a microgravity isolation mount

Acta Astronautica Vol. 15, No. 6/7, pp. 441-448, 1987 0094-5767/87 $3.00 + 0.00 Printed in Great Britain. All rights reserved © 1987 Pergamon Journals Ltd

A MICROGRAVITY ISOLATION MOUNTt

D. I. JONES, A. R. OWENS and R. G. OWEN School of Electronic Engineering Science, University College of North Wales, Dean Street, Bangor,

Gwynedd, LL57 IUT, U.K.

(Received 6 January 1987)

Al~traet--In this paper we discuss the technology required for isolating microgravity sensitive payloads from spacecraft vibrational disturbances. Specifications are given for the expected level of input disturbance and the required level of acceleration which can be transmitted across the mount. An outline is given of an actively controlled platform intended to satisfy this specification with details of the actuators, sensors, control law and the method of transmission of electrical power.

1. INTRODUCTION

It has long been proposed that the microgravity environment of Earth orbit has advantages for experimental work in the fields of fluid science, organic and inorganic materials preparation and the life sciences. Wilhelm[l] reviews some of the preliminary work which has already been performed in materials processing. If space manufacture is to achieve com- mercial viability then further research is required now to establish suitable processing techniques which take full advantage of the unique on-orbit environment. In order to encourage such research it is important to provide, at this stage, a generalised experimental facility which is available on frequently-flown mis- sions and is adaptable to the needs of a wide ranging scientific community. However it has been established[2] that many of the proposed processing techniques are critically dependent upon achieving lower levels of microacceleration than exist in current spacecraft. The experimental facility must therefore take the form of a platform which is effectively isolated from the motion of the parent spacecraft.

The factors responsible for disturbing the microgravity environment have been identified[3], and can be categorised in two ways:

(1) external disturbances affecting the spacecraft as a whole, such as aerodynamic drag and orbital dynamics;

(2) internal disturbances such as rotational move- ments of the spacecraft for attitude control, the effects of crew motion and on-board equipment causing vibration (motors, pumps etc.).

One concept that has been proposed[3] for achieving very low microgravity levels is an unmanned free-flying platform inserted into Earth orbit at a suitably high altitude. This might take the form of a "drag-free" concept whereby an active thruster system causes the platform to follow the position of a

tPaper IAF-86-270 presented at the 37th Congress of the International Astronautical Federation, Innsbruck, Austria, 4-11 October 1986.

small proof mass located freely at its centre of gravity. Very low microgravity levels may be achieved but even this concept suffers from the propagation of vibration, caused by service equipment located on the platform, through the structure to the experiment. The difficulty of access, if human intervention is required, is also a drawback.

On the other hand, the objective of the work described here[4] is to determine what levels of microacceleration are achievable on a platform which is enclosed within a manned, possibly manoeuvering, spacecraft. The platform is separated from the parent spacecraft by a microgravity isolation mount (MGIM), giving good vibration isolation, but is controlled to maintain a long-term position adjacent to its supporting frame. Unlike the autonomous platform, ancillary equipment is now situated on the parent spacecraft, which reduces direct communi- cation of vibration but introduces the difficulty of connecting a power supply and cooling circuits to the platform.

In this paper we discuss what levels of acceleration are allowable on the platform and what levels are present on the supporting frame. A design proposal for the MGIM is then presented and its component parts discussed. Finally, we assess its potential performance and give experimental results from a preliminary rig which has been constructed to verify its operational capabilities.

2. REQUIRED ACCELERATION LEVELS

Studies of the microacceleration levels required for successful experimentation have shown that the low frequency band is critical. Curve (a) in Fig. 1 shows a sinusoidal specification of acceptable acceleration levels, based on the envelope of several curves given by Tiby and Langbein[5]. Their curves were derived from theoretical models which indicate that the allowed acceleration exhibits a constant limit at low frequencies and a square law dependency at high frequencies. Curve (b) is the design specification

441

442 D.I. JONES et al.

t Key

e "~ 10-1 EnveLope of BatteLLe _ / / output Level ~ / c /

-~ 1(~3~. - b EURECA / / c U S. Space Stotion / / i b | usable specif ic

Io-5L

l d 9 F " 1 I I I 10-5 10-3 10-1 101 10 3

Frequency (Hz)

Fig. l. Relationship between sinusoidal specifications.

for EURECA[6] while curve (c) is the proposed specification for the U.S. Space Station. Taking into account each of these curves, we formulated Fig. 2a as an appropriate specification for our work.

For frequencies below 1.5 Hz a constant level of 10-Sg is specified as the minimum to be achieved while between 1.5 and 15Hz an f 2 variation, compatible with Fig. la, is assumed. Above 15 Hz a 10-3g upper limit is imposed in consideration of practical constraints such as damage to delicate in- strumentation. A supplementary (dotted) curve in Fig. 2a continues t h e f 2 variation down to the 10 6g level and represents the ultimate objective of our work.

Our guideline sinusoidal characteristic for the translational vibration of the supporting frame is shown in Fig. 2(b). In practice this vibration characteristic will be produced by several different sources, such as spacecraft subsystems (pumps, steerable an- tennae etc.), crew motion and thrusters which will interact on the nonlinear dynamic components of the spacecraft structure. The result will be a random combination of impulsive and periodic signals with broadband "noise' as a base and probably containing spectral peaks related to the natural modes of the spacecraft structure. Since modelling of such a com- plex system is a considerable task in itself, and in view of the paucity of measured data (especially in the lower frequency range), Fig. 2b was adopted as an envelope which encompasses all these effects. An f 2

1°' I- b /

0.474

tX) x~ - 4 0 /

- 6 0 I

0.01 0.1 1.0 10

Frequency (Hz) Fig. 2. Sinusoidal specification.

I I 100 1000

variation is assumed below 3 Hz and a constant limit of 1 0 - ' g above 3Hz. In fact, comparison with known characteristics for other transportation systems[4] shows it to be a worst case since it exceeds considerably the expected on-orbit vibration environment. This is supported by Hamacher and Merbold[2] where the measured acceleration data presented for the SL-1 mission rarely reaches 10 2g except during orbit trim burns.

Combining Figs 2a and b gives the transmissibility function of Fig. 2c which shows that below 0.03 Hz it is permissible for the platform to follow the outer frame; this defines the break frequency required of the MGIM, Above this frequency at least a - 4 0 dB/decade roll-off is required, up to 1.5 Hz, to maintain the platform acceleration at or below 10 5g.

Power spectral density characteristics for random vibration input, corresponding to Fig. 2, have also been established[4].

3. S T R U C T U R A L D E S I G N

It is a requirement of the proposed MGIM that vibration isolation be effective in all six degrees of freedom, translational and rotational. From Fig. 2a, the peak vibrational amplitude of the supporting frame is 3.9 mm peak. Allowing some margin, our isolating elements must operate over a + 5 mm range in one axis while allowing free motion of similar magnitude in the other two axes. Furthermore, elements acting on the same axis must be well matched in elastic and damping characteristics as any mis- match will induce angular rotation in the platform. The match must be maintained despite changes in environmental conditions and despite ageing. Since the platform is to accommodate various payloads, the isolation elements must compensate for changes in mass and moment of inertia and offset of the centre of mass from the centre of geometry.

The above requirements indicate that an actively controlled element is preferable. The relative merits of passive and active isolators have been discussed by Havenhill and Kral[8] who report that the adapt- ability and superior performance of active isolators are decisive factors in their favour though at the expense of increased power requirement, complexity and weight.

Consequently, our preliminary design study is based on the structural concept shown in Fig. 3 where the payload is affixed to a central platform. The platform is to be actively controlled in six degrees of freedom so that it remains, without physical contact, at a central position within its supporting frame. It is envisaged that the unit will fit into a cube of about 1 m side and have a maximum mass of approximately 200 kg. Both platform and supporting frame should be as rigid as possible and the platform should be well-damped so that any high frequency modes will decay quickly. In conflict with this, low platform

Outer frame

Passive an t i - v ibrat ion mount

J

A c t u a t o r /

module sensor

. / Actuator

PayLoad

J

PLatform

A microgravity isolation mount 443

Fig. 3. Structure of microgravity isolation mount.

- - Locking mechanism

weight is desirable so that payload mass is maxi- raised. This can be achieved with a platform having a closed-cell honeycomb internal structure with a stainless steel surface skin. The central platform allows easy access to the payload and is adaptable to various sizes and shapes of experiment modules.

Modules containing actuators and sensors are situated at each corner, the actuators acting together to control translational motion and differentially to control rotational motion. With this configuration the actuators act directly on the platform and a modular construction facilitates assembly. A locking mechanism is provided which clamps the platform securely for periods of launch and manoeuvre.

4. ACTUATORS AND SENSORS

The actuators in Fig. 3 are effectively small linear motors; a detailed cross-section is shown in Fig. 4. Rare earth permanent magnets establish a flux density in the bore of the stator. A thin, planar armature inserted into the bore produces force on the platform in one direction while allowing free movement along the other two directions to the limits of the bore gap.

PLatform

F i e L d pate

Permanent / " magnets

Outer frame

j Stator Armature

Fig. 4. Cross section of actuator.

I Thin skein of conductors

--End conductors

The scale of forces required of the actuators is very small and can be estimated by assuming a MGIM break frequency of 0.03 Hz (see transmissibility function) and a 200 kg platform. Then:

fo = l/2n x//-K/M (1)

where K = isolation element stiffness and M = platform mass.

This gives K - 7.1 N/m and at the maximum platform excursion of 5 mm, the force is approximately 0.04 N. Since four actuators act on an axis, each must provide about 0.01 N force. We estimate that this may be produced by a samarium cobalt magnet of dimensions 30 x 22 x 10 ram, giving a gap flux density of 0.35 T, and a 200 conductor armature carrying about 5 mA. A more detailed analysis can be found by Jones, Owens and Owen[4] but it is evident that the power requirements are very low and the actuator does not appear to be a limiting factor in this application.

There are eight sensors sited to measure platform displacement relative to the outer frame. The displacement sensors operate as differential capacitance bridges detecting the movement of a central plate affixed to the platform, as shown in Fig. 5. To ensure true balance between the two ratio arms, and also to

Moving plate

--J ill 1

/ Bifitor transformer

Fig. 5. Circuit of capacitive position transducer.

444 D, I. JONES et al.

7ram clearance in aLL directions

\

Pri f ixed to f rame

(14 tu rns )

Fig. 6. Non-contact power transformer.

ensure the long-term stability of the ratio, a bifilar wound transformer is used in the bridge. Stray capac- itances from the sensing plates and connecting leads to ground, and between the primary and secondary windings of the transformer, are eliminated by guard techniques. Conventional phase sensitive detection yields a linear d.c. output which is independent of the dielectric constant of the gap and, by making the central plate much larger than the two sensor plates, is sensitive to motion in one axis only.

5. PLATFORM POWER AND COOLING

The connections between platform and supporting frame must perform three separate functions:

(i) Supply of electrical power to the platform; (ii) Transport of cooling fluid to and from the

platform; (iii) Transmission of control and data signals.

Any physical connection will form a compliant element between the frame and platform thus intro- ducing direct transmission of vibration. It is therefore of interest to investigate to what extent these functions can be performed without recourse to a direct umbilical link.

Electrical power transmission may be substantial for some experiments such as crystal growth from a melt in a furnace. We have assumed a load rating of 1 kW and investigated the use of a transformer with loose-coupled secondary to effect power transmission. Figure 6 shows that the primary winding is wound onto the core of the transformer in the usual manner, but the secondary winding has a 7 mm clearance in all directions between the core and former. A prototype of this transformer has been constructed, the primary winding being driven with a square wave derived from a 150V d.c. supply by a

MOS transistor bridge. The secondary is connected directly to a bridge rectifier and smoothing capacitor with resistive load. Power transfer of 1 kW has been successfully achieved and a typical regulation characteristic is shown in Fig. 7. Calculations indicate that residual forces due to movement of the secondary are very small and this is confirmed by other workers in this field[9].

Cooling the payload is the most difficult task. Applying the Stefan-Boltzmann law of radiation shows that for the 6 m 2 surface arfea of our unit, a surface temperature of 46°C results from dissipating 1 kW with an ambient temperature of 20°C. This implies that the payload must be cladded with radi- ating panels and good thermal coupling maintained between the furnace and panels. For greater power levels, forced liquid cooling is the only realistic method requiring flexible tubing between frame and platform. For the specification given previously, the limit of stiffness for this tubing is of the order of 2-3 N/m and careful dynamic characterisation would be required when designing such an umbilical.

1 5 0

100

50

"-~----x ~ x

T : 1 2 F s

,so: 5.3 9.

[ i i I J I 0 1 2 3 4 5 6

O u t p u t c u r r e n t ( A )

Fig. 7. Transformer regulation characteristic.


Retotive I

( , - , , : o T - j°mp"r 'er j [ ...... J " " - "

/

Fig. 8. Block diagram of system

UmbiUcot stiffness J

PLatform + [ structural [ dynamcs J

using position transducer only.

The transmission of data over 10-15 mm presents little problem and we are currently developing a non-contact optical link which can sustain high data rates.

6. CONTROL TECHNIQUES

Three basic control schemes were investigated for a simplified one degree of freedom model:

(i) A transducer for position control, with velocity damping being derived from the integrated output of an accelerometer.

(ii) An inner acceleration control loop, with velocity damping again being derived from the output of an accelerometer; position control is provided by a relatively stiff mechanical spring.

(iii) Control using a position transducer only with lead-lag compensation to provide damping.

It can be shown[4] that all three schemes can be modelled by the generic third order transfer function:

y s + ~ (2)

= s3 d- Cts2 -~- s .-.I.- 7

where the parameters ct and 7 are related to the chosen natural frequency, the system mass, umbilical stiffness and specified control loop gains. In principle, then, all three systems are capable of providing the transmissibility function of Fig. 2c; However, in practice, there are constraints on the values which ct and 7 are allowed to take.

If we attempt to implement technique (i) with a piezo-electric accelerometer we find that the roll-off in its response at low frequency causes a resonant peak in the overall system response and an un- acceptable sensitivity to changes of payload mass. Using a servo-accelerometer having a fl~tt response down to d.c. overcomes this problem but the long time constant of integration needed for deriving the velocity signal calls for low accelerometer drift and a stable, accurate integration method.

With technique (ii), the inclusion of a passive spring eliminates the need for a position transducer and a high gain in the inner accelerometer loop increases the effective mass of the system. A relatively stiff spring will dominate the response at low fre-

quencies thereby overcoming the disturbance effect of the umbilical. At high frequencies, the response is dominated by the accelerometer signal which regu- lates the platform against disturbances in this frequency range. This technique haS the ad~,antage-6F overcoming the umbilical effect but requires larger actuator forces and, in practice, would have limits on the accelerometer loop gain due to unmodelled high frequency dynamics.

Technique (iii) does not require an accelerometer and the specified frequency response can be realised in a straightforward manner provided the umbilical stiffness is low. The system block diagram is shown in Fig. 8 and can be shown to have the transfer function of eqn (2) with the following substitutions:

oJ~ = (K + C) /M

0~0) 0 ~ a

7~Oo = a K + bC

~: [K + (b/a)C] 7 = K + C

where:

09 o = system natural frequency

a,b = lead-lag time constants

M = platform + payload mass

K = umbilical stiffness

C = feedback loop gain.

Assuming K = 0, normalised frequency response curves for this transfer function are given in Fig. 9 for 7/7 = 30. If the lowest system mass is 50 kg (~ = 1.25) and the highest 200 kg (~t = 2.5), Fig. 9 shows that this technique can give acceptable responses over a 4:1 ratio of mass.

In view of the discussion of the previous section we may expect that a low stiffness umbilical is a feasible proposition and we therefore choose to concentrate our attention on technique (iii).

7. PERFORMANCE ASSESSMENT

Numerical simulation of the system of Fig. 8 was undertaken where the input vibration was a zero

446 D.I . JONES et al.

10 - , ~ - - 0.5

0 ° = 1 ' 2 5

- - 1 0

- 2 0

- 3 0

I I I 0.001 0,01 0,1

0 .03

F r e q u e n c y ( H z )

Fig. 9. Normalised system frequency response.

N I ©

t3 L O)

*o t) t~

E o

(3-

/ /

!

/ /

10 /

3

1

0 . 3

0.1

I /L I I l i I O.001m 0 .003 0.01 0.03 0.1 0.3 1.0 1.5


Fig. 11. Acceleration with 1N/m umbilical.

mean, normally distributed random amplitude corresponding to the specification of Fig. 2b. The simulation output was analysed to give spectral density estimates of the platform vibration amplitude and acceleration. The results for the linear system (i.e. [iii] above) are shown in Fig. 10 and conform exactly to the specification, as would be expected. The recorded r.m.s, value of platform displacement was 0.125 mm and the r.m.s, value of platform displacement was 2.84 #g, in good agreement with the required values.

The inclusion o f a 12 bit quantiser to represent analog to digital conversion and a 1% actuator nonlinearity were shown to be insignificant.

The effect on the platform acceleration of an umbilical modelled as a pure spring of stiffness 1 and 5 N/m is shown in Figs 11 and 12, respectively. In the first case, the effect is confined to a rise in acceleration level in the 0.03 to 0.04 Hz range and does not cause violation of the output level specification. In the case of Fig. 12 a sharp resonant peak occurs at about 0.05 Hz with violation of the specification in the range 0.02 to 0.1 Hz, which is not acceptable. We conclude that a maximum umbilical stiffness of 2-3 N/m is allowable with this technique.

8. EXPERIMENTAL VERIFICATION

In order to obtain experimental verification of the performance of our proposed design we have constructed a one degree of freedom test bed which is shown in Fig. 13. The rig consists of two parallel air tracks which support a platform weighing approximately 8.5 kg with a 5.5 mm range of movement. The platform floats freely on the air stream and carries two actuator magnets similar to those described under "Structural Design". The actuator coils are attached to the inner frame which can be vibrated along the axis of motion. A capacitive position sensor measures the distance between frame and platform.

The control law described in the previous section is implemented as a digital compensator designed for 0.5 Hz bandwidth and incorporating integral control. The higher bandwidth is necessary because of the reduced mass and the high levels of disturbance forces due to the simple air jet system in this preliminary test rig.

Figure 14 shows the sinusoidal frequency response of the platform to an imposed vibration of the supporting frame while Fig. 15 shows its response to

l r

o

E o,3

"6 o.~

O.O01 0 . ( ; ,03 0.01

/

,I Output revel /

/

j

/ I tl I I I 1 I

0 . 0 3 0.1 0 .3 1.0 t .5

Frequency (Hz)

Fig. 10. Platform acceleration--spectral density from simulation.

N 1" X.

gO

E o

0 -

i / /

/ /

/ 1 0 - /

3

1

0.3

0.1 /

l / I 1 I I L I 0.001 0.003 O.Ot 0 .03 0.1 0.3 1.0 1.,5


Fig. 12. Acceleration with 5N/m umbilical.


Actuator comprising magnet Inner frame

ReguLator and coil (attached to vibrator)

",~ ~ ' \ / Outer frame Preseurised '~.'~-~a \ ' / floating freely on

AcceLerometer

m m

\ e LL,ng / / " s c r e w s

Air jets

Fig. 13. Preliminary air-track test rig.

a step demand. Comparison with the expected frequency response (dotted line) shows that the response does not quite conform; the small oscillation present on the leading edge of the step response indicates the same conclusion. Imperfections in the test rig and a constraint on sampling frequency, imposed by the equipment currently used to implement the digital compensator, are responsible for the discrepancy. Nevertheless the performance of the test rig is sub- stantially to specification and has confirmed the validity of the actuator, position sensor and control loop designs.

9. C O N C L U S I O N S

To date our study has concentrated on formulating a design concept for the MGIM. In this paper we have discussed the required performance specification and outlined the mechanical structure and the design of functional components of an MGIM capable of achieving this. Our performance assessment, based

on "~ -20

- 4 0

~ \ -40 dB/dec \ / a t 0.SHz

\ \

\ \

O. 1 1.0

Frequency (Hz)

Fig. 14. Frequency response.

10

I [ I I 10 2O 3O 40

Time (s)

Fig. 15. Step response.

on a simplified 1 degree of freedom computer model, and a pessimistic estimate of the supporting frame vibration, suggest that platform microaccelerations of the order of 5 #g r.m.s, are achievable. In the more benign environment expected in practice it is prob- able that this figure could be reduced significantly.

The one degree of freedom test rig has largely confirmed the potential of our design. Imperfections in the air track limit the usefulness of this preliminary test rig and we are now engaged in constructing a much improved three degree of freedom (planar) version. This is intended to support a 100 kg platform and allow random vibration testing down to the required 0.03 Hz bandwidth. In conjunction we are preparing a six degree of freedom computer model of the MGIM in order to design multivariable control laws and assess the impact of offset centre of mass on system performance.

Acknowledgement--The work presented here was performed as part of ESTEC contract 1-1755/85.

R E F E R E N C E S

I. J. P. Wilhelm, Industrial research opportunities in space. IEEE EASCON '84, 17th Annual Electronics & Aerospace Conf. (1984).

A.A 15/6-'7--K

448 D.I . JONES et aL

2. H. Hamacher and U. Merbold, The microgravity environment of the science double rack during Spacelab- 1 (1987).

3. R. E. Olsen and J. Mockovciak, Operational factors affecting microgravity levels in orbit. J. Spacecraft 18, 141-144 (1981).

4. D. I. Jones, A. R. Owens and R. G. Owen, Microgravity Isolation Mount: Design Report. Report of a study performed under ESTEC contract no. 1755/85 by U.C.N.W., Bangor (1986).

5. C. Tiby and D. Langbein, Allowable g-levels for microgravity payloads. Report of a study performed under ESTEC contract no. 5504/83 by Battelle Institute, Frankfurt (1985).

6. D. Eilers, Microgravity conditions for orbiting plat- forms. Annual DGLR Conference, Germany (1983).

7. Teledyne Brown Engineering, Low acceleration charac- terization of space station environment. Report pre- pared for N.A.S.A., Marshall Space Flight Centre (1985).

8. D. D. Havenhill and K. D. Kral, Payload isolation using magnetic suspension. Paper AAS-85-013, Annual Rocky Mountain Guidance and Control Co~Tference, Colorado (1985)~

9. J. Kiedrowski, Noncontacting power transformer. NASA Space Station Power Management and Distri- bution Workshop, Texas (1984).

a microgravity isolation mount

Documents