US20030127551A1 - Estimating tape pack radius using a kalman filter - Google Patents

Estimating tape pack radius using a kalman filter Download PDF

Info

Publication number
US20030127551A1
US20030127551A1 US09/441,003 US44100399A US2003127551A1 US 20030127551 A1 US20030127551 A1 US 20030127551A1 US 44100399 A US44100399 A US 44100399A US 2003127551 A1 US2003127551 A1 US 2003127551A1
Authority
US
United States
Prior art keywords
tape
angular position
reel
take
responsive
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
US09/441,003
Inventor
John Abedor
Kurt Hallamasek
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to US09/441,003 priority Critical patent/US20030127551A1/en
Publication of US20030127551A1 publication Critical patent/US20030127551A1/en
Assigned to HILLSIDE CAPITAL INCORPORATED reassignment HILLSIDE CAPITAL INCORPORATED SECURITY AGREEMENT Assignors: AMPEX CORPORATION
Abandoned legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B15/00Driving, starting or stopping record carriers of filamentary or web form; Driving both such record carriers and heads; Guiding such record carriers or containers therefor; Control thereof; Control of operating function
    • G11B15/18Driving; Starting; Stopping; Arrangements for control or regulation thereof
    • G11B15/46Controlling, regulating, or indicating speed
    • G11B15/48Starting; Accelerating; Decelerating; Arrangements preventing malfunction during drive change
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B15/00Driving, starting or stopping record carriers of filamentary or web form; Driving both such record carriers and heads; Guiding such record carriers or containers therefor; Control thereof; Control of operating function
    • G11B15/18Driving; Starting; Stopping; Arrangements for control or regulation thereof
    • G11B15/46Controlling, regulating, or indicating speed

Definitions

  • the invention relates to tape systems, and in particular, to systems for determining the tape pack radius of supply and/or take-up reels.
  • Tape systems generally have two reels for storing tape, namely, a supply reel and a take-up reel, a capstan for moving the tape from reel to reel and tension arms for regulating the tape tension.
  • High-performance tape systems also include servo systems, which regulate tape position and velocity. The servo systems rely on estimates of the tape pack radii to determine how to control the rotational speeds of the reels to achieve the desired tape velocity and position. The more accurate the estimates, the more precisely the servo system can control the movement of the tape.
  • Good estimates of the tape pack radii are fundamentally important in controlling all aspects of the system operations. For example, good estimates are important in determining from which reel to draw the tape to wrap around a scanner. An inaccurate estimate could result in an over-rotation of the selected reel. Further, good estimates are important to determining when to decelerate a high-speed rewind operation, again to avoid over-rotation of one of the reels that may result in the breaking of the tape of the detatchment of the tape from the reel. Also, good estimates are important to determine if there is sufficient tape available on the supply reel to complete a record operation. Inaccurate estimates of the reel pack radii can result in incomplete record operations, if the system sufficiently under estimates the tape position.
  • the tape pack radius is calculated from measurements of the angular positions of the reels and the capstan.
  • the position measurements are made by, for example, optical encoders that count the number of slots that pass between a photo detector and a light source as the reel rotates.
  • the calculations produce results that are at best as accurate as the position measurements, which tend to be “noisy.”
  • the measurement noise is due in large part to quantitization errors. At slow speeds these systems tend to produce relatively inaccurate results because the position measurements are comparable to the quantitization errors.
  • the invention is a tape system that uses a Kalman filter to produce both an estimate of the tape pack radius and an estimate of the associated estimation error variance.
  • the Kalman filter produces the estimates based on current and prior angular estimation position measurements of the reel, capstan, and tension arm, and the prior estimates of both the tape pack radius and the estimation error variance.
  • the filter gain is based on the prior estimation error variance and a calculated measurement error variance, which is a function of the current and prior measurements.
  • the system uses two Kalman filters, one to produce the estimates for the supply reel and one or produce the estimates for the take-up reel.
  • the Kalman filter includes a model of the dynamics of the system that it uses to predict what the tape pack radius and estimation error variance will be at the time of a new measurement.
  • the Kalman filter updates the estimate of the tape pack radius when a new measurement is made, based on this predicted radius and a correction factor that is the weighted difference between a “measured” reel pack radius (calculated from reel, capstan, and tension arm measurements) and the predicted radius.
  • This weight which is Kalman filter gain, is a function of the estimation error variance and a measurement noise, or error variance.
  • FIG. 1 is a functional block diagram of a system constructed in accordance with the invention.
  • FIG. 2 is a flowchart of the operations of the system of FIG. 1;
  • FIG. 3 is a flowchart of the operations of the system of FIG. 1 when the system is operating in a coasting mode.
  • a tape system includes a capstan 14 that transfers tape 12 between a supply reel 10 and a take-up reel 16 .
  • a capstan 14 that transfers tape 12 between a supply reel 10 and a take-up reel 16 .
  • tape is wound off of the supply reel and onto the take-up reel and, thus, the tape pack radius of the supply reel decreases as the tape pack radius of the take-up reel increases.
  • Angular position transducers 21 - 25 which may be, for example, optical encoders, are attached to the shafts (not shown) of the capstan 14 , the two reels 10 and 16 , and the tension arms 18 and 20 .
  • the transducers operate in a conventional manner to measure, respectively, the angular positions of the capstan, the reels and the tension arms.
  • a supply Kalman filter 26 a receives measurements from transducers 21 , 22 and 24
  • a take-up Kalman filter 26 b receives measurements from transducers 23 , 22 , and 25 . These measurements are used by the filters to produce estimates of the tape pack radii and estimates of associated estimation error variances, which provide a confidence interval around each tape pack radius estimate.
  • the operations of the filters 26 a and 26 b are discussed in more detail with reference to FIG. 2 below. Also as discussed in more detail below, the estimates produced by the filters are used by a servo system 28 that controls the motion of and the tension in the tape 12 , and by a tape length estimator 30 that determines how much tape is available for a record operation.
  • the tape system includes other well-known elements such as additional tape guides, scanners, longitudinal heads, and so forth, none of which are shown in the drawing. As long as these elements are fixed in position so that they do not cause the tape path to vary, their presence is immaterial to the operations of the Kalman filters 26 a and 26 b . The effects on the operations of the filters of a change in the tape path length, for example, by means of movable guides, is discussed below with reference to FIG. 4.
  • r r is the radius of the take-up reel tape pack
  • ⁇ r is the change in angular position of the take-up reel 16 from the previous sample time to the current sample time
  • ⁇ c is the change in angular position of the capstan
  • ⁇ a is a gain determined by the geometry of the tape path and the nominal position of the take-up tension arm 20
  • ⁇ a is the change in angular position of the arm.
  • the tape pack as a hollow, circular cylindrical mass with width “w”, and a uniform density “d,” i.e. mass per unit volume, where d is the density of the tape itself. If a length of tape “l” units long, with width w and thickness “ ⁇ ” is wound onto the reel, the mass of the tape pack increases by the mass of that length of tape, and the radius of the tape pack increases accordingly.
  • r 1 is the outer radius of the tape pack and r 0 is the inner radius of the tape pack, i.e., the radius of the reel hub.
  • r 2 is the inner radius of the tape pack, i.e., the radius of the reel hub.
  • r(t i ) is the tape pack radius at time t i
  • ⁇ r (t i ) is the angular position of the take-up reel at time t i .
  • r ⁇ ( t f ) r ⁇ ( t 0 ) + ⁇ 2 ⁇ ⁇ ⁇ ⁇ [ ⁇ r ⁇ ( t n ) - ⁇ r ⁇ ( t 0 ) ] eqn . ⁇ 5
  • ⁇ c is the position of the capstan at the current sample time
  • ⁇ c ⁇ is the position of the capstan at the previous sample time
  • ⁇ a and ⁇ a ⁇ are the positions of the tension arm at the current and previous sample times
  • ⁇ r and ⁇ r ⁇ are the positions of the take-up reel at the current and previous sample times.
  • ⁇ m 2 2 ⁇ r - ⁇ r - ⁇ [ r c 2 ⁇ ⁇ var ⁇ ( ⁇ c ) + ⁇ a 2 ⁇ ⁇ var ⁇ ( ⁇ a ) + g ⁇ ( ⁇ c , ⁇ c , ⁇ a , ⁇ a - , ⁇ r , ⁇ r - ) 2 ⁇ ⁇ var ⁇ ( ⁇ r ) ] eqn . ⁇ 6
  • the system ignores measurements that it determines to be unreasonable. As discussed below with reference to FIG. 3, the system considers a measurement to be unreasonable if the associated measurement error variance, ⁇ m 2 , is greater than the maximum variance ⁇ max 2 , or if the three-sigma measurement interval around the calculated measured radius r m , as determined by the measurement model, is not at least partially included within the interval from the minimum tape pack radius, r min , to the maximum tape pack radius, r max . Ignoring the unreasonable measurements greatly improves the robustness of the Kalman filter to capstan slip.
  • the Kalman filters use equations 1 and 6 to produce estimates of the tape pack radii and the associated measurement error variances, respectively, as discussed below.
  • the tape pack radii processor 26 in step 300 must initialize the Kalman filter with (1) an initial tape pack radius estimate, (2) an initial estimation error variance and (3) initial position measurements of the capstan, the tension arm, and the reel, that is, ⁇ c , ⁇ a , and ⁇ r , where ⁇ r , is the angular position of the tape reel under consideration, for example, the take-up reel.
  • ⁇ r is the angular position of the tape reel under consideration, for example, the take-up reel.
  • the processor 26 takes a next set of position measurements ⁇ c , ⁇ a , and ⁇ r , and calculates ⁇ c , ⁇ a , and ⁇ r , using the initial set of measurements as the measurements of the previous sample time, namely, ⁇ c ⁇ , ⁇ a ⁇ and ⁇ r ⁇ (step 302 ).
  • the processor in step 306 , next determines if the calculated radius falls outside of the interval of possible radii, namely, the interval r min to r max . If so, the processor sets the calculated r m to r min if r m is less than r min , or to r max if r m is greater than r max (step 308 ).
  • the processor determines that the angular position measurements are unreasonable (steps 312 , 314 ).
  • the processor further tests the reasonableness of the measurements by determining if at least a portion of the three sigma interval around r m is within the interval of r min to r max (step 316 ). If so, the processor determines that the measurements are reasonable, and uses the calculated radius r m and measurement error variance ⁇ m m 2 in the Kalman filter to determine estimates of the tape pack radius and the estimation error variance. Otherwise, the system determines that the measurements are unreasonable and, as discussed below, uses predicted values for the estimates.
  • step 320 updates the estimate of the tape pack radius as:
  • ⁇ circumflex over (r) ⁇ 30 ⁇ circumflex over (r) ⁇ ⁇ +k( r m ⁇ circumflex over (r) ⁇ ⁇ )
  • ⁇ v 2 is a constant that represents the inaccuracies of, or the noise applied to, the predictive model.
  • the servo controller 28 uses the estimated tape pack radii in conventional feedback and feed-forward loops that control the rotational speeds of the reels and move the tape at a desired velocity.
  • the tape length estimator 30 which determines how much tape is available for record operations, uses as a lower bound for the tape pack radii:
  • the processor then sets the measured values ⁇ c , ⁇ a and ⁇ r as the prior values ⁇ c ⁇ , ⁇ a ⁇ , and ⁇ r ⁇ , respectively, and repeats steps 302 - 320 to calculate updated estimates for the next sample time.
  • the processor operates in a “coasting” mode when the tape path length is altered by anything but the tension arm.
  • the system operates in the coasting mode when the tape is being unwound from one reel and wrapped around the scanner (not shown).
  • the system also operates in a coasting mode when the tape is being withdrawn from the scanner.
  • the processor continues operating in the coasting mode while the system is varying the tape path length. Once the tape path length is held constant, the processor again utilizes the filter gain, and operates as discussed above with reference to FIG. 3.

Abstract

A system for measuring a length of tape available for a record operation has both a tape supply reel which rotates as a tape leaves the supply reel during a tape transfer process, and a tape take-up reel which receives tape from the supply reel, and the tape take-up reel also rotates as it receives the tape during the tape transfer process. There are provided an encoder responsive to movement of the tape, a first angular position transducer to measure an angular position of the tape supply reel, a second angular position transducer to measure an angular position of the tape take-up reel, and a third transducer responsive to the encoder. A Kalman filter, responsive to an angular position measurement by the first angular position transducer, the second angular position transducer, and also the third angular position transducer, calculates the length of tape available for a record operation. The Kalman filter also calculates the radius of the tape pack on the supply reel, and the radius of the tape pack on the take-up reel.

Description

    FIELD OF THE INVENTION
  • The invention relates to tape systems, and in particular, to systems for determining the tape pack radius of supply and/or take-up reels. [0001]
  • BACKGROUND OF THE INVENTION
  • Tape systems generally have two reels for storing tape, namely, a supply reel and a take-up reel, a capstan for moving the tape from reel to reel and tension arms for regulating the tape tension. High-performance tape systems also include servo systems, which regulate tape position and velocity. The servo systems rely on estimates of the tape pack radii to determine how to control the rotational speeds of the reels to achieve the desired tape velocity and position. The more accurate the estimates, the more precisely the servo system can control the movement of the tape. [0002]
  • Good estimates of the tape pack radii are fundamentally important in controlling all aspects of the system operations. For example, good estimates are important in determining from which reel to draw the tape to wrap around a scanner. An inaccurate estimate could result in an over-rotation of the selected reel. Further, good estimates are important to determining when to decelerate a high-speed rewind operation, again to avoid over-rotation of one of the reels that may result in the breaking of the tape of the detatchment of the tape from the reel. Also, good estimates are important to determine if there is sufficient tape available on the supply reel to complete a record operation. Inaccurate estimates of the reel pack radii can result in incomplete record operations, if the system sufficiently under estimates the tape position. [0003]
  • In prior known systems the tape pack radius is calculated from measurements of the angular positions of the reels and the capstan. The position measurements are made by, for example, optical encoders that count the number of slots that pass between a photo detector and a light source as the reel rotates. The calculations produce results that are at best as accurate as the position measurements, which tend to be “noisy.” With optical encoders, for example, the measurement noise is due in large part to quantitization errors. At slow speeds these systems tend to produce relatively inaccurate results because the position measurements are comparable to the quantitization errors. [0004]
  • Certain prior systems have processed the noisy measurement using low-pass filters, in order to smooth them. However, this approach has two significant problems. First, the signals produced by these filters always lag behind the true tape pack radii, or in other words, the estimates are biased. Second, these filters are slow to converge. Moreover, there is an intrinsic tradeoff—the more the filter smooths the output signals, i.e., the estimates of the tape pack radii, the more lag is introduced into the system and the slower the convergence. [0005]
  • SUMMARY OF INVENTION
  • The invention is a tape system that uses a Kalman filter to produce both an estimate of the tape pack radius and an estimate of the associated estimation error variance. The Kalman filter produces the estimates based on current and prior angular estimation position measurements of the reel, capstan, and tension arm, and the prior estimates of both the tape pack radius and the estimation error variance. The filter gain is based on the prior estimation error variance and a calculated measurement error variance, which is a function of the current and prior measurements. Preferably, the system uses two Kalman filters, one to produce the estimates for the supply reel and one or produce the estimates for the take-up reel. [0006]
  • More specifically, the Kalman filter includes a model of the dynamics of the system that it uses to predict what the tape pack radius and estimation error variance will be at the time of a new measurement. The Kalman filter then updates the estimate of the tape pack radius when a new measurement is made, based on this predicted radius and a correction factor that is the weighted difference between a “measured” reel pack radius (calculated from reel, capstan, and tension arm measurements) and the predicted radius. This weight, which is Kalman filter gain, is a function of the estimation error variance and a measurement noise, or error variance.[0007]
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • The above and further advantages of the invention may be better understood by referring to the following description in conjunction with the accompanying drawings, in which: [0008]
  • FIG. 1 is a functional block diagram of a system constructed in accordance with the invention; [0009]
  • FIG. 2 is a flowchart of the operations of the system of FIG. 1; and [0010]
  • FIG. 3 is a flowchart of the operations of the system of FIG. 1 when the system is operating in a coasting mode. [0011]
  • DETAILED DESCRIPTION
  • 1. The System [0012]
  • Referring to FIG. 1, a tape system includes a [0013] capstan 14 that transfers tape 12 between a supply reel 10 and a take-up reel 16. When the reels rotate clockwise, as depicted by the arrows, tape is wound off of the supply reel and onto the take-up reel and, thus, the tape pack radius of the supply reel decreases as the tape pack radius of the take-up reel increases.
  • Angular position transducers [0014] 21-25, which may be, for example, optical encoders, are attached to the shafts (not shown) of the capstan 14, the two reels 10 and 16, and the tension arms 18 and 20. The transducers operate in a conventional manner to measure, respectively, the angular positions of the capstan, the reels and the tension arms.
  • A supply Kalman [0015] filter 26 a receives measurements from transducers 21, 22 and 24, and a take-up Kalman filter 26 b receives measurements from transducers 23, 22, and 25. These measurements are used by the filters to produce estimates of the tape pack radii and estimates of associated estimation error variances, which provide a confidence interval around each tape pack radius estimate. The operations of the filters 26 a and 26 b are discussed in more detail with reference to FIG. 2 below. Also as discussed in more detail below, the estimates produced by the filters are used by a servo system 28 that controls the motion of and the tension in the tape 12, and by a tape length estimator 30 that determines how much tape is available for a record operation.
  • The tape system includes other well-known elements such as additional tape guides, scanners, longitudinal heads, and so forth, none of which are shown in the drawing. As long as these elements are fixed in position so that they do not cause the tape path to vary, their presence is immaterial to the operations of the Kalman [0016] filters 26 a and 26 b. The effects on the operations of the filters of a change in the tape path length, for example, by means of movable guides, is discussed below with reference to FIG. 4.
  • 2. The Theory [0017]
  • We can look at the system as two subsystems, namely, a supply subsystem that includes the [0018] supply reel 10, the supply tension arm 18 and the capstan 14, and a take-up subsystem that includes the take-up reel 16, the take-up tension arm 20 and the capstan 14. The two subsystems have identical theories of operation, and we discuss herein the theory of operation of the take-up subsystem.
  • Let us consider the case when position measurements are made at regular short intervals of, for example, 20 milliseconds. If the tape is moving, the length of tape that is wound onto the take-[0019] up reel 16 must travel past the capstan 14, and we can state that, to a high degree of accuracy,
  • r rΔθrαΔθα −r cΔθc=0
  • where r[0020] r is the radius of the take-up reel tape pack, Δθr is the change in angular position of the take-up reel 16 from the previous sample time to the current sample time, Δθc is the change in angular position of the capstan, μa is a gain determined by the geometry of the tape path and the nominal position of the take-up tension arm 20, and Δθa is the change in angular position of the arm. The radius of the take-up reel is thus: r r = r c Δ θ c - μ a Δ θ a Δ θ r eqn . 1
    Figure US20030127551A1-20030710-M00001
  • If the tape is moving relatively slowly the “deltas,” that is, the changes in the angular positions are comparable in magnitude to the position measurement errors. If the system relies solely on these measurements to determine tape pack radius and ignores both past measurements and the dynamics of the system, errors in position measurements translate directly to inaccuracies in the determination of the tape pack radius. The prior systems discussed above in the Background section all suffer from this problem. [0021]
  • Our system circumvents those problems through the use of a Kalman filter, which has as an input signal a “measured” radius r[0022] m, that is calculated using equation 10. The Kalman filter incorporates a model of how the tape pack radius changes as a function of reel position to predict an estimated tape pack radius. The filter then uses the measured radius to correct this prediction. The derivation of the predictive model is considered here.
  • Consider the tape pack as a hollow, circular cylindrical mass with width “w”, and a uniform density “d,” i.e. mass per unit volume, where d is the density of the tape itself. If a length of tape “l” units long, with width w and thickness “δ” is wound onto the reel, the mass of the tape pack increases by the mass of that length of tape, and the radius of the tape pack increases accordingly. [0023]
  • The mass of the tape pack before the length of tape is wound on is: [0024]
  • dwπr 1 2 −dwπr 0 2
  • where r[0025] 1 is the outer radius of the tape pack and r0 is the inner radius of the tape pack, i.e., the radius of the reel hub. Suppose that the radius of the pack increases to r2 when the length of tape is wound onto the tape pack. The mass of the tape pack thus increases to:
  • dwπr 2 2 −dwπ r 0 2  eqn. 2
  • The increase in the mass of the tape pack due to the tape being wound onto the reel is dwlδ, which is the mass of the additional tape, hence: [0026]
  • dwπr 2 2 −dwπr 0 2 =dwπr 1 2 −dwπr 0 2 dwlδ.
  • Solving for r[0027] 2 we have: r 2 = ( r 1 2 + 1 π l δ ) 1 2 eqn . 3
    Figure US20030127551A1-20030710-M00002
  • Given an initial reel radius r, this equation enables us to approximately predict what the radius will be after we wind on the additional tape of length l. What we are really interested in, however, is an equation that allows us to predict what the radius will be given an initial reel radius and an amount by which the reel rotates. If the reel rotates in the clockwise direction by a small amount Δθ[0028] r and the reel radius is initially r, a length of tape equal to rΔθr is wound onto the tape pack and the tape pack radius increases by Δr, which according to eqn. 3 satisfies: r + Δ r = ( r 2 + 1 π r Δ θ r δ ) 1 2 = r ( 1 + Δ θ r δ π r ) 1 2 r ( 1 + Δ θ r δ 2 π r ) , hence Δ r = Δ θ r δ 2 π r and , in the limit , r . = δ 2 π θ . r eqn . 4
    Figure US20030127551A1-20030710-M00003
  • The desired predictive equation is the integral of eqn. 4, from an initial time to t[0029] 0 the time tf for which the prediction is required: r ( t f ) - r ( t 0 ) = δ 2 π [ θ r ( t f ) - θ r ( t 0 ) ]
    Figure US20030127551A1-20030710-M00004
  • where r(t[0030] i) is the tape pack radius at time ti, and θr(ti) is the angular position of the take-up reel at time ti. Thus: r ( t f ) = r ( t 0 ) + δ 2 π [ θ r ( t n ) - θ r ( t 0 ) ] eqn . 5
    Figure US20030127551A1-20030710-M00005
  • Inaccuracies in this model are accounted for in the Kalman filter by an estimate of “noise” applied to the system—that is, by quantifying as applied noise essentially unmeasurable attributes of the system dynamics. The applied noise is one of the factors that is used in determining the Kalman filter gain, k, as discussed below with reference to FIGS. 2 and 3. [0031]
  • To determine the filter gain, the system also requires an estimate of the measurement noise, or error, variance. The system implicitly assumes that the errors in the various position measurements can be represented by independent, zero-mean random variables ε[0032] i, which each have “small” variances. We know that for a smooth (differentiable) function g that maps “n” real numbers to one real number, we have:
  • var(g(x 1 1 . . . ,xnn))≈
  • [0033] ( ( g x 1 ( x ) ) 2 var ( ɛ 1 ) + + ( g x n ( x ) ) 2 var ( ɛ n )
    Figure US20030127551A1-20030710-M00006
  • From this relation we obtain the approximate measurement noise variance. Based on [0034] equation 1, we define the function g as: g ( θ c , θ c - , θ a , θ a - , θ r , θ r - ) = r c ( θ c - θ c - ) - μ a ( θ a - θ a - ) θ r - θ r -
    Figure US20030127551A1-20030710-M00007
  • where θ[0035] c is the position of the capstan at the current sample time, θc− is the position of the capstan at the previous sample time, θa and θa− are the positions of the tension arm at the current and previous sample times, and θr and θr− are the positions of the take-up reel at the current and previous sample times. If these quantities are perturbed, respectively, by independent, zero-mean random variables εc, εa and εr, the approximate measurement error variance σm 2 is: σ m 2 = 2 θ r - θ r - [ r c 2 var ( ɛ c ) + μ a 2 var ( ɛ a ) + g ( θ c , θ c , θ a , θ a - , θ r , θ r - ) 2 var ( ɛ r ) ] eqn . 6
    Figure US20030127551A1-20030710-M00008
  • The variances var(ε[0036] c), var (εa) and var (εr) are computed by assuming that the quantitization error is a random variable with zero mean that is uniformly distributed over an interval of size 2 π N
    Figure US20030127551A1-20030710-M00009
  • radians, where N is the number of encoder counts per revolution. [0037]
  • The system ignores measurements that it determines to be unreasonable. As discussed below with reference to FIG. 3, the system considers a measurement to be unreasonable if the associated measurement error variance, σ[0038] m 2, is greater than the maximum variance σmax 2, or if the three-sigma measurement interval around the calculated measured radius rm, as determined by the measurement model, is not at least partially included within the interval from the minimum tape pack radius, rmin, to the maximum tape pack radius, rmax. Ignoring the unreasonable measurements greatly improves the robustness of the Kalman filter to capstan slip.
  • The Kalman filters [0039] use equations 1 and 6 to produce estimates of the tape pack radii and the associated measurement error variances, respectively, as discussed below.
  • 3. System Operation [0040]
  • Referring now to FIGS. 1 and 2, when a tape, for example, a cassette tape, that is wound on the supply and take-up [0041] reels 10 and 16, is loaded into the system, the system has no indication of the tape pack radii. Accordingly, the tape pack radii processor 26 in step 300 must initialize the Kalman filter with (1) an initial tape pack radius estimate, (2) an initial estimation error variance and (3) initial position measurements of the capstan, the tension arm, and the reel, that is, θc, θa, and θr, where θr, is the angular position of the tape reel under consideration, for example, the take-up reel. We discuss below the operations of the system in determining the tape pack radius of the take-up reel 16. The system performs the same operations to produce estimates of the tape pack radius of the supply reel 10 and the associated estimation error variance.
  • The [0042] processor 26 uses as the initial estimate of the tape pack radius: r ^ - = r max + r min 2
    Figure US20030127551A1-20030710-M00010
  • which is the radius when one-half of the tape is wound on the reel, and uses as an estimate of the initial estimation error variance: [0043] v - = ( r max - r min ) 2 12
    Figure US20030127551A1-20030710-M00011
  • which is the variance of a uniformly distributed variable over the interval r[0044] min to rmax. This initial estimate of the error variance is relatively large and indicates that the initial estimate of the tape pack radius is very likely to be inaccurate.
  • The [0045] processor 26 takes a next set of position measurements θc, θa, and θr, and calculates Δθc, Δθa, and Δθr, using the initial set of measurements as the measurements of the previous sample time, namely, θc−, θa− and θr− (step 302).
  • Using these delta values, the processor, in [0046] step 304, calculates a “measured” radius, rm: r m = r c Δθ c - μ a Δθ a Δθ r
    Figure US20030127551A1-20030710-M00012
  • The processor, in [0047] step 306, next determines if the calculated radius falls outside of the interval of possible radii, namely, the interval rmin to rmax. If so, the processor sets the calculated rm to rmin if rm is less than rmin, or to rmax if rm is greater than rmax (step 308).
  • The system then in [0048] step 310 calculates a measurement error variance σm 2, using the measured Δθr and the calculated radius rm: σ m 2 = 2 Δθ r [ r c 2 var ( ɛ c ) + μ a 2 var ( ɛ a ) + r m 2 var ( ɛ r ) ]
    Figure US20030127551A1-20030710-M00013
  • If the calculated measurement error variance is larger than the maximum estimation error variance, which is determined by equation 7, the processor determines that the angular position measurements are unreasonable ([0049] steps 312, 314).
  • Assuming the calculated measurement error variance is not greater than the maximum estimation error variance, the processor further tests the reasonableness of the measurements by determining if at least a portion of the three sigma interval around r[0050] m is within the interval of rmin to rmax (step 316). If so, the processor determines that the measurements are reasonable, and uses the calculated radius rm and measurement error variance σmm 2 in the Kalman filter to determine estimates of the tape pack radius and the estimation error variance. Otherwise, the system determines that the measurements are unreasonable and, as discussed below, uses predicted values for the estimates.
  • If the measurements are reasonable, the processor, in [0051] step 318, calculates the Kalman filter gain, k, as: k = v - v - + σ m 2
    Figure US20030127551A1-20030710-M00014
  • where v[0052] is the estimation error variance extrapolated from the previous sample time. The processor next, in step 320, updates the estimate of the tape pack radius as:
  • {circumflex over (r)} 30 ={circumflex over (r)} +k( r m −{circumflex over (r)} )
  • and updates the estimation error variance as: [0053] v + = 1 - k v - .
    Figure US20030127551A1-20030710-M00015
  • These updated estimates are the output signals of the filter. The [0054] servo controller 28 and a tape length counter (not shown) then use these estimates to determine how to control the tape system, as discussed below. Next, the tape pack radius and estimation error variance are propagated to the next sample time. The radius estimate is propagated according to the predictive equation: r ^ - = r ^ + + δ 2 π Δθ r ,
    Figure US20030127551A1-20030710-M00016
  • which comes from equation 5, and the estimation error variance is propagated according to the equation: [0055]
  • v =v +−σv 2
  • where σ[0056] v 2 is a constant that represents the inaccuracies of, or the noise applied to, the predictive model.
  • The [0057] servo controller 28 uses the estimated tape pack radii in conventional feedback and feed-forward loops that control the rotational speeds of the reels and move the tape at a desired velocity. The tape length estimator 30, which determines how much tape is available for record operations, uses as a lower bound for the tape pack radii:
  • {circumflex over (r)} +−3σ.
  • This is a conservative estimate, which prevents the tape length estimator from over estimating the length of the tape available for a given record operation. [0058]
  • If the [0059] processor 26 determines that the measurements are unreasonable, it does not update the filter gain. Further, it does not calculate the updated estimates. Instead, it predicts what the estimates should be based solely on the previous estimates. The processor, in step 315, sets {circumflex over (r)}+={circumflex over (r)} and v+=v and predicts an estimate of the tape pack radius {circumflex over (r)} as: r ^ - = r ^ + + δ 2 π Δθ r
    Figure US20030127551A1-20030710-M00017
  • and the error variance v[0060] as:
  • v =v+σv 2
  • The processor then sets the measured values θ[0061] c, θa and θr as the prior values θc−, θa−, and θr−, respectively, and repeats steps 302-320 to calculate updated estimates for the next sample time.
  • The system continues to determine and, as appropriate, predict estimates for the tape pack radii and the associated error variances as long as estimates are required, that is, as long as the reels continue to rotate. [0062]
  • Referring now to FIG. 4, the processor operates in a “coasting” mode when the tape path length is altered by anything but the tension arm. For example, the system operates in the coasting mode when the tape is being unwound from one reel and wrapped around the scanner (not shown). The system also operates in a coasting mode when the tape is being withdrawn from the scanner. These operations, which significantly change the length of the tape path, are not included in the underlying assumptions upon which the measurement model in the Kalman filter is based. Accordingly, the measured radius r[0063] m produced by that model cannot be used to determine the estimates of the tape pack radius and the error variance when the path length is so altered. The estimates are instead produced using the predictive model.
  • The system thus extrapolates from the current estimates of tape pack radius and estimation error variance to the next sample time by taking a next set of position measurements (step [0064] 402), determining Δθc, Δθa and Δθr (step 404) and estimating the tape pack radius by: r ^ - = r ^ + + δ 2 π Δθ r
    Figure US20030127551A1-20030710-M00018
  • and the error variance by [0065]
  • v =v +v 2
  • where σ[0066] v 2 is a constant that represents the inaccuracies in the predictive model (step 406). It next, in step 408, sets θc−, θa− and σr− equal to the measured values and {circumflex over (r)}+={circumflex over (r)} and v+=v, and at the next sample time again begins the process of predicting the estimates. The system thus returns to step 402.
  • The processor continues operating in the coasting mode while the system is varying the tape path length. Once the tape path length is held constant, the processor again utilizes the filter gain, and operates as discussed above with reference to FIG. 3. [0067]
  • Generally, the system operates in the coasting mode for only a small number of consecutive sample times. By predicting new radii and estimation error variance values during the coasting operations, the processor can determine at the end of these operations reliable updated estimates for the tape pack radii and error variance, as well as a reliable filter gain, which is based in part on these estimates. [0068]
  • While the measurement model could be revised to include the system operations in which the path length is varied, it is not necessary since the time the system spends so little time performing these operations. Thus, the trade off of increased complexity in the model for more accuracy during these limited path-varying operations seems unwarranted. Further, since the estimates produced by the Kalman filter converge rapidly, the accuracy of the system is only minimally reduced by not including in the model the path-varying operations. [0069]
  • The foregoing description has been limited to a specific embodiment of this invention. It will be apparent, however, that variations and modifications may be made to the invention, with the attainment of some or all of its advantages. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true spirit and scope of the invention.[0070]

Claims (12)

What is claimed is:
1. A system for measuring tape pack radii, comprising:
a tape supply reel, said tape supply reel rotating as a tape leaves said tape supply reel during a tape transfer process;
a tape take-up reel for receiving tape from said tape supply reel, said tape take-up reel rotating as it receives said tape during said tape transfer process;
an at least one encoder responsive to movement of said tape;
a first angular position transducer to measure an angular position of said tape supply reel;
a second angular position transducer to measure an angular position of said tape take-up reel;
a third transducer responsive to said at least one encoder;
a Kalman filter, responsive to one or both of an angular position measurement by said first angular position transducer and an angular position measurement by said second angular position transducer and also responsive to an angular position measurement by said third angular position transducer, to calculate an updated estimate of one or both of a supply radius of a tape pack on said tape supply reel and a take-up radius of a tape pack on said tape take-up reel;
a servo-controller, responsive to one or both of said supply radius and said take-up radius, to control rotation of said tape supply reel and said tape take-up reel.
2. The apparatus as in claim 1, wherein said Kalman filter further comprises:
a supply Kalman filter responsive to said first angular position transducer and said third angular position transducer;
a take-up Kalman filter responsive to said second angular position transducer and said third angular position transducer.
3. The apparatus as in claim 1 wherein said at least one encoder further comprises:
a first encoder responsive to an angular position of a supply reel tension arm;
a second encoder responsive to an angular position of a take-up reel tension arm.
4. The apparatus as in claim 1, further comprising:
a capstan, said tape contacting said capstan and said capstan rotating as said tape transfers from said tape supply reel to said tape take-up reel.
5. The apparatus as in claim 3 wherein said at least one encoder further comprises:
a third encoder responsive to an angular position of a capstan.
6. The apparatus as in claim 1 further comprising:
a tape length estimator responsive to said Kalman filter to determine the amount of tape available for a record operation.
7. A system for measuring a length of tape available for a record operation, comprising:
a tape supply reel, said tape supply reel rotating as a tape leaves said tape supply reel during a tape transfer process;
a tape take-up reel for receiving tape from said tape supply reel, said tape take-up reel rotating as it receives said tape during said tape transfer process;
an at least one encoder responsive to movement of said tape;
a first angular position transducer to measure an angular position of said tape supply reel;
a second angular position transducer to measure an angular position of said tape take-up reel;
a third transducer responsive to said at least one encoder;
a Kalman filter, responsive to one or both of an angular position measurement by said first angular position transducer and an angular position measurement by said second angular position transducer and also responsive to an angular position measurement by said third angular position transducer, to calculate said length of tape available for a record operation.
8. A method for estimating a radius of a tape on a supply reel and on a take-up reel, comprising:
measuring a first angular position of a tape supply reel;
measuring a second angular position of a tape take-up reel;
measuring a third angular position responsive to movement of a tape; and,
estimating by a Kalman filter a radius of a tape pack on said supply reel and a radius of a tape pack on said take-up reel, in response to said first angular position of said tape supply reel, said second angular position of said tape take up reel, and said third angular position responsive to movement of said tape.
9. The method as in claim 8 wherein said estimating step by said Kalman filter further comprises:
responding to an initial estimate of said radius of a tape pack on said supply reel;
responding to an initial estimate of a radius of tape pack on said take-up reel; and,
responding to said first angular position measurement, said second angular position measurement, and said third angular position measurement to compute said radius of said tape pack on said supply reel and said radius of said tape pack on said take-up reel.
10. The method of claim 8 further comprising:
making said first angular measurement at a first regular time interval;
making said second angular measurement at a second regular time interval;
making said third angular measurement at a third regular time interval.
11. The method of claim 10 further comprising:
choosing said first regular time interval, said second regular time interval and said third regular time interval each to be approximately 20 milliseconds.
12. A method for estimating a length of tape available for a record operation, comprising:
measuring a first angular position of a tape supply reel;
measuring a second angular position of a tape take-up reel;
measuring a third angular position responsive to movement of a tape; and,
estimating by a Kalman filter said length of tape available for a record operation, in response to said first angular position of said tape supply reel, said second angular position of said tape take up reel, and said third angular position responsive to movement f said tape.
US09/441,003 1996-10-31 1999-11-16 Estimating tape pack radius using a kalman filter Abandoned US20030127551A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US09/441,003 US20030127551A1 (en) 1996-10-31 1999-11-16 Estimating tape pack radius using a kalman filter

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US08/740,637 US6082653A (en) 1996-10-31 1996-10-31 System for estimating tape pack radius using a Kalman filter
US09/441,003 US20030127551A1 (en) 1996-10-31 1999-11-16 Estimating tape pack radius using a kalman filter

Publications (1)

Publication Number Publication Date
US20030127551A1 true US20030127551A1 (en) 2003-07-10

Family

ID=24977416

Family Applications (2)

Application Number Title Priority Date Filing Date
US08/740,637 Expired - Fee Related US6082653A (en) 1996-10-31 1996-10-31 System for estimating tape pack radius using a Kalman filter
US09/441,003 Abandoned US20030127551A1 (en) 1996-10-31 1999-11-16 Estimating tape pack radius using a kalman filter

Family Applications Before (1)

Application Number Title Priority Date Filing Date
US08/740,637 Expired - Fee Related US6082653A (en) 1996-10-31 1996-10-31 System for estimating tape pack radius using a Kalman filter

Country Status (2)

Country Link
US (2) US6082653A (en)
EP (1) EP0840312A1 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110102934A1 (en) * 2009-11-05 2011-05-05 International Business Machines Corporation Cancellation of time-varying periodic disturbances in servo control systems
US8767335B2 (en) * 2012-08-30 2014-07-01 International Business Machines Corporation Accurate radius and velocity measurement of tape transports

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6963467B2 (en) 2001-07-17 2005-11-08 International Business Machines Corporation Method and apparatus for compensating for media shift due to tape guide
US6791781B2 (en) 2001-07-17 2004-09-14 International Business Machines Corporation Method and apparatus for providing linear position (LPOS) estimations
US6762900B2 (en) 2001-07-17 2004-07-13 International Business Machines Corporation Method and apparatus for performing position error signal conditioning of LTO media servo format written-in velocity variation
US6934108B2 (en) * 2003-01-22 2005-08-23 International Business Machines Corporation Servo pattern based tape tension control for tape drives
DE102004009057A1 (en) * 2004-02-23 2005-09-08 Memminger-Iro Gmbh Electronic positive
US8094407B2 (en) * 2009-08-18 2012-01-10 International Business Machines Corporation Servomechanism with adjustable predictor filter

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3834648A (en) * 1972-03-15 1974-09-10 Ampex Apparatus and method for sensing diameter of tape pack on storage reel
US3938041A (en) * 1973-06-21 1976-02-10 Ampex Corporation Tape pack diameter measurement means and method
US4001552A (en) * 1975-04-04 1977-01-04 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration Method and apparatus for measuring web material wound on a reel
JPS6012711B2 (en) * 1977-12-29 1985-04-03 ソニー株式会社 Magnetic tape winding amount display method
US4217615A (en) * 1978-02-02 1980-08-12 Pioneer Electronic Corporation Tape measurement and position system
JPS6014430B2 (en) * 1978-05-08 1985-04-13 ソニー株式会社 Recording/playback device
US4338645A (en) * 1978-11-06 1982-07-06 Hitachi, Ltd. Tape running time indication system
GB2058435B (en) * 1979-01-11 1983-07-27 Bang & Olufsen As System for detection of tape thickness in tape recorders or players
US4399953A (en) * 1981-11-18 1983-08-23 The Okonite Company Wire and cable process control apparatus
JPS5942681A (en) * 1982-09-02 1984-03-09 Mitsubishi Electric Corp Display device of tape quantity
US4731679A (en) * 1984-09-20 1988-03-15 Ampex Corporation Method and apparatus for transporting a recording medium with an adaptive velocity change profile
DE3504311A1 (en) * 1985-02-08 1986-08-14 Deutsche Thomson-Brandt Gmbh, 7730 Villingen-Schwenningen METHOD FOR DETERMINING THE EXPIRED AND / OR STILL AVAILABLE GAME TIME OF A MAGNETIC TAPE
JPS62236192A (en) * 1986-04-07 1987-10-16 Sony Corp Tape position detector
JPH01165084A (en) * 1987-02-23 1989-06-29 Toshiba Corp Detector for position of tape
JPH0668903B2 (en) * 1987-03-13 1994-08-31 三菱電機株式会社 Tape remaining time display device
DE3810365A1 (en) * 1988-03-26 1989-10-05 Schlafhorst & Co W METHOD AND DEVICE FOR DETERMINING THE COIL SCOPE OF CRANES AND FOR EVALUATING THE RESULT
US5248112A (en) * 1988-05-20 1993-09-28 Ampex Systems Corporation Tape transport control system with a feedforward capstan drive to the reel motors
US5276576A (en) * 1991-05-07 1994-01-04 Funai Electric Engineering Company Limited Method and device for obtaining data of tape thinness and/or hub radius of a tape reel

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110102934A1 (en) * 2009-11-05 2011-05-05 International Business Machines Corporation Cancellation of time-varying periodic disturbances in servo control systems
US8213105B2 (en) 2009-11-05 2012-07-03 International Business Machines Corporation Cancellation of time-varying periodic disturbances in servo control systems
US8767335B2 (en) * 2012-08-30 2014-07-01 International Business Machines Corporation Accurate radius and velocity measurement of tape transports
US9064532B2 (en) 2012-08-30 2015-06-23 International Business Machines Corporation Accurate radius and velocity measurement of tape transports

Also Published As

Publication number Publication date
US6082653A (en) 2000-07-04
EP0840312A1 (en) 1998-05-06

Similar Documents

Publication Publication Date Title
US6305629B1 (en) Servo error detection of bi-directional reel-to-reel tape drives using fine line tachometers
US4525654A (en) Tape transport control unit
JPS5920178B2 (en) Direct tape transfer device between reels
US4338645A (en) Tape running time indication system
JPH02126455A (en) Magnetic tape tension controller
US20030127551A1 (en) Estimating tape pack radius using a kalman filter
US20150062744A1 (en) Controlling the transport of a tape within a tape transport system
US6817560B2 (en) Combined tension control for tape
KR950009591B1 (en) Device checking the location of tape
EP0376960A1 (en) Apparatus for providing a profiled tape tension without utilizing a tape pack diameter sensor
US4739950A (en) Constant velocity tape drive system
JPS59186003A (en) Servo control system
JP2992423B2 (en) Tape constant speed controller
US4821581A (en) Method and apparatus for the continuous measuring of bulk material passing over a measuring wheel
US5085379A (en) Method of controlling reel drive
US5309300A (en) Beginning/end of tape detection system
JP2739963B2 (en) Control device for tape transfer device
US5815336A (en) Method for sensing position of magnetic tape during discontinuous travel of the tape
JP2930593B2 (en) Control device for tape transfer device
JPH0899106A (en) Method for tracking thickness deviation on inlet side of cold reversible rolling mill
JP2658125B2 (en) Tape length determination device
JP3018635B2 (en) Magnetic recording / reproducing device
JPS63300450A (en) System for controlling motor speed
JP2610014B2 (en) Tape amount detection device
JPS62252561A (en) Control method for tape shifting device

Legal Events

Date Code Title Description
STCB Information on status: application discontinuation

Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION

AS Assignment

Owner name: HILLSIDE CAPITAL INCORPORATED, NEW YORK

Free format text: SECURITY AGREEMENT;ASSIGNOR:AMPEX CORPORATION;REEL/FRAME:021630/0230

Effective date: 20081003