Various embodiments of the present disclosure are generally directed to an apparatus and method for disturbance rejection in a control system.
In some embodiments, a controller is adapted to position a control object. A disturbance observer generates a disturbance compensation value which is applied to reduce position error resulting from application of mechanical disturbance to the control object. The disturbance observer includes an adaptive filter with at least one dead zone providing a pass-through response with a scalar gain of less than one.
In further embodiments, a data storage device has a data read/write transducer adjacent a rotatable data recording medium. An actuation mechanism supports the transducer and is configured to move the transducer with respect to the medium. A servo controller is configured to generate an initial input signal for the actuation mechanism responsive to position error of the transducer. A disturbance observer generates a disturbance compensation value which is combined with the initial input signal to form a modified input signal which is applied to the actuation mechanism to adjust the position of the control object. The disturbance compensation value reduces position error resulting from application of mechanical disturbance to the data storage device. The disturbance observer includes an adaptive filter with at least one dead zone providing a pass-through response with a scalar gain of less than one.
In further embodiments, a method includes positioning a control object using a controller, determining a position error of the control object, and generating a disturbance compensation value responsive to application of mechanical disturbance to the control object. The disturbance compensation value is generated using an adaptive filter with at least one dead zone providing a pass-through response with a scalar gain of less than one. Once generated, the disturbance compensation value is applied to reduce the position error of the control object.
These and other features and advantages which may characterize various embodiments of the present disclosure can be understood in view of the following detailed discussion and the accompanying drawings.
Various embodiments of the present disclosure are generally directed to adaptive disturbance rejection in a control system, such as but not limited to a servo control system in a data storage device.
Control systems can position a control object using closed loop sensing and actuation. One application that uses complex positioning control solutions is a servo control loop in a hard disc drive (HDD) data storage device. Such servo control loops are operated to controllably position a read/write transducer adjacent tracks on an adjacent rotatable magnetic recording disc in order to facilitate the transfer of data between the disc and a host device. Feedback positioning data are obtained from servo data stored on the disc surface and used to correct transducer position error.
Adaptive disturbance observer (DOB) systems can be employed in disc drive servo loops and other forms of control systems. An adaptive DOB generally operates to compensate for position errors induced by the application of mechanical disturbances (e.g., noise, vibration, mechanical shocks) to the control object. Adaptive DOB circuits sense the disturbances and apply signal processing such as in the form of adaptive filtering to reduce the measured position error to a minimum level.
Disturbance rejection solutions can be difficult to implement in many modern applications due to large, unpredictable disturbances that arise from a variety of factors. For example, in the context of a portable consumer electronic device such as a laptop computer, disturbances can arise from the portable nature of the environment in which the device is used (e.g., a vibratory environment such as on an airplane or in a motor vehicle). Unpredictable disturbances can further be generated by the device itself, such as vibrations caused by high powered audio speakers playing music or other audio content.
In cloud computing and other multi-device storage environments, the aggregation of data storage devices into multi-device enclosures can subject the devices to mechanical disturbances from adjacent storage devices (e.g., rotational and translational vibration effects). Disturbances can further be induced from other sources such as high performance cooling fans, power supplies, etc.
Current generation DOB designs can use an adaptive H-filter that outputs disturbance compensation values that are injected into the servo loop to reduce the effects of the disturbance events. While operable, one limitation associated with such designs is reduced stability margin under different disturbance environments, particularly those subjected to large changes in the disturbance profile.
Accordingly, various embodiments of the present disclosure are generally directed to an apparatus and method for disturbance rejection in a control system, such as but not limited to a servo control loop in a data storage device.
As explained below, in some embodiments a control object is controlled by a closed loop positioning (control) system. During operation, the control object is subjected to the application of external mechanical disturbances having various frequency components.
A disturbance observer generates a disturbance compensation value which is applied to reduce position error resulting from application of the mechanical disturbance to the control object. The disturbance observer comprises an adaptive filter with at least one dead zone providing a pass-through response with a scalar gain of less than one. Other features can include a disturbance estimator that adaptively selects the dead zone gain and/or boundaries based on disturbances encountered during operation.
These and other features of various embodiments of the present disclosure can be understood beginning with a review of
The controller 102 provides top-level control of the device 100 and directs communications and data transfers with a host device (not separately shown in
The control system 110 includes a servo controller 112, a plant 114, a state predictor block 116, an inversion filter 118 and a disturbance observer (DOB) filter 120. Summing blocks 122, 124, 126 and 128 combine the various signals as described below. It will be appreciated that the control system 110 can be realized in hardware, software and/or firmware in a variety of ways. Other elements may be included as well including one or more disturbance sensors, etc., but such have been omitted for simplicity of illustration.
Generally, the controller 112 outputs a control input for use by the plant 114 to position the control object. The plant 114 represents the physical aspects of the control object and actuation mechanism. In a hard disc drive (HDD) environment, the plant may include a voice coil motor (VCM), VCM driver circuitry, a moveable actuator that positions the transducer responsive to the VCM inputs, and readback circuitry that uses the transduced readback signals from servo positioning data on the medium to generate a position signal. Other forms of plants can be used, including multi-input-multi-output (MIMO) control systems that use microactuators or other actuation stages to position the transducer or other control object.
The position signal output from the plant 114 may take the form of a position error signal (PES) indicative of position error with respect to a target position. Feedback position data are supplied to the controller so that, over a succession of servo intervals (samples), the control object (transducer) is nominally caused to follow a desired position/trajectory.
An external disturbance signal is added to the control input by summing block 126. This generally represents the application of mechanical disturbances to the system that induce a corresponding error component in the position of the control object. It is contemplated, albeit not necessarily required, that the disturbance profile will be time varying, multi-frequency and largely unpredictable so that different frequency spectra may be applied to the device over time and under different operational conditions.
A disturbance observer (DOB) compensation module 130 is provided to reduce the effects of the disturbance upon the system. The DOB compensation module 130 includes the state predictor 116, inversion filter 118 and DOB filter (H-filter) 120. Other circuitry may be included in the DOB compensation module as well. Generally, the module injects a disturbance compensation signal (as shown via summing block 124) in an effort to cancel out, or at least reduce, the effects of the disturbance at summing block 126.
The state predictor 116 provides a nominal closed-loop vibration transfer function ĜW (also referred to as DSF). The inversion filter provides an inverse function ĜWi (for example, a ZPET inversion, and can be characterized as Q/DSF). The DOB filter 120 provides an H filter response that is self-tuned and includes a specially configured adaptive dead zone as discussed below to reduce the effects of the injected disturbances and drive PES to a minimum. The overall vibration transfer function DSFDOB for the DOB module can be stated as:
To explain the use and configuration of the dead zone aspects of the DOB filter 120, it will be helpful to first briefly discuss limitations often encountered with traditional adaptation schemes that use H filter configurations. The following model uses a least mean square (LMS) approach in an effort to reduce a position error (ε) to be a minimum value (e.g., zero):
yW=θTϕ(k) (2)
ε(k+1)=eW(k+1)−θ(k)Tϕ(k) (3)
θ(k+1)=θ(k)+γϕ(k)ε(k−1) (4)
where yW is estimated disturbance, θT is an adaptive variables vector of the DOB filter, k are servo samples, ϕ(k) are modified position error inputs to the DOB filter, θTϕ(k) is estimated position from the DSF, e(k+1) is the measured position error and ε(k+1) is the estimated position error. The model is generally operable to reduce disturbance effects, but does not guarantee stable operation since external disturbances can provide significant increases in error sensitivity function response.
Accordingly, a dead zone is incorporated into the DOB filter 120 to address these and other instabilities of the loop. The dead zone is a range of operation of the filter in which a substantially constant signal pass-through response is given with a selected scalar gain (e.g., K<1). The dead zone features dynamic adjustment of the zone boundaries based on the magnitude of the measured position error eW. Since the measured position error generally excludes the effect of the injected compensation value yW, the dead zone boundary can be determined by the level of position error before injecting the compensation from the DOB loop.
The filter 200 can be defined as setting a proportional gain K to the measured position error eW as follows:
δ(k)=ε(k)±KeW(k)
if |ε(k)|>|KeW(k)|; else
δ(k)=0 (5)
where δ is the output of the dead zone filter. The bounds and scalar gain are adaptively selected based on the input measured position error eW. More specifically, as shown by the functional representation in
yW=θTϕ(k) (6)
ε(k+1)=eW(k+1)−θ(k)Tϕ(k) (7)
θ(k+1)=θ(k)+γϕ(k)δ(k+1) (8)
where δ(k+1) is governed by equation (5) above based on the then-existing measured and estimated position values eW(k) and ε(k) and the selected K value.
Operation of the dead zone filter can be better understood with a review of
The filter circuit 230 includes an absolute value (|ABS|) block 236 which operates on the input measured position error eW, gain block 238 which applies gain K to form the upper bound limit +KeW, and inverter block 240 to provide the lower bound limit −KeW. Filter output signal 242 represents the output δ(k) of the filter.
It follows from
The modified input signal is used by the VCM driver 254 to apply drive current to a VCM of an actuator 258, which in turn actuates (moves) a transducer 260 adjacent a corresponding storage medium 262. A position error signal (PES) or other position signal is fed back to the servo controller 252 and the disturbance observer 254. In this way, the disturbance observer operates in parallel to compensate for disturbance using a dead zone as configured in
Because of the adaptive nature of the DOB filter 200, it will be understood that the dead zone is not fixed, but rather, the dead zone location (e.g., nominally centered around 2 kHz) will adjust in both boundary and gain as other frequency disturbances as applied (e.g., 500 Hz, 1500 Hz, 2300 Hz, etc.). The scalar gain, while less than 1.0, will be set to the proper value based on the target stability margin and disturbance signature of the servo system.
The introduction of a relatively large input disturbance of a selected frequency (or relatively narrow frequency range) will be detected by the estimator block 202 and, provided the component exhibits sufficient persistence and magnitude over multiple successive samples, the estimator will select appropriate boundary and gain values to implement a dead zone in the filtering to compensate.
In some cases, one or both of the boundary and gain values will be adaptively adjusted during continued application of the dead zone. With regard to different disturbance frequencies, the system identifies the frequency contents, selects an appropriate gain K to adjust the dead zone boundary and the dead zone filter operates to generate the output delta value. Hence, unlike a notch filter which generally removes a narrow frequency range of interest (e.g., a 60 Hz power noise signal, etc.), the dead band zone filter 206 provides pass through at reduced amplitude for the appropriate frequency interval.
While a fixed amount of scalar gain K has been described, other configurations are contemplated including different scalar gains for different frequency ranges to further equalize minimum stability margin for various disturbance spectra.
It will now be appreciated that the various embodiments of the present disclosure provide a disturbance observer (DOB) in a control system used to position a control object, such as but not limited to a servo control system in a data storage device. An adaptive H-filter of the DOB is provided with a dead zone and adaptive selection logic that selects boundary and scalar gain values to provide pass through range to enhance stability margin. More than one dead zone can be applied depending on the frequency spectra of the applied disturbance.
It is to be understood that even though numerous characteristics of various embodiments of the present disclosure have been set forth in the foregoing description, together with details of the structure and function of various embodiments, this detailed description is illustrative only, and changes may be made in detail, especially in matters of structure and arrangements of parts within the principles of the present disclosure to the full extent indicated by the broad general meaning of the terms in which the appended claims are expressed.
Number | Name | Date | Kind |
---|---|---|---|
4604681 | Sakashita | Aug 1986 | A |
4914564 | Surauer et al. | Apr 1990 | A |
5042752 | Surauer et al. | Aug 1991 | A |
6094602 | Schade, III | Jul 2000 | A |
6138244 | Jin et al. | Oct 2000 | A |
6950270 | Lyle et al. | Sep 2005 | B2 |
6950741 | Ishikawa et al. | Sep 2005 | B2 |
7253987 | Shim | Aug 2007 | B1 |
7260463 | Yasui et al. | Aug 2007 | B2 |
7406820 | Critchley et al. | Aug 2008 | B2 |
7535192 | Takaishi | May 2009 | B2 |
8532847 | Andoh | Sep 2013 | B1 |
8630850 | Suzuki et al. | Jan 2014 | B2 |
20040080860 | Inaji et al. | Apr 2004 | A1 |
20070183076 | Baugh et al. | Aug 2007 | A1 |
20100328810 | Yoneda | Dec 2010 | A1 |
20130106336 | Sonoda | May 2013 | A1 |
Number | Date | Country | |
---|---|---|---|
20160293208 A1 | Oct 2016 | US |