Design and Construction of a Controller for Powered Elbow and Terminal Device for an Amelic Child

CHARLES H. DANKMEYER, JR., CPO*, HARRY PHILLIPS, CP*,AND GERHARD SCHMEISSER, MD"Limb Prosthesis ClinicJohns Hopkins HospitalBaltimore, Maryland

A young boy (DOB 9/30/77) presented bilateral upper-limb amelia, left lower-limb complete phocomelia and right lower-limb proximal phocomelia. He is a healthy, intelligent child with good trunk balance. The patient has remained in his home environment while receiving specialized education outside the home.

Prosthetic Management

Personnel of the Limb Prosthesis Clinic at the Johns Hopkins Hospital have supervised the child's prosthetic management. The initial prescription was for a prosthesis suitable for a left shoulder-disarticulation amputee. The prosthesis was supplied in March 1980 when the patient was 21/z years old. It included a semi-rigid thoracic interface, a specially made passive flexion-abduction shoulder joint, a passive friction elbow (Hosmer E-50), a small friction wrist (Hosmer WE-300N) and a child's voluntary-opening terminal device (Hosmer 12-P). The primary functional goal of the initial prosthesis was to provide opposition for the right lower limb which, although anatomically deficient, was far more complete and functional than any of his other limbs. It was also felt that providing a prosthesis at this time would enable the patient and his family to become accustomed to prosthesis use, and enable the prosthesis team to perform a better evaluation of this child for future prosthesis management.

All attempts to operate the elbow and/or terminal device by body power were unproductive. Harnessing thoracic expansion for these purposes was considered but was rejected as an unjustified burden on his marginal vital capacity. Harnessing his right lower limb was attempted but was rejected due to observed overload of the limited functional potential of this limb. In addition, the harness and cable arrangements interfered with use of a seating device. It was also perceived that these arrangements would hamper the use of a "swivel walker" type of device which was under consideration as a mobility aid.

After the prosthesis was delivered, the patient accepted it well in spite of its active functional deficiencies. His tolerance of the thoracic interface was remarkably good. After this experience the prescription was modified by replacing the voluntary-opening terminal device with a Michigan type of powered terminal device (Hosmer #2107) controlled by a pull-switch activated by a conventional chin nudge control. The existing passive elbow was retained. The combination of a powered terminal device coupled with a passive elbow was also well tolerated. The patient used it from June 1980 until the following year. It was then decided that the patient had matured sufficiently to justify substitution of a powered elbow for the passive one. An Ontario Crippled Children's Centre (OCCC) model #62A electric elbow was selected because of its proven reliability and its availability in an appropriate size.

It is noteworthy that the two electrically powered components selected for use in the prosthesis were chosen for control by one or more motion-activated switches instead of by one or more myoelectric transducers. The Prosthesis Clinic team rejected myoelectric control for this patient because he had never had any experience in upper-limb brachiation. Therefore, locating and testing the several necessary myoelectric signal acquisition sites promised to be a lengthy, complex and possibly unrewarding process. In addition, the intimate interface fit required for successful electrode function was proscribed by the patient's limited vital capacity. Finally, there was the purely pragmatic problem of the lack of any commercially available powered terminal device/elbow unit combination which had been designed for myoelectric control and was of suitable size, weight, durability and reliability for a small child.

Control Objectives

The design objectives for the control system for the prosthesis were to enable a small amelic child to control the motions of both a powered elbow and a powered terminal device to the extent of providing opposition for the right lower limb, to enable transfer of light objects between the right foot and the terminal device and to enable transport of these light objects to the vicinity of his mouth.

Special Conditions

User options for operation of the elbow or terminal device should include separate, sequential or concurrent operation of these powered components. Because of a paucity of useful control sites, a single site with multiple input possibilities was required. The young age, small size, resourcefulness, intelligence and vigor of this child compelled prescription of a prosthesis having high reliability, high durability, small size, light weight, smooth external contour, appropriately adjustable sensitivity and safe simple straightforward operating characteristics.

An additional consideration was the control of two quite disparate powered units, i.e., elbow and terminal device, which have different batteries, charging systems and voltages.

The child demonstrated that by chin nudge motions he could press a paddle-shaped activating lever to operate the momentary contact switch that controls the Michigan terminal device. Observation also suggested that he could readily use additional chin motions to operate a multiaxial input control device to increase his control inputs adequately to operate the OCCC electric elbow as well.


The aforementioned biomechanical considerations suggested using a single chin-operated device for separate, sequential or concurrent control of both the powered elbow and terminal device. A joystick module (the controller) was mounted on the thoracic interface. The module was designed to permit displacement of the actuator rod along its long axis by vertical mandibular or forward flexion head/neck motions and concurrently, or separately, to permit angular displacement of the actuator rod about its lower end by horizontal mandibular of lateral rotational head/neck motions.

A rounded knob was placed on the upper end of the actuator rod to interface with the mandible or head for comfort and safety. The entire module was mounted to the thoracic interface with the knot; almost in a mid-sagittal plane, and angular displacements of the rod were limited to a magnitude which enabled the child to apply vertical forces to the rod regardless of its angle of tilt. A single control motion is sufficient to operate the terminal device since the Michigan hook, being based on a voluntary-opening principle, requires the action of a momentary-closing switch to open. Grasp occurs whenever the control motion is inactive. Grasp is powered by the tension of the rubber bands on the hook.

In contrast, two control motions are required to operate the OCCC elbow. One momentary contact switch controls flexion and another controls extension. By providing a means of mechanically returning the actuator rod to a neutral position, a single motion of the rod in the right direction could control one elbow motion, and a single motion in the left direction could control the other elbow motion. Because left motion of the actuator rod by the patient's head brought his eyes and mouth into closer proximity to the terminal device, it was decided to use this motion to control elbow flexion. Because of the location of the switch inferior to the chin, a control motion consisting of moving the head and chin to the left and inferior could be assigned more safely to command elbow flexion than elbow extension. A normal reflex head-withdrawal reaction would automatically arrest additional elbow flexion. By assigning a right-oriented head motion to command elbow extension, an alarm-withdrawal motion would not only arrest elbow flexion, but it would also cause the terminal device to move away from the face.


The first step in the construction of the control was the fabrication of the actuator rod and switch for the terminal device (Fig. 1 ). The momentary contact switch governing the terminal device is a commercially available subminiature push-button switch with its plastic button removed. Instead of the button, the switch is triggered by a plunger mounted in the end of the actuator rod (Fig. 1, Item 1 ). The plunger is threaded so the exact point of contact may be adjusted. Around the plunger a compression spring acts not only as a return spring on the actuator rod but, also, at full compression, as a limit on the downward travel of the plunger. The switch is thereby protected from excessive force. The actuator rod is retained in its housing by a #4-40 Allen-head screw and retaining spring (Fig. 1, Items 2 & 3 ). Finally, the overall length of the actuator rod can be easily adjusted by means of the stainless steel shaft that terminates in the aluminum knob (Fig. 1, Items 4 & 5 ). By loosening two set-screws the rod may be screwed in or out as necessary to lengthen or shorten it.

Fabrication of the actuator rod essentially completed construction of the terminal-device controller. Delrin, an engineering nylon, was selected as the housing material because of its great strength, durability, corrosion resistance, machinability and because it is a good insulator. The housing is made in three sections, a front (Fig. 2, Item 1 ), middle (Fig. 2, Item 2 ) and back (Fig. 2, Item 3 ); all enclosed between two stainless steel plates (Figs. 2 & 3 ). The actuator-rod assembly was mounted by means of two trunnions (Fig. 2, Item 4 ) between the front and back sections of the housing, allowing the rod to project through an opening in the middle section of the housing and to swing freely from side to side through a range limited by that opening. Blocks of rubber (Fig. 2, Item 5 ) were then inserted inside the middle housing along the sides of the actuator rod. The rubber blocks served to return the actuator rod to a central neutral position. The return force can be adjusted by altering the durometer of the rubber blocks.

In order to actuate the switches that controlled elbow function, the Allen-head retaining screw (Fig. 2, Item 6 ) on the actuator rod assembly was allowed to project through a hole in the front section of the housing to where it engaged an aluminum rocker arm (Fig. 2, Item 7 ) at point A (Fig. 2 ). The rocker arm is located in a recess in the front side of the housing and rotates around the same axis as the trunnions of the actuator-rod assembly. At each extreme of rotation the rocker arm contacted and actuated a microswitch (Fig. 2, Item 8 ) controlling either elbow flexion or elbow extension. The switches were buffered against excessive force by a combination of the limits placed on the motion of the rocker bar by the rocker bar's shape, by the limits placed on the motion of the actuator-rod assembly by the rubber blocks and by the middle section of the housing.

The last step in the construction of the control unit was to make the necessary electrical connections, accomplished by means of ribbon cable of a variety manufactured for computer applications. The cable combined the virtues of durability, compactness and organization. Although only six wires were necessary to connect the control unit to the rest of the equipment, twelve-conductor cable was used to provide redundancy in the event of damage to some of the wires. The ribbon cable was anchored inside the housing by means of a terminal block (Fig. 2, Item 10 ). This effectively isolated the electrical connections inside the control unit from damage due to accidental pulling of the ribbon cable. The tactic was so effective that there were no instances of connections breaking inside the control unit until ne of the authors disassembled the controls while writing this article.

The entire control unit was held together with six #6-32 machine screws running from front to back and was attached to the prosthesis by means of three #8-32 machine screws inserted through the back of the control unit. The unit was attached to the thoracic interface in an optimal location for chin nudge control (Fig. 3 ).

Performance and Conclusion

The prosthesis was successfully used by this child with remarkable enthusiasm and skill for about 12 months. Eventually the development of limited manipulative skill with the right lower limb replaced the need for an upper-limb prosthesis. There were no failures or other problems with the controller. Performance of the device as a simple, durable and thoroughly reliable two-joint controller justifies its consideration for inclusion in prescriptions for powered upper-limb prostheses for children with severe quadrimembral deficiencies.

*Dankmeyer, Inc., 2010 Maryland Avenue, Baltimore, MD 21218

**Department of Orthopedic Surgery, Harvey 616, Johns Hopkins Hospital, Baltimore, MD 21205