Documente Academic
Documente Profesional
Documente Cultură
Abstract— A Bowden cable performance analysis, based on a Furthermore, the orthosis must be wearable and portable,
Design of Experiments (DoE) is presented for orthosis which means that it must be lightweight. Accordingly, the
applications. The need for analysing these cables is based on the transmission system must be as light as possible. It must also
construction of IKO (IKerlan`s Orthosis) with five actuated be flexible so that it can be adapted to the different arm
degrees of freedom (DoF) to help the human arm. The aim is for positions and it must be capable of transmitting sufficient
an individual to be capable of lifting weight without any great force.
effort using this exoskeleton, which should be portable and
readily dressed (and, therefore, lightweight). In order to transmit In order to transmit the power from the actuators to the
the power from the actuators to the joints, Bowden cables are joints Bowden cables are used. This transmission system has
used due to their flexibility and light weight. Transmission already been analysed in other projects, such as [3] and [4]. In
performance has been analysed in terms of load loss and cable principle, the Bowden cables are a good solution since they
deformation in order to estimate actuator requirements and have the desired specifications: they are lightweight, flexible
positioning accuracy. A test bench has been built to measure the and capable of transmitting great force. They are also
cable deformation and load loss occurring between the two ends economical and do not complicate the design of the
of the cable in different situations. This has been applied to exoskeleton to any great extent. Nevertheless, Bowden cable
cables with two different diameters and at different loads. The efficiency must be known, for a correct cable and actuator’s
variables for which the effect has been analysed are weight, the size selection. A DoE test has been used to analyse Bowden
angle formed by the cable at the point where it leaves the sheath, cable performance.
the cable flexion angle, cable length, cable flexion curvature
radius and cable type.
I. INTRODUCTION
The material used in this paper was partly supported by the Spanish
Ministry of Education and Science and European FEDER Fund (research
project DPI2006-14928-C02-01).
TABLE I. SPECIFICATIONS OF THE CABLES SELECTED FOR TESTS
Cable type Diameter Breaking load (provided by the
manufacturer)
1x19 1.2 mm 170 kg
1x19 1.6 mm 227 kg
Figure 2. Cable types. From right to left: 1x7, 1x19, 7x7, 7x19 7x7 1.2 mm 122 kg
7x7 1.6 mm 218 kg
A. Objective and selected variables • Cable type: to analyse the most interesting for use with
the exoskeleton.
The objective of these tests is to analyse which factors
affect the deformation and load loss, and to what extent. This It must be remembered that these variables have been
information is important for optimising the design of the applied to cables with different diameters.
exoskeleton and for pointing out the limitations of the
transmission using Bowden cables. B. Two-level factorial design
Total load transmitted by the cable, load to lift the external In order to analyse the effect of these variables, a two-level
charge plus load loss at the Bowden cable, should be known in factorial design has been used [5]. The values that have been
order to estimate cable and actuator size. given to the variables have been decided by taking into account
the specifications of the orthosis.
TABLE II. BLOCK 1
Weight 600 N 1200 N
Cable-sheath angle 0º 12º
Cable flexion angle 90º 180º
Cable length 350 mm 700 mm
Curvature radius 60 mm
Cable type 1x19 7x7 Figure 4. Loads on both ends of the cable during a cycle
A. Introduction
Each test has consisted of raising and lowering a weight
three times in a 45 mm travel. The unit deformation and load
loss factor (LLF) have been measured on each of the three
times, taking the value of each for been considered as a
replicate measure. Analysing the measures it can be conclude
that replicate measures have a very low dispersion, allowing
considering the mean value for DoE response. These values
have been calculated for raising and lowering the weight. This
information provides four responses, two for LLF and two for
unit deformation, which are the responses that have been
analysed.
Fig. 4 and Fig. 5 show a load and deformation
measurement cycle. The load graph shows a constant load,
which is given by the load cell on the weight side. The other
load changes depending on whether the weight is raised or
lowered. When the weight is raised, the load is greater than the
weight and, when it is lowered, it is less than the weight, due to
friction load between the cable and the sheath. In the
deformation graph, the first steep gradient corresponds to the
moment when the weight begins to be raised (stretching of the
Figure 3. Test bench
TABLE IV. RESPONSES ANALYSED
LLF_raising (F_raise-F_weight)/F_weight
LLF_lowering (F_weight-F_lower)/F_weight
Unit_Deformation_raising D_raise/L
Unit_Deformation_lowering D_lower/L
F: force measured by the load cells
D: deformation, difference between the measurements of the two LVDTs
L: length of cable that suffers the load changes when weight is raised or lowered
B. Results
Here the graphs that corresponding to the unit deformation
and load loss factor (LLF) when the weight is raised for blocks
1 and 2 are shown. The variables with significant effects are
the same variables as when the weight is lowered. Therefore,
results obtained when the weight is lowered are not given here.
Figure 9. Block 2. CD interaction for the LLF when the weight is raised
All the graphs shown here have been created using Minitab.
This program calculates the effects of the variables and the
interactions and draws the graphs using these values.
For the block 1, Fig. 6 and Fig. 7 show the representation of
the effects for the LLF and for the unit deformation when the
weight is raised, as shown on normal probability papers. A line
is drawn matching the least effects to help to identify the
variables with significant effect. It can be seen that these
variables are weight, cable flexion angle and cable type for
LLF and unit deformation; and cable-sheath angle for LLF.
Figure 6. Block 1. Effects for the LLF when the weight is raised
For the block 2, Fig. 8 shows the representation of the
effects for the LLF when the weight is raised, as shown on
normal probability papers, and Fig. 9 shows the CD (cable
flexion angle/curvature radius) interaction for that case; Fig. 10
and 11 show the effects for the unit deformation when the
weight is raised, as well as the CD (cable flexion
angle/curvature radius) interaction. In the Fig. 8 and the Fig. 10
a line is drawn matching the least effects to help to identify the
variables with significant effect. It can be seen that these
variables are weight, cable flexion angle, curvature radius,
cable type and cable flexion angle/curvature radius interaction
for LLF and unit deformation; and cable-sheath angle for LLF.
In the Fig. 9 and the Fig. 11 a line is drawn connecting the two
points of each cable flexion angle value to make easier seeing
the interaction effect.
Figure 7. Block 1. Effects for the unit deformation when the weight is raised
TABLE V. LLF AND UNIT DEFORMATION IN SOME OF THE TESTS
Cable Flexion Curvature LLF Unit
type angle radius deformation
1x19 90º 60 mm 0.245 [-] 0.00327 [-]
1x19 90º 180 mm 0.176 [-] 0.00305 [-]
1x19 180º 60 mm 0.555 [-] 0.00700 [-]
1x19 180º 180 mm 0.456 [-] 0.00523 [-]
7x7 90º 60 mm 0.273 [-] 0.00438 [-]
7x7 90º 180 mm 0.210 [-] 0.00356 [-]
7x7 180º 60 mm 0.603 [-] 0.00908 [-]
7x7 180º 180 mm 0.492 [-] 0.00651 [-]
Load: 1200N
Cable length: 700 mm
Diameter: 1.6 mm
Cable angle at the point where it leaves the sheath: 0º
F2 = F1 ⋅ exp ( µθ ) (1)