control fUnrt; ~nC which are not relevant to the present invention -- such as lO providing a database manager and an a~L CI~L iate operating system environment for a task based application program, then, to simplify the following discussion, all of these functions and the accompanying software therefor have been omitted heref rom .
FIGs. 6A and 6B collectively depict a simplified flowchart of Flow Mea~uL~ 1~ Basic Main Loop 600; the correct alignment of the drawing sheets for these f igures is shown in FIG . 6 . This routine provides 20 the basic flow mea~uL L functions.
Upon entry into routine 600, execution plvceeds to block 610 which reads current raw phase difference mea~uL ~ ~ data (RAW_RATE_A and RAW RATE_B) and state 25 information from circuit 70. DPpDn~l;n~ upon the current mode of each channel pair, RAW_RATE_A and RAW_RATE_B will each provide, in counts, either a single interchannel phase difference measurement or a single ~t value. After block 610 executes, block 620 is executed. This block 30 executes Zero Determination Routine 700 which, in response to the raw phase difference measurements and state information and as ll;crll_se~ in detail below, processes the phase difference data for the channel pair that is currently operating in the measurement mode as a 35 flow based ~t value and processes the phase difference data for the other channel pair as an inter-channel phase difference measurement. This mea,uL~ t is used by this routine to determine the electronic zero value for that 211~5~8 WO 93/01472 -42- PCI/~IS92/05583 latter channel pair. The electronic zero consists of two values, namely the internal phase delay, expressed in the same counts as ~t, as60ciated with each of the two channel pairs. Thereafter, routine 700 det~rm;n~ the 5 mechanical zero for the Coriolis meter. The mechanical zero is an offset value in the ~t mea:-uLI ~s that is obtained, as described below, during a zero flow condition occurring during meter calibration. After these operations are completed, routine 700 then corrects 10 the current ~t value measured for the channel pair operating in the mea~iuL~ 1 mode by the r- '~nic~l zero for the meter and by the most current electronic zero value for that pair -- this electronic zero value having been previously determined while that pair was last 15 operating in its zero mode.
After routine 700 has fully executed, execution ~Loceeds from block 620 to 630. The latter block, when executed, filters the corrected ~t value produced by 20 block 620 through a double pole software filter to remove noise and the like thereby yielding a current filtered ~t value. Execution next proceeds to block 640 which calculates the current volumetric and mass flow rates using the current f iltered ~t value and the temperature 25 corrected rate factor. This, aLuLe factor is updated on a periodic basis through RTD Temperature Processing Routine 1100 which, as described in detail below, executes on an interrupt basis.
Upon completion of block 640, block 650 is executed. This latter block tests the volumetric and mass flow rate values against coLLeayo~lding low flow (cutoff) limit conditions and, if these conditions are met, temporarily sets the volumetric and mass flow rates to zero. Thereafter, execution proceeds to block 660 which, when executed, stores the current volumetric and mass flow values in the database for subsequent use, such as for periodic updating of the displays, totalized flow ....

93/01472 ~ ~ ~16 ~ ~ PCr/US92/05583 readings and/or meter outputs. Execution then loops back to routine 610 and so on.
- A flowchart of Zero Determination Routine 700 5 is collectively depicted in FIGs. 7A and 7B for which the correct A l;; - t of the drawing sheets is shown in FIG .
7. This routine contains four 6eparate sections:
Electronic Zero Determination Routine 710, Electronic Zero C -n~Ation Routine 760, M~rhAn;cAl Zero 10 DetPrminAtion Routine 780, and M~r~hAniCAl Zero C ~ tion Routine 790. As generally discussed above, routine 700, 6pecifically through routine 710, determines the current flow based ~t value for the channel pair currently operating in the mea~uL L mode and 15 det~rm;nP~ the current electronic zero value for the other channel pair operating in its zero mode. Routine 760: -~tes each current mea6ured ~t value from the channel pair operating in the measurement mode by the most recent electronic zero value for that channel.
20 Routine 780 det~rm; n~ the mechanical zero for the meter.
Finally, routine 790 c.uLle~ ~s the flow based ~t value for the current channel pair operating in its mea~uL L
mode by the mechanical zero value f or the meter .
Specifically, upon entry into routine 700 and 6pecifically into routine 710, execution first proceeds to decision block 703. This block determines whether the value of variable STATE indicates that channel pair A-C
is zeroing, i.e. the state of circuit 70 i6 given by the value "6C" (see FIG. 5). This value is provided by circuit 70 upon inquiry by mi~;LU~L~''es~L 80 (see FIGs.
3A and 3B). In the event that this state is now occurring, then execution proceeds, via the YES path emanating from decision block 703 as shown in FIGs. 7A
and 7B, to block 706. This latter block, when executed, updates the value of a totalized rate variable (TOTAL_ RATE) with the current value of RAW_RATE_A. As will be seen at the conclusion of the zeroing interval, 2~ 1169~
O 93/01472 PCr/US92/0~83 this totalized rate value is set equal to zero. Next, block 709 is executed to set the state of a temporary flag (TEMP_STATE) to a value (ZEROING_CHANNEL A) that signif ies that channel pair A-C is presently undergoing 5 zeroing . Once this occurs, execution yl uceeds to block 712 to merely in~L~ ~ the value of a loop counter (COUNTER) by one. Execution then proceeds to ~l~c~ Si~n block 7 3 0 . Alternatively, in the event that the current value of variable STATE indicates that channel pair A-C
10 i8 not zeroing, then execution ~-oceeds, via the NO path emanating from decision block 703, to decision block 715.
The latter decision block tests the state of the temporary flag to determine if zeroing has just terminated for channel pair A-C, i.e. whether the value l5 of this f lag still equals ZEROING CHANNEL_A . In the event that zeroing has just terminated for this channel pair, then decision block 715 routes execution, via its YES path, to block 718. This latter block, when executed, calculates the electronic zero value for 20 channel pair A-C, i.e. ELECT_ZERO A, as a simple average value of the separate mea,,uL~ Ls that have been totalized, specifically the value of the variable TOTAL_RATE divided by the contents of loop counter COUNTER. Once this has oc. uLL~d, execution ploceeds to 25 block 721 which sets the value of the temporary flag to another value, here NOT_ZEROING_CHANNEL A, that signifies that channel pair A-C is not undergoing zeroing.
Thereafter, execution proceeds to block 724 which merely resets the values of both the loop counter and the 30 totalized rate variable to zero. Execution then proceeds to decision block 730. Alternatively, execution also ~ ceeds to this decision block, via the NO path emanating from decision block 715, in the event that channel pair A-C has not been and has not just completed 35 zeroing.
Blocks 730 through 751 provide the same operations as do blocks 703-724 but for determining the 21t 1~98 ~0 93/01472 PCI/VS92/05583 value of the electronic zero for channel pair B-C, i.e.
ELECT_ZERO_B. Specifically, decision block 730 cletPr~nin~c whether the value of variable STATE indicates that channel pair B-C is zeroing, i.e. the state of S circuit 70 is given by the value "46" (see FIG. 5). In - the event that this state is now occurring, then execution pLuceeds, via the YES path emanating from decision block 730 as shown in FIGs. 7A and 7B, to block 733. This latter block, when executed, updates the 10 value of the totalized rate variable, TOTAL_RATE, with the current value of RAW_RATE B. As will be seen at the conclusion of this zeroing interval, this totalized rate value is set equal to zero. Next, block 736 is executed to set the state of the t~ ~ILy flag, TEMP_STATE, to a 15 value (ZEROING_CHANNEL_B) that signifies that channel pair B-C is presently undergoing zeroing. Once this occurs, execution ~L oceeds to block 7 3 9 to merely in- L. ~ l. the value of the loop counter, COUNTER, by one.
Execution then pL-aceeds to routine 760. Alternatively, 20 in the event that the current value of variable STATE
indicates that channel pair B-C is not zeroing, then execution ~Luceeds, via the NO path emanating from decision block 730, to decision block 742. The latter decision block tests the state of the temporary flag to 25 determine if zeroing has just terminated for channel pair B-C, i.e. whether the value of this flag still equals ZEROING_CI~ANNEL B. In the event that zeroing has just terminated for this channel pair, then decision block 742 routes execution, via its YES path, to block 30 745. This latter block, when executed, calculates the electronic zero value for channel pair B-C, i.e.
ELECT_ZERO_B, as a simple average value of the separate measurements that have been totalized, specifically the value of the variable TOTAL_RATE divided by the contents ~5 of loop counter COUNTER. Once this has OC1ULLC:d, execution proceeds to block 748 which sets the value of the temporary flag to another value, here NOT_ZEROING_C~ANNEL_B, that signifies that channel -:

21~ ~6~8 WO 93/01472 -4 6- PCr/US92/05583 pair B-C is not undergoing zeroing. Thereafter, execution ~loceeds to block 751 which merely resets the values of both the loop counter and the totalized rate variable to zero. Execution then proceeds to routine 5 760. Alternatively, execution also proceeds to this routine in the event that channel pair B-C has not been and has not just completed zeroing, i.e. via the N0 path emanating from decision block 742. At this point, routine 710 has completed execution. Tn~l as one of 10 the channel pairs i6 operating in its zero mode at any one time, then the current value of the COL L ~ 1 i n~
variable ELECT_ZER0_A or ELECT_ZER0_B is being rll~t~-rmin/~c~
at that time with the appropriate steps, as described above, being executed therefor.
Electronic Zero C -nc~tion Routine 760 merely corrects ( ~tes) the current ~t meaDuL~ L by the electronic zero value for the particular channel pair that ~lo-luced that mea~,uL~ L. Specifically, upon entry 20 into this routine, execution proceeds to decision block 763 which, based upon whether channel pair B-C or channel pair A-C is currently operating in its measurement mode, respeetively routes execution to block 767 or 769. In the event that execution is routed to block 767, then, 25 this block, when executed, subtracts the electronic zero value for channel pair B-C from RAW_RATE_B and stores the result in variable ~t. Alternately, if execution is routed to block 769, then, this block, when executed subtracts the electronic zero value for channel pair A_C
30 from RAW_RATE_A and stores the result in variable Qt.
After either block 767 or 769 has executed, execution proceeds to Mechanical Zero Determination Routine 780.
Routine 780 det~ n~C the current value 35 mechanical zero value for the meter. Specifically, upon entering routine 780, execution proceeds to decision block 781. This block, when executed, ~1PtorminF-C if a current mechanical zero value is to be found. As noted 93/01472 2 1 1 1 6 ~ g PCr/US92/05583 ~O -47-above, a mechanical zero is de~Prmin~l under no flow conditions during meter calibration. If meter calibration is currently being performed and if a user indicates that no f low is occurring by depressing an 5 appropriate pushbutton on the meter electronics, then - decision block 781 routes execution, via its YES path, to block 784. This latter block executes MPrhAnicAl Zero Routine 800, as discussed in detail below, to determine the current mechanical zero value (MECH_ZERO) for the meter. Once this value has been determined, execution proceeds to Mechanical Zero r ^ncation Routine 790.
Execution also proceeds to routine 790, via the NO path emanating from decision block 781, in the event that meter calibration is not occurring or if it is that the user has not specif ied that no f low is occurring .
Mechanical Zero Routine 790 contains block 792, which, when executed, merely subtracts the current mechanical zero value, MECH_ZERO, from the value of variable ~t with the result being a corrected ~t meaau I t which will be s~lhcPq~lpntly f iltered and used by main loop 600 (specifically blocks 630 and 640 therein as shown in FIGs. 6A and 6B) to determine the current value for mass flow rate. Once block 792 has executed, execution exits from routines 790 and 700, as shown in FIGs . 7A and 7B, and returns to Flow MeaauL ~ L Basic Main Loop 600.
To simplify the software, routine 700 does not include appropriate software for detPrminin~
corrPcpon~l i n~ corrected ~t values for both channel pairs during each "active" interval and, as fli cc~lcced above, comparing the results to detect suf f icient discrepancies therebetween and system errors associated therewith.
Routine 700 can be readily modified by any one skilled in the art to include this software.

2~ ~ ~ 6~
WO 93/01~72 PCI`/US92/OS583 FIGs. 8A and 8B collectively depict a flowchart of ~lechanical Zero Routine 800; the correct alignment of the drawing sheets for this figures is shown in FIG. 8.
As discussed above, routine 800 detPrm;nDc the current 5 value for the Ah~n;~A~1 zero of the meter. In essence and as ~; R~A11cRF-A above, the current value of this zero is detPrm;nP~l by first calculating the standard deviation, ~t, of the Qt values obtained for a no flow condition during meter calibration. This standard deviation 10 provides a measure of the noise appearing on the ~t mea:,u, ~ ts at a no flow condition. Only if the noise is sufficiently low, i.e. the value of the standard deviation is below a minimum threshold value, will the most recent value for the mechanical zero be updated to 15 reflect its current value; otherwise, this current value will simply be ignored. The number of measured ~t values used in determining the standard deviation is ~uveL--ed by any one of three criteria: (a) when the "running"
standard deviation decreases below a ~..v~:~ ye,.ce limit, 20 (b) a user terminates the mechanical zeroing by depressing an appropriate pushbutton, or (c) if a pre-def ined number of measured ~t values has been taken .
In addition, appropriate limits checks are made to ensure that the current value of the mechanical zero lies within 25 pre-def ined bounds prior to replacing its most recent value with its current value.
SpPr;f;~A~lly, upon entry into routine 800, execution proceeds to decision block 803. This block, 30 when executed, tests the status of a f lag ( ZERO STATE) to specify whether the process of detPrm; n; n~ a mechanical zero is currently occurring. This flag is set by JlUp. iate software (not shown) to _ -~rP this process. In the event that this process is underway, 35 decision block 803 routes execution, via its YES path, to block 806. This latter block, when executed, updates the value of a totalized variable (ZERO_TOTAL) with the current ~t value. As will be seeL ~t--r' th~ to~liz--cl 93/01472 2 ~ 9 8 PCr/US92/05583 value is set equal to zero at the conclusion of the zeroing interval. Once block 806 has executed, execution proceeds to block 809 to in~L~ L the contents of a loop counter, ZERO_COUNT, by one. Thereafter, execution 5 proceeds to decision block 820. Alternatively, if a - chAnic~l zero value is not currently being d~t~rm;n~d, i.e the status of the ZERO STATE flag is now not active, then execution proceeds, via the NO path emanating from decision block 803, to block 812. This latter block l0 resets the ZERO_STATE flag to the active state, sets the values of both the ZERO TOTAL and loop counter ZERO_COUNT
to zero, and set the value of variable, MIN_STD_DEV, to a large predef ined number (the exact value of which is not critical as long as it is well in excess of the expected 15 value of the standard deviation). Thereafter, block 816 is executed to reset all the error flags that are associated with the mechanical zero process. After this occurs, execution proceeds to decision block 820.
Decision block 820, when executed, det~rm;n-~c whether a minimum number of measured ~t values has oc:~u~L~d to determine a ` ~n;c~l zero value -- i.e.
specifically whether the current value of loop counter ZERO_COUNT exceeds a predefined minimum value, MIN_ZERO_COUNT which typically equals the decimal value "100". In the event that an insufficient number of ~t values has oc~;u~ Led, then execution exits from routine 800, via path 872 and NO path 822 emanating from decision block 820. Alternatively, if a minimum number of l~t values has occurred, then decision block 820 routes execution, via its YES path, to block 823. This latter block, when executed, updates the standard deviation, ~, of the ~t values that have been currently measured thusfar for use in determining a mechanical zero value and stores the result in variable STD_DEV. Once this occurs, execution proceeds to decision block 826 which tests the resulting standard deviation value against a minimum value therefor. In the event that the resulting 211~698 WO 93/01472 PCr/US92/05583 ~

standard deviation is less then the minimum value, decision block 826 routes execution, via its YES path, to block 829. This latter block calculates a temporary current value for the mechanical zero (MECH_ZERO_TEMP) as being an average of the totalized ~t values obtained thusfar during the current mechanical zero proce6s, i.e.
the value of ZERO_TOTAL divided by the contents of loop counter ZERO_COUNT. Once this occurs, block 829 sets a minimum standard deviation value equal to the current value of the standard deviation. By doing 80, the minimum value of the standard deviation that has been determined thusfar for this current mechanical zero process will always be used, in the manner ~i cc1-ccPd below, to deter~ine whether the current value of the mechanical zero is too noisy and hence unaccept2ble.
Once block 829 fully executes, execution ~lvceeds to deci6ion block 832. Alternatively, execution also proceeds to this decision block, via the NO path emanating from decision block 826, in the event that the current value of the standard deviation now equals or exceeds its minimum value.
At this point, up to three separate tests are undertaken in seriatim through decision blocks 832, 836 and 840 to determine if a sufficient number of measured ~t values has been taken to determine the current mechanical zero value. Such mea~,uL~ ts continue until a sufficient number has o~ u,,ed. In particular, decision block 832 detPrmin~ whether the current value of the standard deviation is less than a c~,-,vt:,y~nce limit. In this case, if the standard deviation has been falling with successive ~t values and has fallen below a predefined limit value, then it is very unlikely that any additional mea~ l.s will adversely impact the ~-h:~n;~ill zero value. Accordingly, if the standard deviation has decreased in this manner, than decision block 832 routes execution, via its YES path, to decision block 843. Alternatively if t~e current v,llue of the 2~1~ 698 0 93/01472 PCr/US92/05583 standard deviation is still higher than the c~lvt:ly~:nce limit, then execution proceeds, via the NO path emanating from decision block 832, to decision block 836. This latter decision block det~rmi nP5 whether the user has 5 depressed a pushbutton or otherwise provided an ~p~JL OIJ~ iate indication to the meter to terminate the current mechanical zero process. In the event that the user terminated this process, then decision block 836 routes execution, via its YES path, to decision block 843. Alternatively, if the user has not terminated the current mechanical zero process, then decision block 836 routes execution, via its NO path, to decision block 840.
Decision block 840, when executed, determines whether a maximum number, MAX COUNT, of the measured Qt values has just ûc~uLLed. In the event that this maximum number of mea~uL~ Ls, e.g. 2000 meaLuL~ ~s, has o~-;u.Led, then decision block 840 routes execution, via its YES path, to decision block 843. Alternatively, if the maximum number of such mea:.uL~ ~s has not occurred, then execution exits from routine 800, via NO path 841 emanating from decision block 840 and via path 872, in order to )L UpL iately process the next successive Qt mea:,uL ~ t .
At this point in routine 800, a current value, though temporary, for the r-^hAn;rAl zero has been det~rm;nPd based upon a sufficient number of s-lo~Pcs;ve Qt measurements. Decision blocks 843, 846 and 849 now determine whether this m--hAn; rAl zero value lies within predefined limits, e.g. illustratively +311sec, and whether this mechanical zero value is relatively noise-free. Specifically, der;c;~ block 843 detPrm;nPc whether the current temporary r ~ ^hAn i C~ 1 zero value is less than a lower limit, i . e. illustratively -311sec. In the event that this limit is negatively ~Yre~ded, then 3s decision block 843 routes execution, via its YES path, to block 854. Since this signifies an error condition, block 854, when executed, sets the value o~ an ~Lu,uLiate error flag, i.e. MECHANICAL ZERO TOO LOW, to 21~ ~98 WO 93/01472 PCr/US92/0~583 true. Alternatively, if the lower limit is not negatively ~Y~-e~ d, then ~ i Rit~n block 843 routes execution, via its N0 path, to rl~ri ~ n block 846 . This latter decision block determines whether the current 5 temporary mechanical zero value is greater than an upper limit, i. e. illustratively +311sec. In the event that this limit is positively ~Yt ee~F~d, then rleci~:ion block 846 routes execution, via its NO path, to block 859.
Since this signifies an error condition, block 859, when lO executed, sets the value of an appropriate error flag, i . e. MECHANICAL ZERO TOO HIGH, to true. I'he upper and lower +311sec limit values were det~rmi n~d empirically as being those values within which all no-flow based ~t values should lie for meters that are currently l5 manufactured by the present assignee. Alternatively, if neither of these limits is ~Y~ d, then decision block 846 routes execution, via its N0 path, to decision block 851. This latter decision block tl~t~rmi n~ whether the temporary mechanical zero value is sufficiently 20 noise-free, i.e. whether all the successive Qt values that are utilized to generate this value possess less than a given variability, by comparing the present minimum standard deviation value against a limit equal to a preset integer multiple ("n") of, typically twice, the 25 convergence limit.
In this regard, the most repeatable value for the mechanical zero tends to occur when the standard deviation reaches its minimum value. It appears that 30 this occurs because the measured ~t values will be corrupted by periodic noise, such as 60 Hz hum and its harmonics, that beats against the sampling rate of the velocity sensor signals ( i . e . counters 75 are read once every tube cycle) thereby creating beat frequencies that 35 appear in the measured l~t values. In normal operation, I
expect that some noise of this type will always be present, though the amplitude of the noise will usually vary from one installation to another. For the range of ~. ~ ~.

2~11698 '0 93/014~2 PCr/US92/0~583 meters manufactured by the present assignee, the velocity signals have fl~nlli tal frequencies in the range of 30-180 Hz. The amplitude of the beat frequencies will be lowest when the noise is in-phase with this 1; n~ rate 5 and will increase as the noise gets ~Lo~L~:s6ively out-of-phase with the sampling rate thereby leading to increased variability and error in the measured no flow ~t values. Hence, the minimum value of the standard deviation is used to determine whether the resulting 10 r- '-nicAl value will be too noisy. Specifically, if decision block 851 detDrmi n~C that the minimum standard deviation exceeds the limit of "n" times the cc,.,v~Ly~ ce limit, then the current temporary mechanical zero value is simply too noisy and is ignored. Since this signif ies lS an error condition, decision block 851 routes execution, via its YES path, to block 862. This latter block, when executed, sets the value of an a~u~liate error flag, i.e. MECHANICAL ZERO TOO NOISY, to true. Alternatively, if the minimum standard deviation is sufficiently low, 20 hence indicating that the t~ - aLy J- -n;cAl zero value, i5 relatively noise-free, then decision block 851 routes execution, via its NO path, to block 865. This latter block updates the -niCAl zero value, MECH_ZERO, as being equal to the value of the temporary 2~ mechanical zero, MECH ZERO_TEMP. Once block 854, 859, 862 or 865 has executed, execution proceeds to block 870 which, in turn, sets the state of flag ZERO_STATE to inactive to reflect that the ---hAn;cAl zero process has been terminated and is now not in progress. Once this 30 occurs, execution then exits from routine 800.
Having described the mechanical zero process, FIG. 9 diagrammatically shows the associated zeroing operations that occur for each COLL~ J~1;n~ range in the 35 standard deviation, ~t, that can be obtained during this process. Specifically, whenever the value of t~t lies within region 910 and hence is less than the CU.IV~:L Je"ce limit (1), zeroing immediately stops and the resulting -~t ~ l~g~
WO g3/01472 _54_ PCI/US92/05583 mechanical zero value i5 accepted- For any value of ~t lying within region 920 and hence greater than the C~ V~L~ Ce limit but less than "n" times that limit, zeroing continues until a maximum number, as given by the s value of variable MAX_COI~NT, of ~t mea~uL- --ts has OC~.;uL L t d . This number, in tube -cycles, def ines a maximum zeroing interval. For any value f C~.~t that lies within region 930 and hence exceeds "n" times the cu..v~L~c:nce limit, zeroing immediately halts. The associated current l0 mechanical zero value is simply ignored in favor of its most recent value.
FIG. 10 diayl tically shows the ranges of acceptable and non-acceptable mechanical zero values. As ~5 shown, erroneous I -ch~nic~l zero values are those which lie either within region 1020 and hence are negatively greater than the negative limit of -311sec or those which lie within region 1030 and are positively greater than the positive limit of +311sec. If the mechanical zero is 20 det-~rm;nP~l as having any of these values, that value is simply ignored. Only those values for the -h~nir~ll zero that lie within region 1010 and hence are situated between the negative and positive limits are accepted.
FIG. 11 shows a flowchart of RTD T G~UL~:
Proc~ccin~ Routine 1100. As riiccllccP~l above, this routine operates on a periodic interrupt basis, every .8 seconds, to provide a digitized flow tube t~ aL
value that is essentially insensitive to temperature drift of the RTD and, using that value, calculates a current value for the tO...~e:Lc~LuLe r -ncated meter ~actor (RF). This value is then stored in the database within the mi~;L~ Ler for subsequent use by routine 600 in det~nm;nin~ a current mass flow rate value.
Upon entry into routine 1100, execution proceeds to block 1110. This block, when executed, causes analog switch 35 to route the RTD voltage to the J~'O 93/01472 _55- 2 ~ 1 ~ 6 9 ~ PCr/US92/05583 input of V/F converter 41 (see FIGs. 3A and 3B) for subsequent conversion. To speci~ically effectuate this, mi~:Lo~ucessuI 80 applies suitable address and control - signals, via leads 82 and 84, to circuit 70 and 5 particularly to control logic 72 situated therein. These signals, in turn, instruct that logic to apply the appropriate select signals over leads 34 to the analog switch. After this occurs and an d~lU~,Liate col~ntin~
interval has elapsed, block 1110, shown in FIG. 11, reads lO the contents of counter 78, shown in FIGs. 3A and 3B, which contains a counted value proportional to the frequency converted analog RTD voltage. Thereafter, as shown in FIG. 11, execution proceeds to block 1120. This block, when executed, filters the contents that have been ~5 read from counter 78 through a two-pole software f ilter and stores the resulting f iltered value in temporary variable V_T0_F.
After this occurs, block 1130 is executed which 20 eliminates a zero offset value from the filtered value to yield a current ~LeuU~IlUy value, CURRENT_FREQ. This zero offset value, FREQ_AT_OV, is a non-zero filtered counted frequency output value pluduced by the V/F converter with a zero input voltage (Vref1) applied thereto.
25 Thereafter, block 1140 is executed to calculate a proportionality factor, FREQ_PER_C, that specifies the number of counts per degree C. This factor is simply given by the difference in the filtered counted value~
for two reference voltages (Vref1 and vref2 which are 30 illustratively ground potential and 1. 9V, respectively) divided by the decimal number "380". Since the counted frequency values for both reference voltages are obtained essentially contemporaneously with any change in the flow tube temperature, then any t~ Lu~ e drift produced by5 the V/F converter will inject an essentially equal error t into both of these counted values. TnA~ C-h as the proportionality factor is calculated using the dif f erence between these counted values rather than the WO93/01472~ 69~ -56- PCl/USg2/05583 magnitude of either value alone, the value of the proportionality factor will be essentially unaffected by any shift in the counted V/F output attributable to temperature drift. The zero offset value (FREQ_AT_OV) S and the filtered counted l.9V reference value (FREQ_AT_l.9V) are both ~Pt~rmin~d on a periodic interrupt basis, again every .8 seconds, by another routine not shown. This routine, which would be readily apparent to anyone skilled in the art, causes circuit 70 l0 to apply appropriate select signals to the analog switch to first route on a time staggered basis either the ground potential (Vref1) or l.9V (Vref2) to the input of V/F converter 41, and then sequentially count the frequency value produced therefrom and thereafter read 15 and filter this value and store the filtered results.
Once the proportionality factor has been determined by block 1140, execution ~L.,ceeds to block 1150. This block calculates the current temperature 20 (TEMP) sensed by the RTD by dividing the current frequency value by the proportionality factor.
Thereafter, execution proceeds to block 1160 which calculates the t~ c-Lule ~ted meter factor RF
using a meter factor value and the current t ~ ~ItUL-`
25 value. For a Coriolis meter, its meter factor is a known constant that is det~in~d empirically during manufacture. Once this temperature ~ ed meter factor is calculated, it is stored in the database for subsequent use in det~rminin~ mass flow rate. Execution 30 then exits f rom routine 1100 .
Those 6killed in the art will now certainly realize that although both channel pairs are operated in parallel, such that one pair is operating in its zero 35 mode while the other pair is operating in its mea~u mode, these channel pairs could be sequentially operated.
In this instance, an operating channel pair would function in its zero and/or mea~uL L modes while the - -~ 93/01472 2 ~11 6 g 8 PCr/US92/05583 other channel pair l~ ; ned in a standby status . The channel pairs could then periodically switch from operating to standby status at the conclusion of each mode or after the operating channel pair sequentially 5 undertook both its zero and mea,,uL~ modes. Since, with sequential operation, one channel pair is always in the standby status at any one time, then, to simplify the circuitry, one channel pair rather than two could be used with that one pair always operating and continuously 10 cycling between its mea~ul~ L and zero modes. In those instances when the one effectively operating channel pair ~=
is undertaking its zero mode, no flow mea~u~ ~s would then be being made. Accordingly, an assumption, in lieu of actual flow mea~ur ~ s, would need to be made 15 regarding the f low that is occurring during that time .
Hence, by eliminating continuous flow mea~;uL ~s, the use of effectively only one operating channel pair at any one time in a Coriolis flow meter -- regardless of whether the meter contains only one physical channel pair 20 that is cycled between its two modes or two pairs with one such pair being inactive at any one time, may provide f low measurements that are somewhat inaccurate . By contrast, since my inventive flow mea~u~l ~ circuit 30 always has one channel pair that, during normal flow 25 metering operations, is actively measuring actual f low at any time, the meter provides very accurate f low measurements at the expense of a slight increase in circuit complexity.
Furthermore, although an "active" interval has been provided within the zero mode for any channel pair during which, for example, dual flow mea~uL. t.s and inter-channel pair comparisons thereof could be made, this interval could be eliminated, if needed, with no 3s adverse affect on meter accuracy. In fact, doing so could either be used to shorten the duration of the zero mode by one zeroing interval ( i . e . the time during which the channel pair _uld other~ise operate in the "active"

2111~98O 93/01472 PCl/US92/05583 interval) or lengthen the time during which that channel pair is actually zeroing by appropriately increasing the number of internal phase delay mea:~u~ Ls that are to be taken then.
Also, those skilled in the art rPcoqni 7e that, although the ~licclo~Pd pmho~;r L utilizes U-shaped flow conduits, flow conduits (tubes) of almost any size and shape may be used as long as the conduits can be 10 oscillated about an axis to establish a non-inertial frame of reference. For example, these conduits may include but are not limited to straight tubes, S-shaped conduits or looped conduits. Moreover, although the meter has been shown as containing two parallel flow 15 tubes, Pmho~ Ls having a single flow tube or more than two parallel flow tubes -- such as three, four or even more -- may be used if desired.
Although a single Pmhorlir L of the invention 20 has been shown and described in detail herein, many other varied Pmho~ Ls that still incorporate the ~Pa~h i n~c of the present invention can be readily fabricated by those skilled in the art.

Claims (22)

I claim:

1. In a Coriolis meter (5) for measuring flow rate of a process fluid flowing therethrough, said meter having at least one flow conduit (130), a method of producing a mechanical zero value for the meter comprising the steps of:
oscillating the conduit while a process fluid to be measured does not flow therethrough;
sensing movement of said conduit and providing first and second signals responsive to said sensed movement;
measuring, in response to said first and second sensor signals, a plurality of successive time periods (.DELTA.t) occurring between corresponding points on the first and second signals while the process fluid does not flow through said conduit so as to form a corresponding plurality of measured no flow .DELTA.t values;
determining a standard deviation of said plurality of measured no flow .DELTA.t values; and producing, in response to said plurality of measured no flow .DELTA.t values and if the standard deviation is less than a pre-defined limit value, a current mechanical zero value for subsequent use in compensating flow based measured .DELTA.t values so as to determine therefrom flow rate of the process fluid then flowing through said meter.

2. The method in claim 1 further comprising the step of determining a number of said measured no flow .DELTA.t values in said plurality and used in determining an intermediate mechanical zero value as being the lesser of either a pre-defined maximum number of measured no flow .DELTA.t values or a total number of the measured no flow .DELTA.t values that have occurred prior to the standard deviation thereof obtaining a value less than a predefined convergence limit.

3. The method in claim 2 wherein said time periods measuring step comprises the step of measuring at least a predefined minimum number of the successive time periods such that said plurality of measured no flow .DELTA.t values contains a corresponding minimum number of values.

4. The method in claim 3 wherein said current mechanical zero producing step comprises the step of calculating said intermediate value as an average of the plurality of measured no flow .DELTA.t values.

5. The method in claim 4 wherein said determining step comprises the step of updating, in response to the occurrence of each successive measured no flow .DELTA.t value, the standard deviation of said plurality of measured no flow .DELTA.t values using said successive measured no flow .DELTA.t value.

6. The method in claim 5 wherein the pre-defined limit is an integer multiple of said convergence limit.

7. The method in claim 3 wherein said current mechanical zero producing step comprises the steps of:
generating said intermediate mechanical zero value in response to said plurality of measured no flow .DELTA.t values; and setting said mechanical zero value equal to said intermediate value if the standard deviation is less than the pre-defined limit value.

8. The method in claim 7 wherein said mechanical zero setting step comprises the step of setting said mechanical zero value equal to said intermediate value if the intermediate value lies within a pre-defined range.

9. The method in claim 8 wherein said current mechanical zero producing step comprises the step of calculating said intermediate value as an average of the plurality of measured no flow .DELTA.t values.

10. The method in claim 9 wherein said determining step comprises the step of updating, in response to the occurrence of each successive measured no flow .DELTA.t value, the standard deviation of said plurality of measured no flow .DELTA.t values using said successive measured no flow .DELTA.t value.

11. The method in claim 10 wherein the pre-defined limit is an integer multiple of said convergence limit.

12. A Coriolis meter (5) for measuring flow rate of a process fluid flowing therethrough comprising:
at least one flow conduit (130);
means (180) for oscillating the conduit;
means (160R, 160L) for sensing movement of said conduit caused by opposing Coriolis forces induced by passage of the process fluid through said flow conduit and for producing first and second signals responsive to said sensed movement of said conduit;
circuit means (30), responsive to said first and second signals, for providing a flow rate value of said process fluid, said circuit means comprising:
means (70, 80) for measuring, in response to said first and second sensor signals, a plurality of successive time periods (.DELTA.t) occurring between corresponding points on the first and second signals while the process fluid does not flow through said conduit so as to form a corresponding plurality of measured no flow .DELTA.t values;
means (823) for determining a standard deviation of said plurality of measured no flow .DELTA.t values; and means (826, 829) for producing, in response to said plurality of measured no flow .DELTA.t values and if the standard deviation is less than a pre-defined limit value, a current mechanical zero value for subsequent use in compensating flow based measured .DELTA.t values so as to determine therefrom the flow rate of the process fluid then flowing through said meter.

13. The meter in claim 12 further comprising means (832, 840) for determining a number of said measured no flow .DELTA.t values in said plurality and used in determining an intermediate mechanical zero value as being the lesser of either a pre-defined maximum number of measured no flow .DELTA.t values or a total number of the measured no flow .DELTA.t values that have occurred prior to the standard deviation thereof obtaining a value less than a predefined convergence limit.

14. The meter in claim 13 wherein said time periods measuring means comprises means (843) for measuring at least a predefined minimum number of the successive time periods such that said plurality of measured no flow .DELTA.t values contains a corresponding minimum number of values.

15. The meter in claim 14 wherein said current mechanical zero producing means comprises means (829) for calculating said intermediate value as an average of the plurality of measured no flow .DELTA.t values.

16. The meter in claim 15 wherein said determining means comprises means (865) for updating, in response to the occurrence of each successive measured no flow .DELTA.t value, the standard deviation of said plurality of measured no flow .DELTA.t values using said successive measured no flow .DELTA.t value.

17. The meter in claim 16 wherein the pre-defined limit is an integer multiple of said convergence limit.

18. The meter in claim 14 wherein said current mechanical zero producing means comprises:

means (806, 809) for generating said intermediate mechanical zero value in response to said plurality of measured no flow .DELTA.t values; and means (826) for setting said mechanical zero value equal to said intermediate value if the standard deviation is less than the pre-defined limit value.

19. The meter in claim 18 wherein said mechanical zero setting means comprises means (843, 846) for setting said mechanical zero value equal to said intermediate value if the intermediate value lies within a pre-defined range.

20. The meter in claim 19 wherein said current mechanical zero producing means comprises means (829) for calculating said intermediate value as an average of the plurality of measured no flow .DELTA.t values.

21. The meter in claim 20 wherein said determining means comprises means (865) for updating, in response to the occurrence of each successive measured no flow .DELTA.t value, the standard deviation of said plurality of measured no flow .DELTA.t values using said successive measured no flow .DELTA.t value.

22. The meter in claim 21 wherein the pre-defined limit is an integer multiple of said convergence limit.