Escolar Documentos
Profissional Documentos
Cultura Documentos
net (TM)
Pgina 1 de 13
The Scaling of
Micromechanical Devices
by
William Trimmer
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 2 de 13
shows four cases for the force law. The top force law scales as S, next
scales as S squared or S2 (I hope this appears as S superscript 2 on your
screen, one never knows with the web), the next as S3, and the bottom as
S4. The scaling of the time required to move an object using these forces
is given as
The top element in equation 2 is S1.5. This is how the time scales when the
force scales as S1. The second element shows that t scales as S1 when the
force scales as S2. The third and forth element show how the time scales
when the force scales as S3 and S4 respectively. This notation is used
consistently throughout this paper. A dash [] means that this case does
not apply.
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 3 de 13
This vertical bracket notation can be used for other scaling laws. For
example, if one had a desire the top element could refer to the case where
the force scales as S6. Or the top element could represent to the case where
the acceleration scales as S1, and the second element represent the case
where the acceleration scales as S2 , ... . These vertical brackets can be
defined for the convenience of the problem at hand. All that is needed is
the initial definition of what each element represents. (Equation 1 is this
definition in our present case.)
Magnetic Forces
This Section examines the scaling of magnetic forces caused by the
interactions of electrical currents. Three cases are examined: A) constant
temperature rise from the center to the exterior of the coil windings, B)
constant heat flow per unit surface area of the coil windings, and C)
constant electrical current density in the coil windings. Assumption A)
leads to forces that scale as S2, assumption B) leads to forces that scale as
S3, and assumption C) leads to forces that scale as S4. These three cases
are depicted in equation 2.5. The derivation of this force scaling requires a
bit of math and will not be given here. This derivation is given in the
appendix of Microrobots and Micromechanical Systems.
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 4 de 13
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 5 de 13
These current scaling are the result of the assumptions in Case A, Case B,
and Case C) and generate the forces:
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 6 de 13
Other forces
There are several other interesting forces. Biological forces from muscle
are proportional to the cross section of the muscle, and scale as S2
Pneumatic and hydraulic forces are caused by pressures (P) and also scale
as S2. Large forces can be generated in the micro domain using pressure
related forces. Surface tension has an absolutely delightful scaling of S1 ,
because it depends upon the length of the interface.
The unit cube
Below is a discussion of how the above force laws affect the acceleration,
transit time, power generation and power dissipation as one scales to
smaller domains. In going from here to there as quickly as possible with a
certain force, one wants to accelerate for half the distance, and then
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 7 de 13
where SF represents the scaling of the force F. Here only the time to
accelerate has been calculated, but an equal time is needed to decelerate,
and both these times scale in the same way. For the forces given in equation
(1), the accelerations and transit times can be expressed as
and
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 8 de 13
Even in the worst case, where F = S4 (the bottom element), the time
required to perform a task remains constant, t = S0 , when the system is
scaled down. Under more favorable force scaling, for example, the F = S2
scaling case, the time required decreases as t = S1 with the scale. A system
ten times smaller can perform an operation ten times faster. This is an
observation that we know intuitively: small things tend to be quick.
Inertial forces tend to become insignificant in the small domain, and in
many cases kinematics may replace dynamics. This will probably lead to
interesting new control strategies.
Power generated and dissipated
As the scale of a system is changed, one wants to know how the power
produced depends upon the force laws. For example, consider the unit cube
above, which is first accelerated and then decelerated. The power, P, or the
work done on the object per unit time is
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 9 de 13
When the force scales as S2 then the power per unit volume scales as S-1 .
For example, when the scale decreases by a factor of ten, the power that
can be generated per unit volume increases by a factor of ten. For force
laws with a higher power than S2 , the power generated per volume
degrades as the scale size decreases. There are several attractive force laws
that behave as S2, and one should try to use these forces when designing
small systems. (Please remember, these force laws depend upon general
assumptions, there is always room to be clever.)
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 10 de 13
For the magnetic case, one may be concerned about the power dissipated by
the resistive loss of the wires. The power due to this resistive loss, PR, is
where A is the cross section of the wire, (rho) is the resistivity of the wire,
and L is the length of the wire. This gives
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 11 de 13
For the magnetic case A) where force scales as S2, the power that must be
dissipated per unit volume scales as S-2 , or, when the scale is decreased by
a factor of ten, a hundred times as much power must be dissipated within a
set volume. This magnetic case is bad if one is concerned about power
density or the amount of cooling needed. If power dissipation or cooling
are not a critical concern, then this scaling case produces more substantial
forces. In the future, superconductors may give us stronger micro
electromagnets.
Summary of the scaling results
The force has been found to scale in one of four different ways: [ S1 ] ,
[ S2 ] , [ S3 ] , and [ S4 ] . If the scale size is decreased by a factor of ten,
the forces for these different laws decrease by ten, one hundred, one
thousand, and ten thousand respectively. In most cases, one wants to work
with force laws that behave as [ S1 ] or [ S2 ] . The different cases that
lead to these force laws, the accelerations, the transit times, and the power
generated per unit volume are given below.
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 12 de 13
and
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004
Pgina 13 de 13
volume also increases for these two laws. The surface tension scales
advantageously, [ S1 ] , however, it is not clear how to use this force in
most applications. Biological forces also scale well, [ S2 ] but may be
difficult to implement. Electrostatic and pressure related forces appear to be
quite useful forces in the small domain.
http://home.earthlink.net/~trimmerw/mems/Scale.html
12/9/2004