P15001: Soft Ankle-Foot Orthotic
/public/

Detailed Design

Table of Contents

Phase Planning

Shared Phase Vision

Our vision for this phase is to finalize critical subsystems and test non-critical subsystems. Key related technical questions related to our subsystems include:
  1. Distance and heel strike sensor
    • How do we integrate distance and heel strike sensing?
    • How do we receive real time response based on heal strike and distance?
  2. Electrical control of air muscles
    • How to control the solenoid from the Arduino?
    • How to supply different voltage levels?
    • Specify the right solenoid
  3. Secure foot – Lower
    • How to apply torque?
    • What possible materials to use?
    • Does the brace remain comfortable when force is applied?
  4. Secure foot – Upper
    • How will the muscle be attached to the brace?
    • Does the brace remain comfortable when force is applied?
    • Are additional components needed for support?
  5. Lower and Upper base attachment
    • What possible materials to use? permanent elastic?
    • Is it abrasion safe?
    • Is it user friendly?
  6. Air muscle lift / operating pressure
    • What are the muscle dimensions?
  7. Air Tank for a full day
    • How much air does the muscle need?
    • Do we need a bigger tank?
  8. Low battery alert
    • Is a hardware approach better than a software approach?
    • Is an audio alert or visual alert better?
  9. Air tank sensor
    • Is using an air tank sensor feasible?
    • Can it be connected to Arduino?
  10. Air tank tubing
    • How will it feed to the air tank?
    • What is an appropriate dimension range?

Some additional plans for this phase for our group are as follows:

Prioritized Tasks



Return To Top

Week 15 Plan



Return To Top

Bill of Materials


Return To Top

System Design

Mechanical System Concept

Lower Attachment

During this phase, new attachment/lift methods were analyzed. Modifications were made to the ankle brace and a stage 2 test was completed. The modifications that were made addressed issues that were identified in the previous phase, phase 3.

Pugh Charts

The following charts were generated to analyze the proposed attachment and lift methods.



Return To Top

Stage 2 Test

Questions to be Answered by Test
a) How much force does it take to lift the foot, using the chosen attachment method?
b) Is there any apparent slippage of the brace when a force is applied?
c) Does the foot lift as soon as the force is applied? If not, how much force is needed before movement is observed?
d) Does the brace remain comfortable when a force is applied? Are there any specific ways that the force is applied that causes the brace to be uncomfortable or constrict blood flow?

Procedure

1. Have volunteer sit on table top with their right foot hanging freely
2. Place modified ankle brace on volunteer’s foot
3. Strap the volunteer’s leg to the table using two Velcro straps, to ensure that the leg remains still when the force is applied
4. Hook the ribbon loop onto the spring force gage
5. Using a spring force gage, a second volunteer will lift the volunteer’s foot
6. Have another volunteer read off the force measurements and the test volunteer record the measured forces.

Setup


Results


Comfort Survey


Conclusions

a) It takes approximately 8lbs of force to lift the foot
b) The brace does not slip when a force is applied
c) No movement occurs until at least 2lbs of force is applied
d) The brace remains comfortable during the lift. The volunteer indicated that no discomfort was experienced during the test. The feeling of the brace during the lift can be compared to that of wearing a thick sock.

Next Steps

1) Discuss ways to address the issue of a delayed lift (i.e. foot does not raise as soon as a force is applied)
2) Complete same test with enhanced attachment method



Return To Top

Upper Attachment

During this phase, the additional strap that was found to be needed in stage 1 of the process was tested to determine its functionality. The strap was added to our assembly, and the test completed in the subsystem design phase was iterated.

Stage 2 Test

Questions to be Answered by Test
a) Does the attachment point about 2 inches below the top of the brace still provide the support needed to counteract the force from the McKibbon muscle?
b) Does the new non-elastic, Velcro strap provide enough support to support the McKibbon muscle function?
c) Is there any apparent slippage of the brace when a force is applied?
d) Does the brace remain comfortable when a force is applied? Are there any specific ways that the force is applied that causes the brace to be uncomfortable or constrict blood flow?

Procedure
Very similar testing procedure completed as compared with Stage 1 of testing.

1. Have volunteer sit on table top with their right foot hanging freely
2. Place modified leg brace on volunteer’s foot
3. Strap the volunteer’s leg to the table using existing two Velcro straps, as well as new strap sewn to the top of the brace
4. Hook safety pin 2 inches below the top of the brace to simulate attachment point
5. Using a spring force gage, a second volunteer will pull down on the attachment point
6. Record force measurements when slippage occurs, as well as document the distance the brace slipped, if at all.

Setup


Results
Results of phase 2 of this test showed the no significant slippage occurred on the brace when about 7 pounds of force was applied. The attachment point broke at this point, so no additional force could be added, but after an anticipated initial slippage to have the brace settle on the leg, subsequent pulls did not significantly alter the position of the brace.

Conclusions

a) The attachment point will still need to be about 2 inches below the top of the brace
b) The new non elastic Velcro strap should be sufficient enough to counteract the force from the muscle
c) There will be an initial slippage, but the strap prevents any major slippage after the initial slip

The foot lift calculation model was compared to 7 foot lift tests (including the two conducted in this phase, DDR) in order gauge the accuracy of the model:

Next Steps

1) Iterate test again once prototype muscle attachment piece machined and added to brace (see design below). This will allow us to use an actual air muscle with the forces that we will see in our system, as well as closely simulate the function of our final prototype

Full Report with additional plots: P15001_Feasibility_Testing_Report_Upper_Muscle_Attachment_2

Return To Top

Muscle Optimization Stage II

Motivation - The purpose of this test was to design and construct an optimized muscle for the integrated system. This was achieved largely through iterative testing and by researching past muscle optimization studies.
Procedure and Results - These tests were conducted following the same general procedure and data processing methods as in Muscle Optimization Stage I. A series of 11 formal, distinct, tests were conducted during this phase consisting of at least 5 different muscles with various dimensions. The table below describes the deflection, dimensions, and estimated steps for each muscle:

The figure below depicts the raw results of all 11 tests color coded by muscle design.

In order to make the data more transparent, the following plot was created using only M4, M5, & M7 with the estimated number of steps shown in the legend:

Conclusions - The optimal muscle should have the smallest diameter possible while still allowing for >1 inch of deflection. This was most closely achieved by M7 (L0=6", Inflated-Dia=1/2", Silicon tubing). It demonstrated the potential to achieve 1 inch deflection at 55psi. The expected life of this optimized muscle is shown in the table below by varying the tank dimensions:

Full Report with additional plots: P15001_Muscle_Optimization_Stage_II



Return To Top

Muscle Optimization Stage II Addition (Single Weave Sleeve)

Motivation - This sub-test was added on later in the phase in order to test the performance of a single weave muscle sleeve as opposed to a triple weave sleeve as previously tested. The sleeve appeared to have a small diameter and a comparably large strain rate.
Procedure and Results - These tests were conducted following the same general procedure and data processing methods as in Muscle Optimization Stage I and II. Two muscles were tested (M8=latex and M9=silicon) under 2 weight loadings (7.51 and 10.1lb). The initial lengths were: M8=7.5in and M9=4.8in. These muscles performed best under high pressure and the charts below illustrate the estimated number steps and the recorded deflection for three pressures:

Conclusions - The single weave sleeving appears to hold potential benefits for the optimal muscle design. M8 performed exceptionally well but the tubing was latex (allergen) and the initial length exceeded our constraints. With thinner silicon tubing and proper length adjustment single weave sleeving appears to be an optimal design.

Next Steps -
1.Purchase thin silicon tubing.
2.Construct an optimized single weave sleeve muscle with an initial length of approximately 6in (subject to change).

Full workbook with data tables: P15001_Muscle_Optimization_II-Single_Weave_Sleeve

Return To Top

Number of Steps Research

Motivation - The purpose of this research was to find a reasonable and practical number of steps that our average user will take in a typical day.
Procedure and Results - Original findings were based on a New York Times article claiming that average American adults take around 5,000 steps per day. However since “Nearly three-quarters of all strokes occur in people over the age of 65” it seemed that the majority of our target market would fall under the category of 60 years old or more. Research from two different sources both agreed that 4,000 steps is the approximately average daily steps for people above the age of 60.

Conclusions - Based on these findings we conclude that it is safe to assume the majority of our end users will not average more than 4,000 steps/day. Furthermore, based on our observations from the Nazareth clinic we conclude that it is reasonable to reduce the target number of steps by 25% since most of our clients are not likely to be as physically active as their peers. Thus the target number of steps is 1,500 steps per leg.

Full Report with additional information and citations: P15001_Number_Of_Steps_Research

Return To Top

Tank Selection

Motivation - The purpose of this research was twofold:
a.) To research the cost of 4,500 Psi compressed air tanks and the corresponding estimated number of steps.
b.) To perform a sense check on the CAIR calculation model and the assumption that a larger tank will provide the addition volume as estimated.
Procedure and Results
The budget research was conducted by researching paintball tank prices online at at least two different suppliers. If available, the tank's weight, length, and diameter were recorded.
The tank life research was conducted by finding a third party paintball tech website with tables estimating the number of shots from a paintball gun with a give tank. The number of shots data from four tanks were selected and the corresponding number of steps was calculated using the CAIR Capacity model. As shown in the tables below, there is a consistent 20% difference between the estimated number of shots and the calculated number of steps:

Conclusions - Based on these findings we conclude that:
a.) A 90ci 4,500 Psi carbon fiber paintball tank is currently available for $160.95. Manufacturer specs: 2 inches long x 4 1/2 inches diameter - 1 3/8 inches bottle to ASA.
b.) The calculation method used in the CAIR capacity model is the same method used by a third party paintball tech. This is reassuring to know because it indicates that even if the muscle volume calculations are off, the estimated life of a particular muscle with a larger tank can still be determined using a tank exhaustion test.

Detailed Tank Selection Information: PDFa
Paintball Shots Information: PDFb

Return To Top

Fishing Line Deflection Test

Motivation - The purpose for this test is to find:
a) What grade or fishing wire should be used?
b) How much deflection is absorbed by the elasticity of fishing wire?
Test Setup
Two different grades of fishing line were tested in this experiment: 8lb transparent line and 60lb green line. The test was performed in the BAD Lab by suspending weights from an overhang as shown in the figure below:

Procedure and Results - The 8lb transparent fishing line broke under 7.5 and 5.6lbs of force before any accurate strain readings could be recorded. The green 60lb line was tested in four steps: 1. The wire was loaded incrementally and corresponding length readings were taken, 2. The wire was subjected to additional human body weight in an effort to test the upper limit but no failure was observed, 3. Measurements were taken as the loads were removed, 4. The data seemed to indicate a post-loading permanent deformation so incremental loads were added again to acquire additional data as shown in the figure below:

Conclusions
a) The green 60lb fishing wire far outperformed the transparent 8lb line which broke under less than 5.6lbs of weight. The 60lb line did not even break under 20lb of weights plus downward bodyweight force.
b) The deflection absorbed by the elasticity of the fishing line was found to vary with loading and fishing wire length. For approximately 10.5” of fishing line at 10.1lbs, the deflection was measured to be less than 1% strain or 0.09” It is recommended that the fishing wire be pre-stretched in order to reduce initial deformation if strain is a serious concern in later phases.

Full Report with additional plots and tables: P15001_Feasibility_Testing_Report_Fishing_Line_Deflection

Return To Top

Muscle Attachment Design

Purpose - An assembly to attach the muscle to the brace needed to be designed that fit the following criteria:
a) The assembly must be small and lightweight
b) The assembly must be safe without dangerous abrasive protrusions
c) The assembly must allow the muscle to be easily attached and removed so the AFO can be washed
d) The assembly must be feasible to design during MSD and not overly complex

Pugh Analysis
During the course of this phase, a brainstorming session was conducted and the following Pugh chart was made according to the selection criteria listed above:

Design
A preliminary design was completed for the muscle attachment piece. The design consists of 2 parts:

1. Integrated plug for the McKibbon muscle
2. Attachment base to be attached to brace

The first piece, the integrated plug, combines the plug on the McKibbon muscle with an attachment piece. The plug portion of the design serves as the air inlet to the muscle. It was designed to have the same functionality as existing plugs for McKibbon muscles, with similar dimensions. The plug was then combined with the attachment piece to easily slide into the base, with 2 screw holes to easily secure the muscle to the base.

The second piece, the attachment base, was designed to easily mate to the integrated plug. The plug easily slides into the cavity on the base, and the 2 screws holes should align to easily connect the piece together. Holes were drilled around the outside to easily sew the base to the brace.
Annotated photos of the 2 pieces are shown below:


Next Steps
This was a very preliminary design for the piece. Once the prototype is made, the design can be iterated and optimized.

1) Note positives and negatives from completed prototype
2) Complete stress and fatigue analysis on current design
2) Refine design with changes needed in mind


Return To Top

Electrical System Concept

System Architecture



Return To Top

Schematic

Critical Systems

Optional Systems



Return To Top

Top Level Pseudocode



Return To Top

Subsystem: New Heelstrike Sensor

Motivation - The motivation of this test was use the new distance sensor that has the higher pressure limit and make sure that the functionality of the system remained.
Procedure - The nice part about this testing was that in the previous testing, a header was created that allows for the sensor to be easily swapped out. Therefore, the test included swaping the new sensor, raising the threshold that the code was measuring heelstrike at, and to preform more tests on the system. There were no major code changes
Results - It is clear that the heelstike and distance sensors are working together. It is also clear that the new sensor works the same as the old sensor which was a desired outcome.
The output was also monitored and is seen below.

Conclusions - First and foremost, it is clear that this method of monitoring Heelstrike does work and that the system works together. The reason that we would be doing this test is to make sure that we could use a higher rated sensor. The reason that we would want to use a higher rated sensor is to make sure that the sensor performance does not degrade over time by operating outside of the tolerance limits.
Next Steps - The next step would be to test the system on different terrains.

Return To Top

Subsystem: Heelstrike and Distance Sensing (Part 1: Levelground)

Motivation - This test was to combine the two working subsystems that we had for gait sensing. These two subsystems were the heelstrike and the distance sensing.
Procedure - The first part of this testing was to find a way to attach the Heelstike sensor. This was done by creating a header that would allow the sensor itself to not be sodered and make it so they sensor was easy to remove. After this was done, a class function was written for the distance sensing and the SD card logging. This was used during the test to verify that the system was working.
Code - The rest of the test was to create a function that will test the value of the ADC input for the distance sensing and to create log, and close files for the SD card.

Results - It is clear that the heelstike and distance sensors are working together. This testing was also done with the lower threshold sensor, which explains why there is almost clipping at the bottom of the heelstrike curve.
The output was also monitored and is seen below.

Conclusions - First and foremost, it is clear that this method of monitoring Heelstrike does work and that they system works together. It is possible that as we do more testing and work with the heelstrike sensor more that we will look into filtering, but at this point it would seem that responsiveness of the system is more important than the small spikes we are seeing.
Next Steps - The next step would be to test the system on different terrains.

Return To Top

Subsystem: Heelstrike and Distance Sensing (Part 2: Stairs)

Motivation - With a working subsystem that combines heelstrike and distance sensing, the next portion of this test was to test the two sensors on stairs.
Procedure - The same procedure as above applies to testing the device on stairs.
Results - Although the distance and heelstrike sensors can work really when walking on levelground, this does not appear to be the case when walking on stairs. This is mostly due to the fact that all the weight was applied near the toes when walking on stairs. As a result, the heelstrike sensors could not pick up enough force to set off the system as shown on the two plots for upstairs and downstairs.

Conclusions - It is clear that walking on different terrains (especially stairs), causes a shift in pressure distribution on the foot. Most of the pressure is near the toes when walking on stairs, both ascending and descending.
Next Steps - If the force sensor could not pick up enough force when walking on stairs, a second sensor could be necessary for varying terrain. Or even an automatic timer which could sense when heelstrike could occur based on additional calibration tests.

Return To Top

Subsystem: Heelstrike and Distance Sensing (Part 3: Levelground with Toestrike)

Motivation - After the stair testing was done, it was clear that to accurately adapt to terrain, an additional sensor would be needed. This sensor would be placed on the ball of the foot.
Procedure - The sensor that was used was of the same type that was used by the heelstrike. This means that the same type of header could be created for the sensor and that it could be wired in the same way. It also means that the class function that was created could be used for toestrike as it was for heelstrike.
Results - It is clear that all three sensors now work. The difficult with these sensors is going to be placing them in the right location in order to ensure that there is consistent readings coming from them.

Conclusions - It is clear from when the toestrike sensor is triggering in the gait cycle that all three sensors are working in synchronicity.
Next Steps - The next test would be to see how the system works while on stairs

Return To Top

Subsystem: Heelstrike and Distance Sensing (Part 4: Stairs with Toestrike)

Motivation - Since it was verified that an addition sensor placed under the ball of the foot resulting in all three sensors working in synchronicity, it was then necessary to see how the system works on stairs.
Procedure - The same procedure as above applies to stairs.
Results - Since walking on stairs does not produce enough force on the heel and a lot of the force is shifted to below the toe, the force sensor by the toe was picking up most of the force for triggering. In addition, the distance and heelstrike sensor were working properly as was tested earlier.

Conclusions - It has been verified that walking on different terrain types causes a shift in pressure distribution. An additional sensor placed below the toe helps detect when toe is striking the terrain, especially when foot pressure if shifted on varying terrain .
Next Steps - The next test would be using all three sensors and the coding to predict and ultimately, detect what terrain a test subject is walking on.

Return To Top

Subsystem: Gait Articulation

Motivation - This test was to test the time it took to dorsiflex and plantiflex the foot in order to ensure the McKibben muscle and solenoid would actuate the foot at a natural rate of speed.
Procedure and Results - Multiple slow motion videos were taken of an individual. The videos were used to determine roughly how long it took an individual to plantifex their foot and to dorsiflex it. In addition the video was analyzed to determine which period of the gait cycle needed to be plantiflexed. A video was taken at the same frame rate of an unloaded McKibben muscle to determine how quickly. The test yielded that it took about one eighth of a second to both plantiflex and dorsiflex the foot. The video of the muscle showed that it was constricting and releasing slightly faster than one eighth of a second. The arduino code was modified to reflect this information. Then the functionality of the heal strike sensor, arduino, solenoid and muscle was tested and confirmed, as seen in the video.
Conclusions - The foot can be articulated at the same rate as an individual's natural motion and it can be articulated at the desired time of a person's gate cycle using the heal strike sensor and arduino.
Next Steps - Build and test a prototype to test on an individual to determine if plantiflexion and dorsiflexion feel natural and are occurring at the desired time during a gait cycle.
Video 1 -Gait Time Analysis
Video 2 -Muscle Articulation Time Analysis


Return To Top

Subsystem: Solenoid Hold State

Motivation - This test was to develop away to have an unpowered hold state so that when the orthotic has diminished its pressurized air supply it will become a passive orthotic, holding the foot in a dorsiflexed state.
Procedure and Results - The first step was to conduct research and contact a subject matter expert. Richard Bullers was contacted from SMC, a company that specialized in solenoids. Rich suggested a five port two position double acting solenoid to achieve an unpowered hold state. This type of solenoid does not have a returning spring but a separate solenoid to move the pin in order to open and close ports, using this solenoid with one of its ports sealed would yield an unpowered hold state. The arduino code was modified to reflect the need to output a signal to contract the muscle and a separate signal to vent the muscle . A five port two way solenoid valve was acquired from the bad lab and tested in the same configuration as in the gait articulation test. The functionality of the hold state was verified as seen in the video.
Conclusions - An unpowered hold state can be achieved.
Next Steps - We need to block the second port along with refining the air muscle design to ensure there are no leaks. We also need to find a way to exhaust the vented air.
Video - Muscle Articulation with Old State

Return To Top

Subsystem: Data Storage

Motivation - This test was performed to help determine how much much of memory capacity will be necessary for a micro SD card to hold through typical usage of the device.
Procedure and Results - A five minute walk was performed with the device and current SD card. After five minutes of walking, the SD card had used approximately 2.95 MB of data storage. A screenshot of the properties is shown below.

After Scaling the outcome to four hours which is how long this device would typically be used in one day (assuming data storage drops linearly), the SD card has used approximately 141.6 MB or .1416 GB.
Conclusions - A typical day's worth of this device does not use up much space on a micro SD card.
Next Steps - The most we'll need for memory capacity on a micro SD card is 4 GB.

Return To Top

Subsystem: System Housing

In order to hold the PCB board that contains the distance sensor and force sensor, PCB housing.

Next Steps
1) Research various assembly options for attaching the IR sensor
2) Look for various adhesives for waterproofing the system
3) Add the thru hole types and screws

Return To Top

System BOM

Return To Top

MSDII Test Plan



Return To Top

Build, Assembly, and Debug Plan

During this phase, progress was made on the Build, Assembly and Debug Plan. The additions on the mechanical side consist of Upper Muscle Attachment Base and Integrated Plug build plans as well as preliminary assembly and integration instructions. We plan on updating this document much more in the next phase.
Build, Assembly, and Debug Plan

Return To Top

Risk Assessment



Return To Top

Resources

Past Projects


Return To Top

Detailed Design Review

The Previous Action Items taken during the last review have been updated.

The following Agenda was used during the technical review on November 18.

The following Notes and Action Items were taken during the phase 4 Detailed Design Review.


Return To Top

Home Planning & Execution Problem Definition Systems Design Subsystems Design Detailed Design Gate Review
MSD I
Build Preparation Build & Test Integrate & Assemble System Validation System Verification Final Review
MSD II