ELECTRONICALLY COMMUTATED HYDRAULIC MACHINE AND OPERATING METHOD TO REDUCE GENERATION OF RESONANCE EFFECTS

Abstract

A hydraulic apparatus including an electronically commutated machine having a plurality of working chambers which are controlled on each cycle of working chamber volume to carry out active or inactive cycles of working chamber volume allows only a plurality of defined fractions of cycles to be active cycles to avoid generating frequencies of active cycles which cause low frequency resonances. The demand signal may be quantised into fractions m/n where n is an integer below a threshold selected to avoid repeating patterns of active cycles of more than a cut-off length.

Claims

1. A method of operating an apparatus, the apparatus comprising a prime mover and a plurality of hydraulic actuators, a hydraulic machine having a rotatable shaft in driven engagement with the prime mover and comprising a plurality of working chambers having a volume which varies cyclically with rotation of the rotatable shaft, a hydraulic circuit extending between a group of one or more working chambers of the hydraulic machine and one or more of the hydraulic actuators, each working chamber of the hydraulic machine comprising a low-pressure valve configured to regulate a flow of hydraulic fluid between the working chamber and a low-pressure manifold and a high-pressure valve configured to regulate the flow of hydraulic fluid between the working chamber and a high-pressure manifold, the hydraulic machine being configured to actively control at least the low-pressure valves of the group of one or more working chambers to select the net displacement of hydraulic fluid by each working chamber on each cycle of working chamber volume, and thereby the net displacement of hydraulic fluid by the group of one or more working chambers, responsive to a demand signal, the method comprising controlling the said valves to cause each working chamber to carry out either an active or an inactive cycle of working chamber volume during each cycle of working chamber volume, wherein the fraction of working chambers which carry out active cycles is variable and is selected from of a plurality of discrete fractions.

2. The method according to claim 1, wherein the plurality of discrete fractions are selected to avoid generating any repeating patterns of active and inactive cycles of working chamber volume with a length greater than a predetermined maximum repeat pattern length.

3. The method according to claim 2, wherein the plurality of discrete fractions do not include any fractions with a denominator greater than a predetermined maximum denominator, when expressed as irreducible fractions.

4. The method according to claim 1, wherein the demand signal to which the hydraulic machine responds is quantised, having one of a plurality of discrete values.

5. The method according to claim 1, wherein the discrete fractions are expressed as irreducible fractions, the denominators range up to a maximum which is selected to avoid generating repeating patterns of working chamber actuation with a frequency less than a predetermined minimum.

6. The method according to claim 1, wherein the smallest non-zero fraction in the plurality of discrete fractions is 1/n, and typically the second smallest fraction in the plurality of discrete fractions is 1/(n−1), where n is an integer.

7. The method according to claim 1, wherein the smallest non-zero fraction in the plurality of discrete fractions is selected taking into account that two or more working chambers have the same phase or that there are uneven phase differences between two or more working chambers.

8. The method according to claim 1, wherein the discrete fractions are determined by simulation or experiment, typically wherein discrete fractions are included in the plurality of discrete fractions in response to simulation or experiment showing that the frequency content of the resulting high pressure manifold pressure, or valve activation currents, or other signals, meets one or more acceptable frequency spectrum criteria and/or where the frequency content below a cut-off frequency is below a threshold, or where the effect of the selection of active and inactive cycles is found to be acceptable, or excluded if they do not meet such criteria.

9. The method according to any one preceding claim 1, wherein the discrete fractions and/or the plurality of discrete values of the quantised demand signal are calculated during runtime and/or calculated in real time taking into account pre-determined parameters and/or current measured parameters.

10. The method according to claim 1, wherein the plurality of discrete fractions are varied responsive to the speed of rotation of the rotatable shaft or another operating parameter of the apparatus, optionally wherein the method comprises switching from using a first plurality of discrete fractions to a second plurality of discrete fractions when the speed of rotation of the rotatable shaft exceeds a threshold.

11. The method according to claim 1, wherein the timing of the opening or closing of at least the low-pressure valves are regulated to vary the fraction of maximum stroke volume which is displaced by each working chamber during each active cycle, optionally wherein this enables a continuous range of displacements per revolution of the rotatable shaft to be generated although the fraction of working chambers which carry out active cycles is limited to be one of a plurality of discrete fractions.

12. The method of calculating a plurality of discrete fractions for use in the method of claim 1, the method comprising inputting a minimum allowable frequency, a target operation speed of rotation of a rotatable shaft and data indicative of the number and/or phase difference between working chambers of the machine, and/or phase difference between working chambers in a group, calculating an integer number, n, of working chamber decision points between active cycles which will lead to the generation of frequencies of cylinder activation only in excess of the minimum allowable frequency, and including 1/n in the plurality of discrete fractions.

13. The method according to claim 12, further comprising including within the plurality of discrete fractions a plurality of fractions having denominators being integers up to n and numerators being integers up to n−1, after removing duplicate values.

14. The method according to claim 12, comprising removing one or more discrete fractions from the plurality of discrete fractions to avoid the generation of repeating cylinder activation patterns with frequency components below a specific value.

15. The method according to claim 11, further comprising storing the plurality of discrete fractions on a solid-state memory device for retrieval during operation.

16. A solid-state memory device storing a plurality of discrete fractions calculated according to the method of claim 15.

17. An apparatus comprising a prime mover and a plurality of hydraulic actuators, a hydraulic machine having a rotatable shaft in driven engagement with the prime mover and comprising a plurality of working chambers having a volume which varies cyclically with rotation of the rotatable shaft, a hydraulic circuit extending between a group of one or more working chambers of the hydraulic machine and one or more of the hydraulic actuators, each working chamber of the hydraulic machine comprising a low-pressure valve configured to regulate a flow of hydraulic fluid between the working chamber and a low-pressure manifold and a high-pressure valve configured to regulate the flow of hydraulic fluid between the working chamber and a high-pressure manifold, the hydraulic machine comprising a controller configured to actively control at least the low-pressure valves of the group of one or more working chambers to select the net displacement of hydraulic fluid by each working chamber on each cycle of working chamber volume, and thereby the net displacement of hydraulic fluid by the group of one or more working chambers, responsive to a demand signal, the controller configured to control the said valves to cause each working chamber to carry out either an active or an inactive cycle of working chamber volume during each cycle of working chamber volume, wherein the apparatus is configured such that fraction of working chambers which carry out active cycles is variable and is selected from a plurality of discrete fractions.

Description

DESCRIPTION OF THE DRAWINGS

[0096] An example embodiment of the present invention will now be illustrated with reference to the following Figures in which:

[0097] FIG. 1 is a schematic diagram of an apparatus according to the invention, including an electronically commutated hydraulic machine and actuators;

[0098] FIG. 2 is a schematic diagram of an electronically commutated hydraulic machine;

[0099] FIG. 3 illustrates the procedure carried out by the electronically commutated hydraulic machine of FIG. 2 to determine the net displacement by each cylinder sequentially;

[0100] FIG. 4 is a schematic diagram of data processing to implement quantisation;

[0101] FIG. 5 is a plot of quantised output in response to a received demand signal as a function of time;

[0102] FIG. 6 is a flow diagram of a procedure for creating a quantisation table (plurality of discrete fractions); and

[0103] FIGS. 7A through 7C show the variation in fractions of cylinders active (7A), part stroke size as a fraction of maximum (7B) and scale factor (7C) with displacement demand, Fd.

DETAILED DESCRIPTION OF AN EXAMPLE EMBODIMENT

[0104] With reference to FIG. 1, an apparatus 1, for example a hydraulic excavator or other vehicle, includes an electronically commutated hydraulic machine 10 (hereafter “ECM”), comprising a first group 10A and second group 10B of working chambers, each group is respectively fluidly connected to valve block 8 via first fluid connection A and second fluid connection 21B and so that the groups of working chambers may be separately connected to one or more of high pressure manifolds 22A, 22B or 22C.

[0105] Thus as shown in this FIG. 1 embodiment the ECM 10 comprises two groups 10A and 10B, each group comprising one or more working chambers, though the number of working chambers is not illustrated in the Figure. The ECM 10 functions as the said hydraulic machine, which will be described further below with reference to FIG. 2.

[0106] The ECM may be a pump, or a motor, and in this example is operable as a pump or a motor. The ECM is driven by a prime mover 2, through a rotating shaft 4. A low-pressure manifold 6 extends from a tank to a low-pressure side input to the ECM. On the high-pressure side, the ECM has a valve block 8, which is actuatable to selectively connect different groups 10A, 10B of one or more working chambers of the electronically commutated machine to high-pressure manifolds 22a, 22b and 22c to thereby vary which working chambers are connected to each high-pressure manifold.

[0107] All of the working chambers (whether group 10A, or group 10B, or both groups at once, or one or more further groups) which are connected to a high-pressure manifold (so that they displace working fluid into or out of the same high-pressure manifold) function together as the group of one or more working chambers connected to one or more hydraulic actuators through the hydraulic circuit and it is the net displacement of working fluid by the one or more working chambers of the group connected to a particular one or more actuators which are controlled together to control or respond to the actuator, responsive to the demand signal. The invention is equally applicable where there is no option to change the allocation of working chambers to actuators.

[0108] Each of these high-pressure manifolds extends to an actuator, such as a further hydraulic machine 11. Machine 11 could be fixed displacement, or it could be variable displacement with valves that are electronically or mechanically (hydraulically) actuated and controlled, which drives a load 12, such as one or more wheels of the vehicle, through a further shaft 14, or another kind of hydraulic actuator 16, 18, for example the bucket of an excavator, or a ram etc. The actuators may function only as sinks or only as sources of hydraulic fluid but some or all may function as either a sink or source depending on a direction of actuation of the actuator. When driving an actuator the working chambers of the ECM which are connected to the actuator carry out pumping cycles and when driven by an actuator the working chambers of the ECM which are connected to the actuator carry out motoring cycles.

[0109] The apparatus comprises an apparatus controller 100 which receives control signals from an operator through one or more manual controls, and feedback signals, such as actuator position signals, or pressure signals from the individual hydraulic actuators 11, 16, 18 and/or high-pressure manifolds 22A, 22B, 22C and/or fluid connections 21A, 21B. The apparatus controller 100 processes these signals and controls the apparatus by calculating continuously variable demand signals for each group of working chambers, and sending these to the ECM. Furthermore, in the example shown the apparatus controller may also periodically send control signals to the valve block 8 to reconfigure which working chambers are connected to which actuators, for example in response to changes in current or possible future load, thereby changing which working chambers are in which group of one or more working chambers. However, valves in the valve block may alternatively be actuated via pilot pressure via hydraulic joysticks.

[0110] FIG. 2 is a schematic diagram of part of the ECM embodiment shown in FIG. 1, and shows a single group of working chambers currently connected to one or more actuators through a high pressure manifold 54. FIG. 2 provides detail on the first group 10A, said group comprises a plurality of working chambers (8 are shown) having cylinders 24 which have working volumes 26 defined by the interior surfaces of the cylinders and pistons 28 which are driven from a rotatable shaft 30 by an eccentric cam 32 and which reciprocate within the cylinders to cyclically vary the working volume of the cylinders. The rotatable shaft is firmly connected to and rotates with a drive shaft. A shaft position and speed sensor 34 determines the instantaneous angular position and speed of rotation of the shaft, and through a signal line 36 informs the ECM controller 50, which enables the ECM controller 50 to determine the instantaneous phase of the cycles of each cylinder.

[0111] The working chambers are each associated with Low Pressure Valves (LPVs) in the form of electronically actuated face-sealing poppet valves 52, which have an associated working chamber and are operable to selectively seal off a channel extending from the working chamber to a low-pressure hydraulic fluid manifold 54, which may connect one or several working chambers, or indeed all as is shown here, to the low-pressure hydraulic fluid manifold of the ECM. The LPVs are normally open solenoid actuated valves which open passively when the pressure within the working chamber is less than or equal to the pressure within the low-pressure hydraulic fluid manifold, i.e. during an intake stroke, to bring the working chamber into fluid communication with the low-pressure hydraulic fluid manifold but are selectively closable under the active control of the ECM controller via LPV control lines 56 to bring the working chamber out of fluid communication with the low-pressure hydraulic fluid manifold. The valves may alternatively be normally closed valves.

[0112] The working chambers are each further associated with a respective High-Pressure Valve (HPV) 64 each in the form of a pressure actuated delivery valve. The HPVs open outwards from their respective working chambers and are each operable to seal off a respective channel extending from the working chamber through valve block 8 to a high-pressure hydraulic fluid manifold 22, 58, which may connect one or several working chambers, or indeed all as is shown in FIG. 2. The HPVs function as normally-closed pressure-opening check valves which open passively when the pressure within the working chamber exceeds the pressure within the high-pressure hydraulic fluid manifold. The HPVs also function as normally-closed solenoid actuated check valves which the ECM controller may selectively hold open via HPV control lines 62 once that HPV is opened by pressure within the associated working chamber.

[0113] Typically, the HPV is not openable by the ECM controller against pressure in the high-pressure hydraulic fluid manifold. The HPV may additionally be openable under the control of the ECM controller when there is pressure in the high-pressure hydraulic fluid manifold but not in the working chamber, or may be partially openable.

[0114] In a pumping mode, the ECM controller selects the net rate of displacement of hydraulic fluid from the working chamber to the high-pressure hydraulic fluid manifold by the hydraulic motor by actively closing one or more of the LPVs typically near the point of maximum volume in the associated working chamber's cycle, closing the path to the low-pressure hydraulic fluid manifold and thereby directing hydraulic fluid out through the associated HPV on the subsequent contraction stroke (but does not actively hold open the HPV). The ECM controller selects the number and sequence of LPV closures and HPV openings to produce a flow or create a shaft torque or power to satisfy a selected net rate of displacement.

[0115] In a motoring mode of operation, the ECM controller selects the net rate of displacement of hydraulic fluid, displaced by the ECM, via the high-pressure hydraulic fluid manifold, actively closing one or more of the LPVs shortly before the point of minimum volume in the associated working chamber's cycle, closing the path to the low-pressure hydraulic fluid manifold which causes the hydraulic fluid in the working chamber to be compressed by the remainder of the contraction stroke. The associated HPV opens when the pressure across it equalises and a small amount of hydraulic fluid is directed out through the associated HPV, which is held open by the ECM controller. The ECM controller then actively holds open the associated HPV, typically until near the maximum volume in the associated working chamber's cycle, admitting hydraulic fluid from the high-pressure hydraulic fluid manifold to the working chamber and applying a torque to the rotatable shaft.

[0116] As well as determining whether or not to close or hold open the LPVs on a cycle by cycle basis, the ECM controller is operable to vary the precise phasing of the closure of the HPVs with respect to the varying working chamber volume and thereby to select the net rate of displacement of hydraulic fluid from the high-pressure to the low-pressure hydraulic fluid manifold or vice versa.

[0117] Arrows on the low pressure fluid connection 6, and the high-pressure fluid connection 21A indicate hydraulic fluid flow in the motoring mode; in the pumping mode the flow is reversed. A pressure relief valve 66 may protect the group within the ECM from damage.

[0118] In normal operation, the ECM intersperses active and inactive cycles of working chamber volume to meet the demand indicated by the received demand signal.

[0119] FIG. 3 illustrates the procedure carried out by the ECM controller 50 to determine the net displacement by each cylinder sequentially. The procedure begins 200, whereupon a plurality of stored variable algorithmic accumulators are set 202 to zero. A variably algorithmic accumulator is maintained for each independently controlled group of one or more cylinders (functioning as the group of one or more working chambers) so that each group may respond to an independent demand signal. The ‘algorithmic accumulator’, is more commonly known in computer science as an ‘accumulator’, however a different term is used here to differentiate from the entirely different concept of a hydraulic accumulator. The variable algorithmic accumulator stores the difference between the amount of hydraulic fluid displacement represented by the displacement demand and the amount which is actually displaced.

[0120] The rotatable shaft of the ECM then rotates until it reaches 204 a decision point for an individual cylinder. For the example shown in FIG. 1, there are eight cylinders which are phased equally apart without any redundancy, and so each decision point will be separated by 45 degrees of rotation of the rotatable shaft. The actual period of time which arises between the decision points will therefore be the period of time required for the rotatable shaft to rotate by 45 degrees, which is inversely proportional to the speed of rotation of the rotatable shaft. In some embodiments there will however be different phases between working chamber activation decision points and there may be a plurality of working chambers which can be independently controlled but which always have the same phase.

[0121] At each decision point, the ECM controller reads 206 the demand signal in the form of a displacement fraction, Fd, received from another controller (e.g. the apparatus controller) or calculated internally using signals from the hydraulic circuit, for each group of working chambers of the ECM. For each group of working chambers, the ECM controller then calculates 208 a variable algorithmic sum which equals the relevant algorithmic accumulator plus the demanded displacement for that group. The sum takes into account the period of time since the previous decision point—which can be variable bearing in mind variations in the speed of rotation of the rotatable shaft and possible variation in the phase between working chamber decision points.

[0122] Next, the status of the cylinders which are being considered is checked 210 with reference to a database 220 of working chamber status. For each cylinder 24, if it is found that the cylinder is broken or is part of a different group of cylinders which are not connected to the actuator or actuators, then no further action is taken for that cylinder at this time. Once each cylinder (if any) which have to be further considered at the decision point has been considered, the method then repeats from step 204 once the next decision point is reached.

[0123] For each cylinder for which the decision point is relevant, the algorithmic sum for the relevant group of working chambers is compared 212 with a threshold. This value may simply be the maximum volume of hydraulic fluid displaceable by the cylinder, when the only options being considered are an inactive cycle with no net displacement or a full displacement active cycle in which the maximum displacement of hydraulic fluid by the cylinder is selected. However, the threshold may be higher or lower. For example, it may be less than the maximum displacement by an individual cylinder, for example, where it is desired to carry out a partial cycle, in which only part of the maximum displacement of the cylinder is displaced.

[0124] If algorithmic sum is greater than or equal to the threshold then it is determined that the cylinder 24 will undergo an active cycle 214. Alternatively, if algorithmic sum is not greater than or equal to the threshold then it is determined that cylinder 24 will be inactive 216 on its next cycle of cylinder 24 working volume, and will have a net displacement of zero. The accumulator value will be calculated 218 according to the displacement subtracted from algorithmic sum.

[0125] Control signals are then sent to the low 52 and high 64 pressure valves for the cylinder 24 under consideration to cause the cylinder to undergo an active or inactive cycle, as determined. (In the case of pumping, it may be that the high-pressure valves are not electronically controlled and the control signals only concern the low pressure valves). The control signals are transmitted across the respective control line 56 (low pressure) and 62 (high-pressure) for the particular valve associated with the cylinder under consideration.

[0126] For each group of working chambers (cylinders) this step effectively takes into account the displacement demand represented by the displacement demand signal, and the difference between previous displacements represented by the displacement demand signal previous net displacements determined by the ECM controller (in this case, in the form of the stored error), and then matches the time averaged net displacement of hydraulic fluid by the cylinders to the time averaged displacement represented by the displacement demand signal by causing a cylinder to undergo an active cycle in which it makes a net displacement of hydraulic fluid, if algorithmic sum equals or exceeds a threshold. In that case, the value of the error is set to SUM minus the DISPLACEMENT by the active cylinder. Alternatively, if algorithmic sum does not equal or exceed the threshold, then the cylinder is inactive and algorithmic sum is not modified.

[0127] The procedure restarts from step 204 when the next decision point is reached for one or more cylinders.

[0128] It can therefore be seen that, for each group of working chambers, an algorithmic accumulator maintains a record of the difference between the displacement which has been demanded, and the displacement which has actually occurred. On each cycle, the demanded displacement is added to the displacement error value, and the actual selected displacement is subtracted. The algorithmic accumulators effectively record the difference between demanded and provided displacement and an active cycle takes place whenever this accumulated difference exceeds a threshold. Because a separate algorithmic accumulator is maintained for each distinct group of one or more cylinders which are connected together to the same high-pressure manifold, the pressure in or flow through each high-pressure manifold connected to respective actuators can be independently controlled.

[0129] One skilled in the art will appreciate that the effects of this displacement determination algorithm can be obtained in several ways. For example, rather than subtracting the selected displacement from the algorithmic accumulator variable, it would be possible to sum the displacement which has been demanded, and the displacement which has been delivered, over a period of time, and to select the displacement of individual cylinders to keep the two evenly matched.

[0130] It can be seen that when a demand signal is low, the algorithm will lead to highly pulsatile pressure ripple as periodic active cycles will be interspersed periodically between inactive cycles. If the demand signal is a fraction 1/n of maximum demand and it remains constant at that fraction, then every nth cycle of working chamber volume will be an active cycle, with the remainder being inactive cycles, and there will be pulsatile flow with a frequency of the frequency of working chamber activation decisions points 204 divided by n. There will be similar effects when the demand signal is for example near, but less than, 100% of maximum demand, as occasional inactive cycles will take place periodically between otherwise continuous active cycles.

[0131] Although these vibrations typically initiate with relatively low amplitude, the amplitude of the vibrations can increase over time, especially if the frequency of the vibrations is at (or close to) a resonant frequency of the vehicle (or part of the vehicle). These vibrations can cause damage if the amplitude increases beyond a predetermined maximum amplitude.

[0132] According to the invention, the demand signal passed to the ECM controller and used as the input for the above algorithm is quantised, such that it has only one of a predetermined group of discrete values which, as we will explain, are selected to avoid generating repeating patterns of cylinder activation in excess of a predetermined length and so frequency components below a cut-off frequency.

[0133] FIG. 4 is a schematic diagram of data processing implemented by the apparatus controller 100 and by the ECM controller 50, which together implement the invention. It will be apparent to one skilled in the art that the function of the apparatus controller and ECM controller may be combined, or still further distributed. Apparatus control program module 300 (represented by computer code executed by the apparatus controller) processes feedback signals 310, 312, 314 received by the apparatus controller from the actuators and high-pressure manifolds. These signals may include pressure measurements, actuator position or speed measurements etc. The apparatus controller also receives operator command signals 316 which can be input through a user interface such as a touch screen or keyboard, or a manual control, such as a joystick or lever used to control an actuator (e.g. to control the operation of hydraulic actuators of an excavator and/or to drive the vehicle). This data is used to calculate current displacement demand signals 301A, 301B, 301C for each of the groups of working chambers. In this example the displacement demand signals are expressed as Fd (fraction of maximum displacement per rotation of the rotatable shaft). These signals are then digitally processed by the apparatus controller to implement hysteresis using hysteresis logic 302A, 302B, 302C which outputs partially processed displacement demand signals 303A, 303B, 303C.

[0134] Hysteresis is useful to prevent chattering between adjacent quantisation steps and is used for all quantisation methods. The level of hysteresis in systems with no integral term, such as negative flow control systems. is specific to the compliance of the system and the relationship between pressure and displacement (which in some cases may be a proportional gain). Hysteresis is not effective with systems that have integral terms when using quantised active cycle fractions and with only full pumping strokes available; it only acts to modify the frequency of the displacement cycling. When designing a hysteresis system it is preferable to take into account that a human operator will effectively compensate for slight errors between displacement of hydraulic fluid produced and demanded, for example by adjusting a joystick position to achieve an actuator position. In some embodiments, hysteresis is only provided when the displacement demand is being reduced, and not when it is being increased. This is especially useful in the variable stroke volume embodiment described below.

[0135] Accordingly, the continuous demand signal which is fed to the ECM controller 50 and processed using the algorithm described with reference to FIG. 3 is quantised and has typically also been processed to introduce hysteresis.

[0136] The partially processed displacement demands are then quantised 304A, 304B, 304C.

[0137] With reference to FIG. 5, instead of passing the originally calculated displacement demand signal 400 to the ECM controller 50, the demand signal is quantised, i.e. made to correspond to one of a plurality of different displacement fractions 402A, 402B, 402C, 402D, 402E, etc. This is carried out with reference to a solid state memory storing a data structure 306 setting out a plurality of discrete fractions. Active cycle fractions could also be calculated during run time without having a stored table. The quantised demand signals 305A, 305B, 305C are then passed to the ECM controller 50. The active cycle fraction could also be calculated during run time without having a stored table.

[0138] The discrete fractions are selected to avoid the generation of patterns of active, or inactive, cycles of cylinder volume, with a frequency content which is below a determined cut-off frequency, assuming a predetermined minimum speed of rotation of the rotatable shaft. Pressure pulsations in a hydraulic line will arise from and have the same frequency content as that found in the enabling patterns comprised of active and inactive cycles of cylinder working volume. This vibration can be transmitted to components. The aim of the quantisation control method is to prevent mechanical components of a system/vehicle (either directly, or indirectly, e.g. via excitation of the operator) from being excited at their natural frequency. This may arise if cylinders are enabled at the same frequency as the natural frequency of the mechanical component where there is some form of path (e.g. a mechanically coupling path) for the vibration to be transferred from the pump (or the connected hoses/pipes), to the mechanical component.

[0139] Using quantisation to remove frequencies from the cylinder enabling patterns will prevent certain displacement levels from being commanded by the ECM controller. The displacement levels can be defined in terms of volume of fluid per shaft revolution or fraction of maximum displacement of fluid per shaft revolution. When the continuous displacement level demand is not equal to one of the discrete displacement levels, the nearest discrete displacement level is selected and there is a resulting error between the continuous displacement demand and discrete displacement level. In this instance, there will therefore be an error between the demanded volume and the delivered volume of the pump. This will not be an issue in systems whereby an error is tolerated between the exact volume of fluid produced and the volume delivered.

[0140] Additionally, the human operator will effectively compensate for the slight errors between the volume of oil demanded and produced. The operator will adjust the joystick positions to achieve the desired actuator positions.

[0141] The importance of the ‘minimum frequency’ is that no other frequencies below it will be present in cylinder enabling patterns, when using quantisation. If the ‘minimum frequency’ selected is above the natural frequency of the component then the mechanical component will not resonate at its natural frequency.

[0142] To this end, the group of discrete fractions may consist of fractions having denominators up to an integer n, where n is selected so that, at the expected speed of shaft rotation, the frequency of selection of an active cycle of cylinder volume, at displacement fraction 1/n, is above the cut-off frequency.

[0143] For example, if a machine has 12 equally spaced cylinders, and rotates at 1000 rpm, then a cylinder selection decision is reached every (60/1000)/12=5 milliseconds. If the largest denominator is 5, a cylinder will carry out an active cycle every 25 milliseconds, at an active cycle fraction of 1/5 and so the smallest frequency which will be present is 40 Hz. This can be seen in the following example pattern of cylinder activation:

TABLE-US-00001 TABLE 1 Cylinder Fd Accumulator active or Time Phase Cylinder demanded value inactive? (ms) (degrees) 1 0.2 0 Active 0 0 2 0.2 0.2 Inactive 5 30 3 0.2 0.4 Inactive 10 60 4 0.2 0.6 Inactive 15 90 5 0.2 0.8 Inactive 20 120 6 0.2 0 Active 25 150 7 0.2 0.2 Inactive 30 180 8 0.2 0.4 Inactive 35 210 9 0.2 0.6 Inactive 40 240 10 0.2 0.8 Inactive 45 270 11 0.2 0 Active 50 300 12 0.2 0.2 Inactive 55 330

[0144] The above table demonstrates that where Fd=1/n (in this case 5), a pattern is generated with a repeat every n cylinders. For m/n, where m and n are both integers, expressed as an irreducible fraction (i.e. where m and n have no common divisor other than 1) there will again be a pattern (during which m cylinders undergo active cycles) with a sequence length of n.

[0145] For example, the group of allowable fractions may be each fraction m/n which is an irreducible fraction and where n is from 1 to a predetermined maximum integer (in this example 5), and m is less than n (for each value of n). An example table for n=5 is shown below:

[0146] Allowable Fd:

TABLE-US-00002 TABLE 2 0 ⅕ ¼ ⅓ ⅖ ½ ⅗ ⅔ ¾ ⅘ 1

[0147] More generally, for x cylinders which are equally distributed around a shaft with a minimum operating rotational speed r (rotations per second), a pattern which repeats every n cylinders will generate oscillations with a frequency of xr/n.

[0148] Tables with a greater maximum sequence length n can be generated by including each irreducible fraction m/n for each m up to n−1 and each n up to a determined maximum.

[0149] A larger repeating pattern length (determined by n) will lead to proportionately lower frequencies but a larger table length.

[0150] For example, for n=12 the corresponding table will be:

TABLE-US-00003 TABLE 3 0 1/12 1/11 1/10 1/9 ⅛ 1/7 ⅙ 2/11 ⅕ 2/9 ¼ 3/11 2/7 3/10 ⅓ 4/11 ⅜ ⅖ 5/12 3/7 4/9 5/11 ½ 6/11 5/9 4/7 7/12 ⅗ ⅝ 7/11 ⅔ 7/10 5/7 8/11 ¾ 7/9 ⅘ 9/11 ⅚ 6/7 ⅞ 8/9 9/10 10/11 11/12 1

[0151] In practice, the fractions may be stored in binary form which will require some rounding depending on the number of significant bits which are stored. Alternatively, the displacement fraction may be calculated without the use of a stored table.

[0152] It is notable that in such tables, the smallest non-zero fractions, and therefore the smallest values of Fd which are implemented by ECM will be 1/n, 1/(n−1), 1/(n−2) . . . (until the next number in the sequence is larger than or equal to 2/n). The largest non-unity fractions will be (n−1)/n, (n−2)/(n−1), (n−3)/(n−2).

[0153] The largest displacement band gives a value for the largest displacement gap that will be present in the table—typically this will be 1/n. This gives an indication of the coarseness of the displacement steps after quantisation and therefore how acceptable the quantisation table will be. Very coarse steps in displacement may prevent accurate control of actuators which may be of particular concern in vehicle applications.

[0154] In an experiment, using an excavator having an operator cab which can oscillate at low frequencies (about 3 to 15 Hz), with the ECM driving hydraulic actuators fluidly connected to the ECM, the quantisation table of Table 2 (maximum sequence length of 12, i.e. n=12) did not excite the cab but gave sufficiently coarse displacement steps to provide an unacceptable user experience; increasing the sequence length to 24, i.e. n=24 did not excite the cab but provided displacement steps which gave an acceptable user experience; further increasing the sequence length to 36, i.e. n=36 provided an acceptable step size but the frequency content excited the cab.

[0155] Thus, there is a trade-off between minimum frequency versus the coarseness of the quantised displacement levels. Gaps of up to 5 to 10% will be acceptable in some applications. Gaps can be reduced by selecting an ECM with a greater number of working chambers at different phases, or by choosing a prime mover with a higher minimum shaft speed (or restricting the minimum shaft speed), and selecting a higher maximum denominator.

[0156] Although resonance of the cab itself is a primary concern in the particular case of a hydraulic excavator, the excitation and resonance of other bodies is also of concern.

[0157] For example, movement of a vehicle cab can cause resonance of the individual operator which in turn may cause unintended movement of the joystick, thus potentially making the situation worse. The present invention is especially useful for avoiding low frequency resonance effects.

[0158] Furthermore, some displacement fractions may be deemed unallowable, for example due to a risk of exciting further resonances, and displacement fractions which are deemed unallowable can be excised from the table of displacement fractions.

[0159] In the example of Table 1, cylinders are equally spaced in phase and there is no redundancy (n cylinders are configured so that their volume cycles are spaced apart by phase 360/n°). However, ECMs are known in which the working chambers are not spaced equally in phase and/or where there is redundancy, by which we refer to a plurality of working chambers having the same phase as each other. The latter is common where cylinders are driven by a multi-lobe cam, for example, meaning that one or more working chamber cycles take place within a single rotation of the rotatable shaft. Unevenly phased working chambers may occur due to the design of the ECM or due to the allocation of working chambers to different groups of working chambers during operation.

[0160] For example, an ECM has 24 cylinders which are equally spaced in phase (360/24=15° apart). Groups of three cylinders which are 120° apart have high pressure outputs which are commoned together, giving eight independent outputs. Three of these independent outputs are connected to a first high-pressure manifold, four of these independent outputs are connected to a second high-pressure manifold, and 1 independent output is connected to a third high-pressure manifold.

[0161] The phasing of the 9 cylinders connected to the first high-pressure manifold may be as follows:

TABLE-US-00004 TABLE 4 Cylinder number 1 2 3 4 5 6 7 8 9 Phase (°) 0 30 60 120 150 180 240 270 300 Phase — 30 30 60 30 30 60 30 30 difference

[0162] Table 4 shows that in this embodiment the phasing between consecutive cylinders is sometimes 30° and sometimes 60°. There is therefore unequal phasing between cylinders.

[0163] In the example in table 4, the repeating cylinder phase pattern length, is 3. This number indicates how many cylinders are required to repeat the cylinder phasing. The phase difference between cylinder 1 and cylinder 2 is 30. The phase difference between cylinder 2 and cylinder 3 is 30. The phase difference between cylinder 3 and cylinder 4 is 60. This pattern then repeats. Since it takes 3 cylinders to repeat this pattern, the repeating cylinder phase pattern length is 3.

[0164] A machine may also be designed to have cylinders with duplicate phasing (redundancy). The table below shows a 6-cylinder machine with a redundancy of 2

TABLE-US-00005 TABLE 5 Cylinder number 1 2 3 4 5 6 Phase (°) 0 0 120 120 240 240

[0165] The quantisation tables for such machines should be created taking into account a requirement to guarantee that a certain maximum sequence length limits the lowest frequency in the expected manner.

[0166] If all working chambers have a redundancy of greater than 1 then the denominators of the fractions may be selected to be multiples of the redundancy. Thus, where the redundancy is 3, the table may include the fractions 1/3, 1/6, 1/9, 1/12, 1/15 etc. If the machine has unequally spaced working chambers, one option is to select all the denominators which are multiples of the repeating cylinder phase pattern length. This will give the same minimum frequency as an equally spaced machine of the same number of cylinders.

[0167] It is therefore the case that in order to limit the lowest frequency, the allowable displacement levels with machines or services with unequal phasing or redundancy will be reduced. This will result in further coarseness in the quantisation table.

[0168] The following table shows the effect of working chamber redundancy in a machine with 12 cylinders. The table indicates which cylinders carry out active cycles at a displacement fraction of 1/3.

TABLE-US-00006 TABLE 6 Redundancy = 1 Redundancy = 2 Redundancy = 3 Enabled Enabled Enabled cylinder cylinder cylinder Cylinder phase Cylinder phase Cylinder phase phase/ Active difference phase/ Active difference phase/ Active difference degrees cycle? (degrees) degrees cycle? (degrees) degrees cycle? (degrees) 0 N 0 N 0 N 30 N 0 N 0 N 60 Y 90 60 Y 120 0 Y 90 90 N 60 N 90 N 120 N 120 N 90 N 150 Y 90 120 Y 60 90 Y 90 180 N 180 N 180 N 210 N 180 N 180 N 240 Y 90 240 Y 120 180 Y 90 270 N 240 N 270 N 300 N 300 N 270 N 330 Y 90 300 Y 60 270 Y 90 0 N 0 N 0 N 30 N 0 N 0 N 60 Y 90 60 Y 120 0 Y 90 90 N 60 N 90 N 120 N 120 N 90 N 150 Y 90 120 Y 60 90 Y 90 180 N 180 N 180 N 210 N 180 N 180 N 240 Y 90 240 Y 120 180 Y 90 270 N 240 N 270 N 300 N 300 N 270 N 330 Y 90 300 Y 60 270 Y 90

[0169] Table 6 shows that when there is a redundancy of 1, there is a repeating pattern every 90° (i.e. the pattern repeats four times per rotation of the rotatable shaft and so at four times the frequency of rotation of the rotatable shaft); and when there is a redundancy of 3, there is a repeating pattern every 90° (i.e. the pattern repeats four times per rotation of the rotatable shaft and so at four times the frequency of rotation of the rotatable shaft). However, when there is a redundancy of 2, the phase difference between enabled cylinders is sometimes 120 degrees and the phase difference between enabled cylinders is sometimes 60 degrees. This causes a repeating pattern every 180° (i.e. the pattern repeats every half rotations of the rotatable shaft and so at twice the frequency of rotation of the rotatable shaft); In the examples with a redundancy of 1 and a redundancy of 3, an enabling fraction of 1/3 causes a frequency at 4 times the frequency of the shaft rotation. An enabling fraction of 1/3 causes a lower frequency at 2 times the frequency of the shaft rotation.

[0170] From this example it is clear that lower frequencies become present when denominators that are not integer multiples of the redundancy are used in the quantisation tables. If it was desired to remove frequencies below 2 times the frequency of shaft rotation, it would not be possible to use an enabling fraction of 1/3 in the case whereby the cylinder phasing had a redundancy of 2.

[0171] In such cases, the quantisation tables for the embodiments with redundancy >1, consist of fractions with denominators which are multiples of the redundancy. E.g. for n up to 18, the following fractions are calculated, then sorted and duplicates are removed: 1/3, 2/3,3/3, 1/6, 2/6, 3/6, 4/6, 5/6, 6/6, 1/9, 2/9, 3/9, 4/9, 5/9, 6/9, 7/9, 8/9, 9/9, 1/12, 2/12, 3/12, 4/12, 5/12, 6/12, 7/12, 8/12, 9/12, 10/12, 11/12, 12/12, 1/15, 2/15, 3/15, 4/15, 5/15, 6/15, 7/15, 8/15, 9/15, 10/15, 11/15, 12/15, 13/15, 14/15, 15/15, 1/18 2/ 18 3/18 4/18 5/18 6/18 7/18 8/18 9/18 10/18 11/18 12/18 13/18 14/18 15/18 16/18 17/18 18/18.

[0172] Reduced to irreducible fractions this gives:

[0173] Allowable Fd:

TABLE-US-00007 TABLE 7 0 1/18 1/15 1/12 1/9 2/15 ⅙ ⅕ 2/9 ¼ 4/15 5/18 ⅓ 7/18 ⅖ 5/12 4/9 7/15 ½ 8/15 5/9 7/12 ⅗ 11/18 ⅔ 13/18 11/15 ¾ 7/9 ⅘ ⅚ 13/15 8/9 11/12 14/15 17/18 1

[0174] More generally, with reference to FIG. 6, the procedure for determining the quantisation table starts by calculating 500 the repeating cylinder pattern, which will depend on the relative phase difference of the individual cylinders, and whether and to what extent there is redundancy between the cylinders. Broken cylinders may also be taken into account, prior to operation (as in FIG. 6, for example by simulation or experiment), or during operation. In a simple example where there is no redundancy and each cylinder is equally phased apart, then the repeating phase difference is simply the phase spacing between the cylinders. If the cylinders are not equally spaced then the repeating pattern should be calculated by identifying the number of cylinders required to produce a repeating phase difference pattern and then summing the phase differences between all of the cylinders. This is used to determine the phase difference between repeating arrangements of working chambers. In the examples of Tables 4 and 5 this is 120°. For a machine with c cylinders which are equally spaced, with redundancy r, this is 360*r/c. The number of cylinders required to generate a repeating pattern is also determined. In the examples of Tables 4 and 5 this is 3.

[0175] In the next step, allowable denominators of the displacement fractions are calculated 502. This is calculated using the minimum expected operating shaft speed of rotation and using the number of cylinders and phase difference between repeating patterns of cylinders calculated in the previous step, and this step also includes the minimum acceptable frequency. From this, the allowable denominators, which do not lead to repeating patterns having a frequency below the minimum frequency, can be calculated. In the example of FIG. 6, with redundancy 3, with a shaft speed of 1500 rpm, and a minimum frequency of 15 Hz, the allowable denominators are 3, 6, 9, 12, 15.

[0176] Thereafter, allowable Fd values (i.e. displacement fractions which are selected as one of the available quanta) are calculated 504, using these denominators. Typically for each allowable denominator n, the quantisation table will include each m/n where m is each integer from 1 to n for each value of n.

[0177] Next, the calculated fractions are processed by removing duplicates 506 and sorting them into numerical order. In the next stage, which is optional, some Fd values may be filtered out 508 of (removed from) the calculated list, because they may generate some other resonance, e.g. of another component of the apparatus.

[0178] Thereafter there is a validation step 510, in which the calculated allowed Fd values are analysed to determine whether they provide sufficiently smooth operation for a user.

[0179] The final set of calculated FD values is then stored 512 in memory and used during operation of the machine. As mentioned above, there may be different tables of allowable Fd for different shaft speeds, or operating modes of the apparatus, for example for when different groups of working chambers are connected to an individual high-pressure manifold.

[0180] In the above examples, the apparatus controller 100 has created quantised demand signals and avoided the generation of repeating patterns of cylinder activation beyond a predetermined length without a requirement to modify the ECM controller 50 or to change the algorithm (sigma-delta algorithm) which it employs. Accordingly, the precise cylinders which are actually caused to carry out active cycles in order to implement the demanded displacement are determined by the ECM controller.

[0181] Typically, they are not predetermined and will vary from one use to the next, depending on the time history of shaft revolutions and demanded displacement.

[0182] As we have explained, it is normally advantageous for electrically commutated hydraulic machines to intersperse active and inactive cycles of working chamber volume to meet fractional displacement demands and commonly each active cycle has the same net displacement, which is the maximum net displacement of each working chamber. However, with reference to FIGS. 7A to 7C we will now describe an embodiment in which stroke volume of the working chambers, during active cycles is reduced by amending valve timing. Although this can be less efficient in some ways, this can be combined with the quantisation approach discussed above, leading to a reliable machine which suppresses the generation of undesired, e.g. low frequency, vibrations which can provide a wide (and in some embodiments) continuous range of displacement fractions.

[0183] In these embodiments, the net displacement during an active cycle is reduced below 100% of the maximum displacement by varying the timing of the active control of the low-pressure and high-pressure valves. Methods for doing so are known from WO 2004/025122. For example, during a pumping cycle, the timing of the closure of low-pressure valve may be delayed from its usual phase, shortly after the point of maximum cylinder volume (top dead centre). For a short delay this gives a slightly reduced displacement. If the closure of the low-pressure valve is delayed until close to the point of minimum cylinder volume (top dead centre), the displacement is reduced to a small fraction of maximum displacement. In the case of a motoring cycle, the low-pressure valve is opened and the high-pressure valve closed earlier than would otherwise be the case during the expansion stroke (from top dead centre to bottom dead centre), reducing the volume of working fluid which is received from the high-pressure manifold.

[0184] This step usually occurs late in the expansion stroke and bringing it forward slightly will slightly reduce the displacement whereas bringing it forward to shortly after the point of minimum cylinder volume will greatly reduce the net displacement.

[0185] In operation, for any given value of received (e.g. calculated or input) displacement demand Fd (x-axis), the Fd is multiplied by a scaling factor 406, which is intended to ensure that the quantised displacement fraction chosen will always be greater than the demand, so that by reducing the volume delivered by each cylinder (by adjusting the valve timing) the actual demand displacement can still be achieved. FIGS. 7A and 7B employ the scale factor. As can be seen in FIG. 7A, the fraction of cylinders which carry out active cycles is quantised as before, thereby suppressing the generation of unwanted frequency components. However, the valve timings are amended such that the total net displacement more closely matches the demanded displacement. It is desirable that the stroke size of each cylinder is kept as close as possible to 100%, in order to achieve this the scale factor used to ‘round up’ the displacement must itself be a function of the Fd, as seen in FIG. 7C. By using a function of this type it is possible to keep the quantised Fd at an approximately fixed level above the Fd demand, and thus ensure that the stroke size is maximised.

[0186] By a way of example, 0.9 on the y-axis of FIG. 7B corresponds to net displacement by each cylinder being 90% of the volume delivered when using a full stroke (maximum). The quantisation of the displacement demand is useful to control the frequency content of the machine output, as a result of the cylinder enabling algorithm, and it can be seen from the left hand side of FIGS. 7A to 7C that at low displacement demands the frequency of cylinder activation does not drop below a threshold (approximately 0.1), instead the part stroke volume decreases. This avoids the generation of pulse patterns with very low frequency content, whilst still enabling the output displacement closely matched the input displacement demand. A similar effect can be seen at high displacement demand, where the method avoids generation of low frequency patterns of cylinder inactivation.

[0187] As shown in the above and in FIG. 7A, it can be seen that the quantised demand sent to the ECM controller 50 is always higher than the continuous displacement demand. The requirement is that the quantised active cycle fraction is higher than the continuous displacement demand, and means the part stroke size can be 1 or lower in order to achieve the continuous displacement demand exactly. If the quantised demand were lower than the continuous displacement demand, then the part stroke size required to achieve the continuous displacement would need to be larger than 1, which is impossible.

[0188] Although it is possible to leave gaps in displacement which cannot be met, if it is ensured that the quantised demand signal is larger than the continuous displacement demand throughout the displacement range then gaps can be avoided. In this example, this is achieved by multiplying the continuous displacement demand by the scale factor shown in FIG. 7C although this is not the only possible approach. For example, a bias could be applied to the continuous demand and this may also vary throughout the displacement range.

[0189] In an alternative embodiment, gaps are addressed by selecting the discrete active cycle fraction closest to the continuous displacement demand and applying no upwards hysteresis, but only downwards hysteresis. These methods prevent requesting a part stroke fraction higher than 1 of the enabling cylinders. For reasons previously mentioned, it is preferable that the part stroke size is as close to full stroke size as possible throughout the displacement range.

[0190] Hysteresis can prevent jumping between quantised steps when the demanded displacement is noisy. In the case shown in FIG. 7A, where the Fd (straight line continuous displacement demand) signal is smooth, hysteresis (or scaling) may be omitted and it would be sufficient to round up to the nearest quantised step which is above the (straight line continuous displacement demand) Fd. Unfortunately, in reality the demand signal will contain noise, and thus we need some hysteresis, by which we mean a difference in threshold on the decision to change up a step versus to change down a step. Applying hysteresis to the quantiser prevents switching back and forth between steps if there is noise, provided there is sufficient hysteresis. If the noise level is bigger than the steps themselves, no amount of hysteresis alone will help.

[0191] An alternative and potentially preferable approach is to employ backlash. Backlash prevents the output signal from changing when the rate of change of the input signal changes sign. This usually has a single parameter, called ‘deadband’, which is the amount of the difference between input and output which will cause the output to start following input again. This type of signal processing often causes an offset between the input and output signals. It is possible to correct for the offset by using scaling, such as that shown in the graph of FIG. 7C. The scaling function is of type y=n/x+1 where n is half of the deadband width.

[0192] In the above examples, the discrete values in the quantisation tables correspond to the discrete fractions of working chambers which will carry out active cycles of working chamber volume. This arises because the units of the demand signal are displacement fraction. However, this is not essential.

[0193] Further variations and modifications may be made within the scope of the invention herein disclosed.