22_System Design Finalization

Some of the primary design decisions that had to be made for my Level 3 Electrical/Control Subassembly were:

(a) which motor actuation design would best fit our size requirement for the device and still supply enough force to extend a finger at low power consumption

(b) which microprocessor would best fit our size requirement while accommodating EMG analog inputs as well as motor control & feedback and user inputs

(c) how motor feedback would be implemented, whether with self-contained servo feedback or external feedback on regular motors

(A lot of system thought and design decisions were summarized in my System Design Report, so I will use some excerpts from my Individual System Design Report)

The primary requirements that had to be taken into account were: (a) device thickness of no more than 1 inch, (b) linear force of no less 500g supplied by each motor (roughly twice the force needed to extend a finger in our early tests), (c) feasible cost of no more than $30 per motor, (d) allows linear movement of no less than 30mm (10mm more than the change in distance needed to extend each finger in early tests), and (d) planar area required by actuation design needs to allow for six actuation components to be compactly housed in forearm assembly.

            The four actuation systems considered were (a) a gear bar, (b) a screw shafted motor, (c) a rotating arm, and (d) a spooling shaft hub.  Figure 1 below shows the outcome of weighing

Figure 1: Decision Matrix for Actuation System

the criteria for each system in a decision matrix.  Every system considered assumed that a string attached to the system would be the object of linear movement.  The major deciding points for each system were:

Gear Bar:  The gear bar would require at least 60mm of linear space to provide the 30mm of linear movement, but could be accomplished fairly simply by attaching a geared hub to a motor shaft.

Shaft Screw (Figure 2):  The smallest motors that could be found with threaded shafts were too large for the device; this meant that threaded shafts would have to be connected to the motor shaft with mounting hubs.  This would add another 10mm to the actuation assembly in addition to the length of the motor and 30mm of linear movement making this assembly about 70mm long for a 30mm length motor.

Figure 2: Shaft Screw Actuation Concept

Rotating Arm:  To achieve 30mm of linear motion with this assembly would need at least 15mm of space in addition to half the motor thickness.  Thinnest motor considered would make the total space needed 22mm leaving very little room for maintaining the 25.4 mm thickness of the device.

Spooling Hub:  This assembly was initially not considered because of potential problems with tangling if the wire wrapped around the spool were to fall off the spool when tension was not present in the line.  In simplest form, however, a string could be attached to the rotating shaft (Figure 3) with the end of the shaft up against a wall-piece to prevent the string from ever falling off the end when un-tensioned.

Figure 3: Spooling Actuation Design (spooling hub:top - spooling shaft:bottom)

Without that problem, this system offered the best space efficiency as the string could simply be attached to the motor shaft and rotated around, linearly pulling the string.  The only space requirement would depend on the motor size.

The Spooling Hub assembly was chosen as the method for linearly retracting a wire attached to the assembly (note: precision linear actuators were considered as well, but cost at least $70 per motor for the size needed).

A microcontroller would be needed for reading EMG signals as well as for controlling the motors, indication LED’s, and reading feedback data from the motors.  Three primary microcontrollers have been considered (Figure 4): the Arduino Micro, the Teensy 3.2 (Figure 5), and the Mojo v3 FPGA.  The Arduino and Teensy are roughly the same size at 18mm wide and about twice as long.  The FPGA was considerably larger, but was considered mostly for its ability to implement a filtering circuit on the FPGA chip.  Because of our unfamiliarity with FPGA’s and its size, it was decided that the benefit of consolidating our filter circuit did not offset the disadvantages enough to use the FPGA.  Between the Arduino and Teensy, the primary difference was that the Teensy was considerably more powerful with more GPIIO and analog I/O pins.  The lower power Arduino would conserve more battery life, but because processed EMG data is going to be analyzed constantly in addition to the other microcontroller functions, it was decided that the faster processing speed would outweigh extra power consumption.

Figure 4: Decision Matrix for Microcontrollers

Figure 5: Teensy 3.2 Microcontroller Tentatively Chosen for device

One of the primary factors determining the method for implementing motor feedback was the type of motor.  If a servo motor could be used, then motor feedback would be self contained in the motor and absolute positioning would be inherent in the motor’s operation (as long as its load limits aren’t exceeded). 

Other motors such as steppers and brushless DC motors would  require something like a potentiometer (similar to a servo’s) to keep track of the motor’s absolute position.  Encoders would be helpful in precision control of the motor movement, but they would only provide relative position feedback; this would work, but the default motor position (device is un-contracted) would need to be calibrated the first time the device was used, then from then on relative movement information would need to be tracked.  But this would easily become offset over time; so an absolute positioning feedback that can indicate a unique value for all motor rotational positions within some range would be needed.  Encoders may still be used, but if potentiometer feedback is precise enough, then they might not be needed.

Some of the reasoning for not choosing servo or stepper motors to supply force to the actuation setup are explained in the following excerpt from my Individual System Design report: (note: stepper motors were the initial choice for the project because of their significantly higher torque at low speeds, but power consumption and size did not offset this advantage. Low powered geared DC motors could accomplish better results by sacrificing higher rpm which aren’t needed past 60rpm.)

Figure 6: Decision Matrix for Motor Type

Figure 7: Gearing Up 90-Degree Servo Motors for more rotation

            In Figure 8 above, a conceptual layout of the electronics housing is pictured.  The motors pictured are the 298:1 geared DC motors mentioned earlier.  This layout is a flattened version of the actual housing.  Dividing the assembly along its horizontal axis, each half will be angled downwards to better conform to a user’s arm; the actual width of the assembly would probably be closer to 9-10cm.  The string is not shown in the diagram, but it would attach to the motor shafts and be routed through guides to the exit ports at the front of the assembly.