IETdo

www.ietdl.org

Published in IET Science, Measurement and TechnologyReceived on 21st August 2008Revised on 15th December 2008doi: 10.1049/iet-smt:20080124

ISSN 1751-8822

Dynamic analysis of a bistable actuatorfor digital hydraulicsJ.-P. Uusitalo1 L. Soderlund1 L. Kettunen1 M. Linjama2

M. Vilenius2

1Department of Electronics, Tampere University of Technology, P.O. Box 527, FI-33101 Tampere, Finland2Department of Intelligent Hydraulics and Automation, Tampere University of Technology, P.O. Box 527, FI-33101 Tampere,FinlandE-mail: jukka-pekka.uusitalo@tut.fi

Abstract: A fast and small electromagnetic actuator, which requires a small amount of work to switch betweentwo stable positions, for the use in digital hydraulics, is presented. A dynamic study of the bistable actuator aspart of a new kind of hydraulic on/off valve is carried out. A prototype is built and measured to verify theanalysis. The results show that with careful elaboration the design is feasible. Furthermore, the responsetimes depend on driving frequencies and an optimal current density for minimal energy consumption in cyclicaction can be found. A single on/off valve is a critical part of a digital hydraulic system.

i

1 IntroductionDigital hydraulics means the use of parellel on/off valves inhydraulic control. If digital hydraulics is to be madeeconomically competitive, the valves (or other means)controlling the flow need to be developed [1–3]. Therequirements of the valves are sufficient switching rate,small size and small energy consumption. The commercialvalves do not yet fulfil these requirements.

The most often used valve in digital hydraulics is a directlydriven on/off seat valve including a solenoid and a spring asthe actuator for bidirectional movement. In the literatureideas to reduce solenoid valves switching time and volumewith help of a bistable construction or add on electronicshave been presented in [4, 5]. For instance, in the Sturmandigital latching spool valve [6] both electromagnetic andhydraulic parts are optimised and highly integrated.

To caracterise valves we introduce a so-called flow densitynumber. It is defined as the flow capacity divided by thevolume of the valve, and it is one of our key characteristicsin valves. The miniaturised bistable seat valve presented in[7] has one of the highest flow density values with acompetitive response time among all directly driven on/off

Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243: 10.1049/iet-smt:20080124

valves. This valve is a highly integrated combination of anelectromagnetic actuator that exploits permanent magnetsand a non-leaking seat type hydraulic part with reducedforces [8]. This paper presents a more profound dynamicanalysis and an optimised version of the actuator presentedin [7].

2 Background2.1 Why bistable electromagneticactuators?

Digital hydraulics has a need for small actuators, which stillprovide the same high opening force for a seat valve. Thechallenge in all designs, including the electromagneticactuators, is also to simultanously reach a sufficiently highoperating frequency rate.

A direct driven seat valve offers the best sealed andpredicted valves, but it requires a big opening force. Whenthis option is chosen, the best choice for the actuator is theelectromagnetic one. An electromagnetic actuator has highpower and work densities. In this case, a relatively smallneed of opening (in the hydraulic part) compared to thesize of the actuator also benefits this choice, since the airgaps stay small.

235

& The Institution of Engineering and Technology 2009

23

&

www.ietdl.org

Electromagnetic actuators can be made smaller byincreasing the used maximum current density. The limitingfactor is the heat production of the coils compared to thesize of the actuator and thus the highest possible rmscurrent available.

The traditional electromagnetic on/off actuators aremonostable devices and they need external energy in orderto maintain one of their two positions. A spring holds thearmature in one position and the electromagnetic forceusing constant power holds it in the other one. This resultsin a big coil with a large time constant.

Bistable electromagnetic actuators need to produce the sameopening force as the monostable actuators, but the force isneeded only transiently, that is, as long as the armature ofthe valve is changing its position. Therefore bistableactuators need less rms power resulting in smaller and fastercoils, or on the other hand, an improved maximum operatingfrequency.

The bistability of electromagnetic actuators is reached withpermanent magnets. The use of Nd-Fe-B permanent magnetprovides high energy density rates, which provides highforces.

For the actuator presented in this paper, no add onelectronics, for example boosting or cooling, is used. Theycould increase the level of reachable current densities anddecrease switching time, as in [9], but this paper is limitedto study the actuator design and its use in hydraulics.

2.2 The necessity of rapid cycled action

When considering the performance of a digital (on/off)valve, there are two important frequencies one must takeinto account. One is the maximum continuous operatingfrequency and the other is maximum operating frequency.The difference between these two is described in Fig. 1,

Figure 1 Different operating modes for a solenoid and abistable actuator

In case of a solenoid valve (a) the only possibility to changeoperating frequency, is to lower it. In case of a bistable valve,mean frequency of 25 Hz (or lower) can be achieved withtraditional steady switching (b) or the valve can have a fromswitch to switch changing random frequency with a limit of100 Hz maximum frequency (c)

6The Institution of Engineering and Technology 2009

where possible operating modes for a solenoid valve and abistable valve with the same hydraulic properties arepresented. Traditional solenoid valves do not have the samequantities because they are based on different operatingprinciples. The difference between driving a solenoid or abistable valve with voltage pulses is described in Fig. 2.

The miniaturised bistable seat valve presented in [7] hadone problem. The maximum operating frequency (in cycles)was measured to be only 45 Hz, while the response timesof the static models predicted much more. The problemwas the lack of dynamic models in rapid cycled action.Another problem was probably a too low coercive forcevalue of the used permanent magnet in 808C.

When a valve is driven at a higher frequency than it cancontinuously tolerate, it is sometimes referred to as ‘overfrequency’ (Fig. 1c). Over frequencies are important indigital hydraulic valves due to controllability of the oil flow,which improves the fidelity of the whole digital hydraulicsystem. Fig. 3 presents a digital hydraulic valve package,often referred to as digital flow control unit (DFCU). ADFCU with four of the valves studied in this paper, can becontinuously controlled at 100 Hz. Additional valvesincrease flow control possibilities. Any single valve is notheated up too much because it is not needed all time, butthere are replacing parallel valves in the DFCU.

For these reasons, it is very important to study the valve atrapid cycled action – over frequencies – using dynamicmodels, although a valve cannot continuously be operatedin this mode.

3 Technical design and analysis3.1 Basic construction of the valve

The basic construction of the valve consists of two coils, amagnetic circuit, an armature with the permanent magnet

Figure 2 Comparison of control signal for a hydraulic valveand corresponding voltage pulse to actuator

In the control signal switching the value to ‘1’ means opening thevalve and switching the value to ‘0’ means closing the valve

IET Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243doi: 10.1049/iet-smt:20080124

IETdo

www.ietdl.org

and the hydraulic part: a shaft attached to the armature, a seatand flow paths. A part of the magnetic circuit is used to createa pressure vessel around the moving parts to isolate the coilsfrom the hydraulic oil (Fig. 4). The pressure vessel does nothold the whole 210 bar of maximum operating pressure byitself, but it is supported by the coil formers and otherparts of the magnetic circuit.

The use of the magnetic circuit as a pressure vessel providesan unnecessary path for the magnetic flux to go around thepermanent magnet. This leads to a loss of force. The loss,however, can be minimised by making the pressure vesselthin and using the other parts of the valve to mechanicallycomplete it as described earlier. This causes the pressurevessel to be saturated close to the permanent magnet(Fig. 5) and the loss of force is minimised. The choice for

Figure 3 Digital flow control unit (DFCU) with five parallelvalves

Figure 4 Valve construction (from [7])

1. permanent magnet, 2. upper coil former, 3. lower coil former,4. magnetic circuit, 5. pressure vessel (a part of magneticcircuit), A. input pressure, B. force compensating pressure,C. output pressure, D. seat, E. armature

Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243i: 10.1049/iet-smt:20080124

this kind of construction of the pressure vessel is made forthe sake of mechanically simple design in prototypeconstruction.

In this valve, the hydraulic forces needed to be exceeded bythe actuator differ from those in a traditional seat valve. Thetraditional (normally closed) solenoid seat valves have usuallytwo reasons for the need of external power in the non-stableposition: the spring and the hydraulic properties of the seatvalve. The problem with the hydraulic properties, thespontaneous closing especially at high-flow rates, has beenlately solved using several hydraulic restrictions in thehydraulic part [8].

The use of several hydraulic restrictions as a pressurecompensator reduces forces and even gives them a bistableform. The sign of the pressure force changes from heavyclosing to small opening when the valve is opened (Fig. 6),at least when there is a considerable pressure difference overthe valve. This type of seat is more suitable for a bistableactuator construction than a typical seat valve and it istherefore the choice to our valve.

The geometry of the magnetic circuit is designed usingstatic magnetic field calculations [7]. The most importantcriteria in static calcultations is to find a force higher thanthe hydraulic force.

3.2 Study on dynamics and responsein cyclic action

As mentioned earlier, static magnetic models are used todesign an adequate geometry. However, the time intervals

Figure 5 Figure shows a typical situation, where thepressure vessel is saturated close to the permanent magnet

In this case there is a current in the coils, but the pressure vesselis close to saturation even without the current. The values in thepicture are the values of relative permeability

237

& The Institution of Engineering and Technology 2009

23

&

www.ietdl.org

of practice are usually so small that dynamic effects have to betaken into account and carefully studied.

Here is a detailed, but still intuitive description of whathappens when the valve is at rest (steady state) and thenopened with a voltage pulse leading to a time dependingdynamic case.

1. The fast raising current induces an increasing magneticfield according to the Biot–Savart law. The relatively bigchange of the magnetic flux in the magnetic circuit induceshigh screening currents.

2. Owing to the dynamic effects, the generated magnetic fluxruns in the skin of the magnetic circuit, which saturatesquickly.

3. When the increase of the current slows down, thescreening currents decrease, the skin depth increases andthe magnetic flux continues to increase in a wider cross-section of the magnetic circuit.

4. The magnetic field induced by the coils will have asignificant effect on the permanent magnet only after theskin depth is bigger than the wall of the pressure vessel. Inpractice this point is reached almost instantly.

5. When the pushing and pulling forces exerted on thepermanent magnet are bigger than the hydraulic forces, thearmature starts to move.

6. The moving magnet causes a changing magnetic flux inthe actuator. This change induces electric currents in theconducting parts. In practice, when the speed of themagnet is high enough, the current in the coils may start toreduce instead of growing. In hydraulic systems this isoften called the back-EMF effect.

Figure 6 Pressure forces of a typical seat valve comparedwith a seat valve with reduced axial forces

8The Institution of Engineering and Technology 2009

7. If back-EMF shows in the current, the reducing currenthas some effects in the actuator. It causes a reducingmagnetic flux in the magnetic circuit. This then induces ascreening current in the opposite direction than previously.Normally the screening currents are not useful in theactuator, but here they help to keep the generated force up.

8. In this valve, the pressure forces drop rapidly when thearmature moves towards the opened position. This offers ahigh acceleration phase for the armature. In high-speedmovement, however, a new repulsing force emerges. Theviscosity friction, which is approximately linear to the speedof the armature, will have significant effect on themovement before the armature reaches the opened position.

9. When the armature reaches the other end and stops, theback-EMF disappears and the current of the coils start toraise again. This then causes a third screening current, tothe same direction as the first one. Ideally, the voltage ofthe coils should be cut immediately after the movement, sothat the last raise of current would not happen at all.

10. When the voltage of the coils is turned off, the currentstarts to decrease. This induces a fourth screening current,again to the other direction than the previous one. Thiscauses the magnetisation to remain in the inner parts of themagnetic circuit for some time.

The screening currents have a strong influence on thedistribution of the flux. After a short voltage pulse, the fluxcaused by the coils may not have reached the whole cross-sectional area of the magnetic circuit. Also, the same fluxmay not have disappeared before the next voltage pulsewith opposite poles is driven (the valve is closed).

When the voltage is set up to opposite poles, the actuatorrepeats the events that occurred before, but with reversedfluxes, currents and so on. The only difference is that theremaining flux (and screening currents) of the first pulse keepthe permanent magnet attached to its position for a longertime, until the previous flux is outdone. This remaining fluxis a big problem at the first closing during the first cycle. Thefirst pulse kills off the flux created by the permanent magneton the other side of the permanent magnet and strengthensit on the other. It is only after the second pulse, that thefluxes on both sides have been strengthened once. This iswhy the second switch is the slowest. During the next cycles,the response times continue to be higher than on the firstswitch, but they reach a steady level after just a few switches.

In case of this actuator, the amplitude of the magnetic fielddensity does not reach a steady level in the middle parts of themagnetic circuit until about 15–20 cycles with the maximumdriving frequency (constant 100 Hz). They will bemagnetised only after about 0.15 s from the start of cycling.However, this does not seem to effect the response time ofthe valve.

IET Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243doi: 10.1049/iet-smt:20080124

IETdoi

www.ietdl.org

3.3 Used models

When a hydraulic part is connected to the actuator andmovement is calculated, the exact modelling task is acoupled problem between time-depending electromagnetic,fluid mechanical and mechanical models. However, themodeling task is simplified. Static fluid mechanicalmodelling results from [8] are used as parameters for theelectromagnetic solvers.

There are two type of dynamic electromagetic models usedin predicting the valve behaviour: (1) The transient modelsand (2) the linear motion models.

The transient model covers the effects of screeningcurrents induced by changing current in the coils, but itdoes not model the movement (back-EMF effect).However, the response times are approximated usingmodels with two separate positions and interpolationbetween the same two transient magnetic models. Thismanner predicts well the starting moment of the armature’smotion and fairly the response times. This model is notused to model cycling since earlier pulses would be difficultto take into account. The benefit in using these models isthe relatively short computing time. For these reasons,transient models are used to predict response time andpower consumption for a wide range of current densities.

The linear motion model takes all the needed dynamicalelectromagnetic phenomena into account, including themovement (back-EMF effect), magnetic diffusion andearlier pulses in cycling. The drawback in linear motionmodels is the long computing time. In this paper, it is usedfor only two selected current densities, the selection beingbased on the transient models.

In both modelling cases, the used software OPERA-2dsolves the vector diffusion equation

� curl1

mcurl Að Þ

� �� s

@A

@t¼ curl H c

� �þ s grad Vð Þ (1)

with the magnetic vector potential A as the unknownvariable. The problem is given in a 2D rotationallysymmetric case as following.

Domain V: For all x [ V

�@

@z

1

m

@Af

@z

� ��@

@r

1

mr

@ rAf

� �@r

0@

1A� s

@Af

@t

¼@Hc,r

@z�@Hc,z

@rþ s

1

r

@V

@f(2)

is satisfied and at the boundary

Af ¼ 0 (3)

Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243: 10.1049/iet-smt:20080124

The task is to find Af, such that for all x [ V (2) and (3) aresatisfied. The current density in (2) has been split into thedriving source (voltage driven model), s(1=r)(@V =@f), andthe induced currents, �s(@Af=@t). Hc is the coercive forceof the permanent magnet. The equations are given in theLagrange coordinate system.

3.4 Modelling cases

Before the dynamic modelling the permanent magnet ischanged to Neorem 491a type of Nd-Fe-B compound,since this magnet has a higher coercive force at 808C.Unfortunately, it also has a lower remanence flux densityvalue leading to a problem of too small opening force. Thegenerated force is still higher than the needed openingforce (48 N at 210 bar), but the difference is so small thatit leads to a grown response time.

A too low force of the actuator can be increased by (1)increasing the size of the whole actuator construction or (2)increasing current density in the limits of avoiding too highheat generation rates and demagnetisation of thepermanent magnet. Since the aim is a small and fastconstruction the second option is chosen.

The further increase of current density was originallyavoided in the static models, since the models predicted theused magnetic circuit material (armco iron) to be very closeto saturation already in the previously used 40 A/mm2.When the current density was set to 45 A/mm2 in thenew actuator, the rms power consumption actually droppeddue to faster response and shorter needed pulse time. Theleast power consumpting current density with itscorresponding response time was then searched withtransient models.

After the choice of the optimal current density, two currentdensities are chosen for the modelling in cyclic action withthe linear motion model. As the next modelling case, alsowith the linear motion model, one of the two chosencurrent densities is used to drive the valve in rapid cycles.

While the previous models are calculated with a pressuredifference of 210 bars, the last modelling case is donewithout a pressure difference (but valve still assumed filledwith oil). This is modelled with the linear motion model at308C temperature, with constant 100 Hz driving frequencyand 4 ms voltage pulses. 24 V driving voltage is assumed.After that, a prototype is built and measured.

In most modelling cases, when a hydraulic input pressureof 210 bar is used, the output pressure is assumed to beclose to 1 bar when the valve is closed and100 bar (Dp ¼ 100 bar) when the valve is opened. Thereare also measurements with constant Dp ¼ 0 bar, asexplained above.

239

& The Institution of Engineering and Technology 2009

24

&

www.ietdl.org

4 Results4.1 Modelling results

First, transient models are used to estimate valve behaviour.Fig. 7 represents the rms power consumption with differentmaximum current density rates at constant 25 Hz. In thispaper, the frequency refers always to a complete cycle, thatis, one period includes one opening and one closing. Thelengths of voltage pulses are set to match the responsetimes given in Fig. 8. From these two pictures it can beseen, that the rms power has a minimum somewhere closeto 45 A/mm2 whereas the response time curves formdecreasing slopes.

The decreasing response time is explained with theincreasing force. The needed minimum opening force isgenerated faster with higher current densities. For eachcurrent density, a different coil structure using 24 V asinput voltage is calculated.

The existence of a minimum in the power consumingcurve is more complicated. First, a valve that generates justenough force to open the flow path, consumes a lot ofenergy during one opening, since it takes time to reach theforce needed and the opening pulse becomes long. In thecase of our valve this would happen at about 35 A/mm2

(Fig. 9). When the current density is increased to 40 and45 A/mm2, the needed opening energy drops because theforce needed is generated faster and the voltage pulse canbe made shorter. This explains the drop in the beginningof the rms power curve (Fig. 7).

At 50 and 55 A/mm2 the pulses required are still shorterthan previously, but the growth of transient maximum

Figure 7 Modelled rms power consuming at constant 25 Hz

Transient and linear motion models. The dash line shows the limitof tolerable heat production. At this point the permanent magnetstays under 808C in most operating cases. That is, the oiltemperature is maximum 708C and the surrounding airmaximum 508C

0The Institution of Engineering and Technology 2009

power during pulses has a bigger effect on the systembehaviour and the needed opening energy starts to increase.Also at high current densities, the saturation of themagnetic circuit causes the maximum generated force toincrease slower (Fig. 9). This leads to a growing rms powerconsuming rate, since the increasing of current does nothave the same benefit as before.

The current densities of 55 and 75 A/mm2 are chosen forfurther modelling with linear motion models. A 55 A/mm2

is chosen, since it is close to the valley in powerconsuming, but with smaller response time than 50 A/mm2 current density. A 75 A/mm2 is chosen, since it has

Figure 8 Modelled response times of opening and closingthe valve at 25 Hz

Modelling is done with transient models

Figure 9 Modelled maximum forces at each current densityand linearly (actually affinely) growing force calculatedfrom a line passing through points (0 A/mm2, 26.23 N)and (35 A/mm2, 48 N)

IET Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243doi: 10.1049/iet-smt:20080124

IETdoi:

www.ietdl.org

the smallest response times of the modeled cases, where thepower consumed by the coils can still be emitted (Figs. 7and 8).

The calculated response times of a cycled constant 25 Hzdrive with current densities of 55 and 75 A/mm2 arepresented in Fig. 10. Calculations show that the biggestchange in the response time is at the first closing after thefirst opening. This is because the actuator first needs to winthe magnetisation due to the first pulse, as explained above.The time constant of the coils, given as t ¼ L=R, istransiently increased for the same reason. The inductanceL, given as

L ¼dF

di(4)

is bigger when the change of flux (F) is big.

When the actuator is applied in practice, the used drivingpulses need to be longer than the computational or measuredminimums. This is because pressure peaks and for example,vibration can alter the environment transiently in thehydraulic system, and a certain safety margin is needed.One possibility would be to measure the armatures positionand set the pulse time accordingly.

The used pulse times in linear motion constant 25 Hzmodels were 3.5 ms for opening and 2.5 ms for closing at55 A/mm2. At 75 A/mm2 the used values are 3 and 2 ms,respectively. The rms power consuming is calculated fromthe modeled current. The results are shown in Fig. 7 withsmall ‘x’ marks. Even thought longer pulses are used, therms powers are smaller than in transient models. This isbecause of bigger time constants due to earlier pulses andtaking into account the back-EMF effect. Thisphenomenon lower the currents. In linear motion

Figure 10 Comparison of response times (in ms) betweentwo linear motion models at constant 25 Hz

Used current densities are 55 and 75 A/mm2. Switches arecounted One cycle is two switches

Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–24310.1049/iet-smt:20080124

models, the power consumption at 55 A/mm2 is still muchlower than at 75 A/mm2, and it is therefore our choice forthe optimal current density.

After choosing 55 A/mm2 as the optimum current densitythe behaviour of the valve in rapid constant 100 Hz cycling ismodelled. It is found out (Fig. 11), that the starting positionof the valve has a big effect on the response time during thefirst two or three cycles. The magnetisation of the middleparts in the magnetic circuit in later cycles (Fig. 12) hasnegligible effect on the response times. Two examples ofthe magnetisation in the magnetic circuit are plotted inFig. 13. The pulse times in constant 100 Hz cycling were4 ms for opening and 3 ms for closing.

Figure 11 Modelled response times with linear motionmodels using a current density of 55 A/mm2 and afrequency of constant 100 Hz

Two cases are modelled: initial state open and initial state closed.Switches are counted. One cycle is two switches. Response timesstay in the reached steady level after the first ten switches

Figure 12 Development of magnetisation in the magneticcircuit calculated with linear motion models

Valve is driven at constant 100 Hz, meaning 20 full cycles in 0.2 s.The field values are plotted at different time steps. In the end ofthe last cycle (t ¼ 0.2 s) the value for Bz in the symmetry axis is0.76 T, whereas initially it is 0.28 T. The plotting points arelocated on a radially oriented line in the magnetic circuit abovethe permanent magnet. The line is shown in Fig. 13

241

& The Institution of Engineering and Technology 2009

24

&

www.ietdl.org

4.2 Measured results and comparison

In the prototype (Fig. 14), the winding of the coils is setusing 24 V as the input voltage. The coils have a totalnumber of 165 turns. At the maximum temperature of thecoils (1208C), the current is 14.6 A. This corresponds to acurrent density of 55 A/mm2. The size of the magnet is asfollows: outer diameter 13.8 mm, inner diameter 5 mm,height 6 mm. The dimensions of the actuator are describedin Fig. 4 and [7].

The forces of the actuator were measured with a digitalscale. The behaviour in a hydraulic system and responsetimes in general were measured with a dSPACE DS1102controller with sample rate 10 kHz. The hydraulic testcircuit is presented in [7].

Whereas the models were done in a high temperature case(808C for the permanent magnet and 1208C in the coils)

Figure 14 The prototype valve

Figure 13 Development of magnetisation in the magneticcircuit

Valve is driven at constant 100 Hz, meaning 20 full cycles in 0.2 s.The field values are plotted at different time steps, the left one isduring fourth cycle, 0.5 ms from the start of the closing pulse. Theright one is the same during the 20th cycle. The correspondingtimes are: 0.0355 and 0.1955 ms. The plotting points for Fig. 12are shown as dashed lines above the permanent magnet in themagnetic circuit

2The Institution of Engineering and Technology 2009

same temperatures were not reached in the measurements.Measured valve behaviour at a temperature of 408C in thewhole actuator is presented in Fig. 15. The used pressuredifference was 210 bar and driving frequency constant75 Hz. At a higher frequency of constant 100 Hz, thepressure oscillations make it hard to find out whether or

Figure 15 Measured valve behaviour at a temperature of408C in the coils

Initial pressure difference over the valve is 210 bar, drivingfrequency is constant 75 Hz, and pulse times are 6 ms each.Driving voltage is 15 V in this case (target voltage is plotted,true driving voltage has an overshoot). The pressure data showsstrong oscillation

Figure 16 Modelled and measured valve behaviour at atemperature of 308C in the whole actuator

Pressure difference over the valve is 0 bars all the time, drivingfrequency is constant 100 Hz, and pulse times are 4 ms each.There is an overshoot of about 13% at the voltage control ofthe measurements

IET Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243doi: 10.1049/iet-smt:20080124

IEdo

www.ietdl.org

not the valve opens at each cycle. At high flow rates, there wasalso another anomaly in the hydraulic part. It was found outthat the armature had a lower maximum flow rate than on aprevious measure with the same part [7].

Since there were problems in measuring the correspondingdata at high pressure rates and temperatures, anothermeasurement-model correspondence was investigated. Thepressure difference was set to 0 bars and temperature to308C in the whole actuator. The results of modelled andmeasured current and voltage are presented in Fig. 16. Theactual driving voltage has a 13% overshoot. It can be seen,that inductance variations are well modelled. The back-EMF effect is not clearly visible in the measured data. Thisis probably due to a strong viscous damping close to theend of the movement. This was not taken into account inthe models.

5 Conclusions and discussionIt was found that the highest possible switching frequency inoil flow control is significantly increased from the solenoiduse. Of course, additional electronics can change thesituation.

The measured and modelled properties show strongcorrespondence although not a complete match. Taking into account the known differences between these two, it isconcluded that the measurements support the models.

From the models, it is concluded that the response timesdepend on the operating frequency and that a currentdensity with the smallest rms power consumption can befound. It should be noted, that in a random drive with amaximum frequency of 100 Hz and a mean frequency of25 Hz the rms power consuming properties would slightlydiffer from a corresponding constant 25 Hz drive due todifferent pulse times. This has been nor modeled neithermeasured. The modes are presented in Fig. 1.

The mechanical strength of some parts in the valve may beunreliable. The permanent magnet covered with nickeltolerates the oil well, but is exposed to collisions when thearmature moves. This may result in demagnetisation, but isunlikely because of hydraulic damping. Also, thepermanent magnet may gather metal parts from the oil, butthis seems unlikely since the actual flow path is far awayfrom the permanent magnet.

The hydraulic part seemed to have some defect. This wasprobably due to some wearing and tearing resulting fromheavy dry cycling. The problems in the hydraulic part –the wear and tear and pressure oscillation could probably besolved, but they do not affect the study focusing on theactuator.

T Sci. Meas. Technol., 2009, Vol. 3, Iss. 3, pp. 235–243i: 10.1049/iet-smt:20080124

6 AcknowledgmentThe research was supported by the the Finnish FundingAgency for Technology and Innovation (Grant no. 40334/04). The authors thank Ville Ahola (IHA/TUT) for thehydraulic measurements.

7 References

[1] LINJAMA M., LAAMANEN A., VILENIUS M.: ‘Is it time for digitalhydraulics?’. Proc. 8th Scandinavian Int. Conf. Fluid Power,SICFP’03, Tampere, Finland, 7–9 May 2003, pp. 347–366

[2] LINJAMA M., VILENIUS M.: ‘Design and implementation ofenergy saving digital hydraulic control system’. Proc. 10thScandinavian Int. Conf. Fluid Power, SICFP’07, Tampere,Finland, 21–23 May 2007, vol. 2, pp. 341–360

[3] LINJAMA M., VILENIUS M.: ‘Digital hydraulics – towardsperfect valve technology’. Proc. 10th Scandinavian Int.Conf. Fluid Power, SICFP’07, Tampere, Finland, 21 – 23May 2007, vol. 1, pp. 181–196

[4] KAJIMA T., KAWAMURA Y.: ‘Development of a high-speedsolenoid valve: Investigation of solenoids’, IEEE Trans. Ind.Electron., 1995, 42, (1), pp. 1387–1390

[5] KALLENBACH E., KUBE H., ZOEPPIG V., FEINDT K., HERMANN R., BEYER

F.: ‘New polarized electromagnetic actuators as integratedmechatronic components – design and application’,Mechatronics, 1999, 9, (7), pp. 769–784

[6] JOHNSON B., MASSEY S., STURMAN O.: ‘Sturman digitallatching valve’, in PALMBERG J.-O. (ED.): Proc. 7thScandinavian Int. Conf. Fluid Power, SICFP’01, LinkpingUniversity, Linkping, Sweden, 2001, vol. 3, pp. 299–314

[7] UUSITALO J.-P., LAUTTAMUS T., LINJAMA M., SODERLUND L., VILENIUS

M., KETTUNEN L.: ‘Miniaturized bistable seat valve’. Proc. 10thScandinavian Int. Conf. Fluid Power, SICFP’07, Tampere,Finland, 21–23 May 2007, vol. 3, pp. 379–391

[8] LAUTTAMUS T., LINJAMA M., NURMIA M., VILENIUS M.: ‘A novelseat valve with reduced axial forces’. Proc. BathWorkshop on Power Transmission and Motion Control,PTMC’06, Bath, UK, 13–15 September 2006, pp. 426–438

[9] MIKKOLA J., AHOLA V., LAUTTAMUS T., LUOMARANTA M., LINJAMA M.,VILENIUS M.: ‘Improving characteristics of on/off solenoidvalves’. Proc. 10th Scandinavian Int. Conf. Fluid Power,SICFP’07, Tampere, Finland, 21 – 23 May 2007, vol. 3,pp. 343–354

243

& The Institution of Engineering and Technology 2009