Excerpt from Continuous Passive Motion CPM by Robert Salter

Engineering Aspects of the Design and Construction of Continuous Passive Motion Devices for Humans

By John Saringer, P Eng


The continuous movement of a joint by external mechanical means poses so many unique challenges that ideal solutions are nonexistent. Every approach represents a compromise. Different emphasis of the design goals inevitably leads to a different approach to building a device. The net result is the broad array of continuous passive motion (CPM) machines designed and built by various hospitals and companies throughout the world.

Criteria usually considered in the design and construction of these machines are as follows.

A variety of basic options are available to the designer and builder of CPM machines. Each selection results in a greater or lessor achievement of the above goals and a unique device results from every unique selection of options. With the passage of time the practical day-to-day experience with machines in use causes one or other approach to be favoured over others. Later-generation products then emerge looking more and more alike. There is also a significant creative and innovative element to the process that can not be ignored. The art of creative product design lies precisely in being able to find simple and creative solutions to constantly broaden and redefine the range of what is considered possible. The following discussion summarizes the current state of the art in CPM design with the options as now evident from the machines built to date.


There are essentially two ways in which a limb may be grasped for the purpose of flexing and extending a joint. The first is the approach taken by a physiotherapist when manipulating the limb manually. The limb is held and supported on both sides of the affected joint and pushed, pulled, and torqued to create the required movement. ROM is controlled subjectively and is usually limited to the amount of force the therapist deems safe to apply to the joint or by the range the patient is comfortable with. CPM devices that try to mimic this approach are called free linkage machines.

The second approach involves the use of a splint whereby the limb is cradled in a frame. This frame has adjustments to accommodate a range of limb sizes and is generally loosely bound to the limb to allow for some movement of the limb relative to the frame. The linkage is hinged at locations roughly coinciding with the joints of the body. The frame then resembles a splint with moving hinges. A motor drive acts on the frame to cause articulated back and forth movement of the hinges, and concurrently, similar movement of the joints. Devices employing this approach are called motorized splints.

Motorized splints are themselves further classified as either anatomic or nonanatomic. An anatomic frame, as the name suggests, is a frame that attempts to mimic the natural movement of the patient's joint with the hinge axes coinciding with the joint axes. The match is never very close due to the complex and variable nature of joints; nevertheless, when the attempt to achieve anatomic alignment has been made, the device is called anatomic.

A nonanatomic motorized splint consists of a frame where no real attempt is made to mimic anatomic movement. Hinge locations are often far removed from joint axes and the motion of the limb only roughly correlates with motions of the frame. This approach relies on relative motion between the limb and frame to constantly compensate for the misalignments.

All three of the above types of machines are available commercially and each offers significant advantages over the others under certain circumstances. The following table briefly summarizes how well each approach generally satisfies the criteria listed at the beginning of the chapter. A discussion follows.

Free Linkage
Mot. Splint
Mot. Splint
Patient size adjustability   Excellent   Poor   Fair
Longterm comfort   Excellent   Good   Fair
Versatility   Good   Poor   Poor
Light weight   Good   Poor   Fair
Range of motion   Poor   Excellent   Good
Multi axis motion   Good   Poor   Fair
Joint loading   Constant   Variable   Variable
Precise ROM control   Very poor   Excellent   Fair
Ease of use   Good   Fair   Good
Joint stability   Very poor   Good   Fair

Adjustment to a variety of patient sizes is most easily achieved with a free linkage machine where there is no splint to size to the limb. An anatomic splint is invariably the least adjustable.

Absence of multiple support points rubbing on a limb generally favours the free linkage approach for long-term comfort. Anatomic frames cause less rubbing, however, the nonanatomic device benefits from changing pressure points during the movement cycle.

The absence of a bulky frame allows greater versatility, the capability to ambulate, and lower weight with free linkage machines as compared to either type of motorized splint.

Figure A.1.   On a nonanatomic machine, the measured ROM at the patient's knee (A) may differ from the indicated ROM (B) by 20º or more.

Full and precise range of motion is easiest to achieve with an anatomic frame because full travel to the frame hinges translates directly into joint movement. In both other approaches there is no such direct relationship between the hinge motion and the joint and hence ROM control is compromised. Free linkage machines rarely attempt to calibrate angular joint movements because there is no hinge point that can be used to measure angles. Nonanatomic machines are inaccurate because the hinge angles, which can be precisely measured and controlled, only roughly approximate joint movement. This problem is most evident when trying to move stiff joints. In this case actual joint movement can be far less than what is attempted by the machine. Differences between actual joint angles and that shown on the ROM control dials or hinge goniometers can be in error by 5 to 10º for an anatomic device and 20º or more for a nonanatomic frame. Displayed angles on machines invariably show the largest achievable angles, which are rarely obtained in practice. Nonanatomic motorized splints built for the knee, for example (see Fig. A.1), typically display ROM control of -10 to 120º while actual knee ROM measured with a goniometer to the patient's knee would typically show O to 100º of movement at most.

Stabilizing a joint that is being moved, the forces imparted to a joint other than torsion during movement and the ability to manipulate a second axis of rotation simultaneously are all closely interrelated. An anatomic device with its hinge points roughly coinciding with the joint often employs more rigid and constraining methods of support to the limb precisely for the purpose of ensuring the anatomic alignment. The end result is relatively stable and constrained single-axis movement which can protect the joint if it is unstable due to ligament or tendon damage but also limits a second axis of motion when desired. Forces on the joint vary in direct relation to how closely the device is anatomically aligned to the joint and how rigidly the limb is constrained. Poor alignment and rigid restraint result in part of the misalignment being forcefully taken up in the joint. In anterior cruciate ligament (ACL) repairs to the knee an anatomic splint has distinct advantages over the alternatives, however, in the case of elbow reconstruction where flexion and extension as well as supination and pronation are required (two axes of movement) the anatomic splint is the least practical.

Free linkage machines offer the least support to the joint and are unsuitable for unstable joints because the joints are largely unrestrained. This works to advantage, however, when the joint is stable because the joint can be manipulated in more than one axis and there is no frame misalignment that can cause excessive loading of the joint. Loading is light, cyclic, predictable and usually compressive in flexion and distractive in extension.

Nonanatomic frames offer a compromise between anatomic and free linkage machines. Limbs are usually less restrained in their frames, as they must be to accommodate misalignment, providing for less support against joint instability. Loading to the joint is inconsistent, largely unpredictable, but never very high because the limb is generally free to slide around.

Ease of use is somewhat affected by the choice of approach. Anatomic frames generally require far greater care in setup and regular monitoring to ensure that anatomic alignment is maintained. If this care is not taken the advantages of the device are defeated. Nonanatomic frames are more tolerant of setup errors and hence are easier to use by untrained staff. Free linkage machines are much easier to adjust to different patient sizes and bulk because of the absence of frames but require greater care in range of motion setup.


After the question as to how to support a limb has been addressed, the next question is how to drive the limb support back and forth through a desired range. There are two options. First, you can hold on to the end-points of the frame as far away from the joint (and hinge) as possible and push and pull to bend the frame at its hinge or the limb at its joint. A push-pull mechanism is called a linear actuator and it is generally attached to extreme ends of a linkage to provide maximum leverage, i.e., a minimum drive force creating maximum torque at the hinge point. The second option is direct rotary drive to the hinge. In this case leverage is very poor. The two types are easily illustrated, e.g., operating a screwdriver is an example of rotary drive whereas using a wrench illustrates leveraged linear drive. Long wrenches give much greater leverage than short ones. Without exception a rotary drive gives better functional performance in safety, range, and consistency of motion and linear actuators are favored where force and torque requirements are too high for a rotary drive to handle practically. Each of these aspects is separately discussed below.

In the area of force and torque capability there is no contest. The very high leverage available to a linear drive makes it the obvious and straightforward choice when a heavy limb such as a leg is to be moved. Of the more than 30 machines that have been built to drive a knee joint to date all but two use linear drives. For smaller joints such as for the hand, torque requirements are small and hence both types of drive are feasible. Medium-sized joints such as the elbow are more easily driven with linear drives but rotary drives can be substituted without great hardship.

Safety is greater using a rotary drive for two reasons. First, the connection between the driven hinge and the motor is more direct with less danger of uncontrolled movement. The frame in a linear drive is usually adjustable between the drive connection points and if this adjustment loosens then the frame can suddenly collapse. The second factor is that a linear drive involves a scissoring-type mechanism which invariably increases risk of pinch points and entrapment of uninvolved limbs.

Range of motion is generally unrestricted and more precisely controlled with a rotary drive, hence it is possible to build rotary drives to manipulate limbs through their full anatomic range. This is not possible with a linear drive because linear drives can only effectively act in narrow rotational ranges before the angle of push or pull is ineffective. A linear drive moving through more than 90º becomes inefficient since most of its delivered force beyond this angle pushes and pulls on the frame rather than flexing the hinges and joints.

Rotational speed and smoothness of motion are much better with a rotary drive because angular speed does not vary during the movement cycle and there is direct coupling through gearing of the motor to the joint. A linear drive operating at a constant motor speed delivers a variable rotary speed which depends on angle. At full extension the rotary speed at the joint is much higher than when it is flexed, with the secondary consequence that the linear drive spends much more time in a flexed position rather than extended than does its rotary counterpart.

Figure A.2.   In this free linkage design, a mechanism (A) is used to create an additional rotary movement (8) by coupling to the primary movement.


It is desirable whenever possible for a CPM machine to be capable of moving a joint through its entire physiologic range of motion. While this is rarely required, the capability of doing so allows applications for the broadest possible range of indications and patient population. Most joints move through more than one axis and some, like the shoulder and hip, move in three.

Building a device to move in two axes is very difficult and the approach usually taken is to couple the secondary axis to the movement in the primary axis. In this way only one motor and drive system is required since the movement in one plane forces motion in the second. Figure A.2 shows a free linkage elbow unit where the changing angle of the elbow joint is used to cause supination and pronation of the wrist resulting in flexion, extension, and rotation of the elbow joint.

Commercial products to move a joint in three planes are only available set up to move one or two axes at a time. Experimental units with three motor drives to move all three axes of the shoulder have been built but their bulk, complexity and cost have so far prevented their commercialization.


The motor power requirements of a CPM device are governed by the maximum speed and loading on a machine. The smallest of motors can drive the largest machine provided that the reduction gearing is high enough. Because gears are much cheaper than motors it is always easier to gear the machine down to a lower speed than it is to increase motor power. This is particularly true of the larger machines built for the lower limb where power requirements are highest. CPM manufacturers frequently compromise speed and load capability in an effort to reduce cost, increase longevity, and reduce weight of their product.

Clinical requirements for speed (cycle time) are currently vague and poorly understood. It is logical to assume that speed should influence the therapeutic value of CPM in areas such as venous flow in the limb, synovial fluid flow, swelling, pain, and wound healing; however, given current knowledge on the subject there is no way in which one speed can be strongly recommended over another. Experience at The Hospital for Sick Children, Toronto suggests that higher speeds in the range of 1 cycle/min or faster produces excellent results. How much this speed can be reduced without compromising performance is unknown. 





Figure A.3.   A light weight, highly efficient unit for the hand.



Figure A.4.   A light weight, highly efficient unit for the jaw.

Given the uncertainty as noted above, CPM machines should be equipped with variable speed ranges with a maximum speed of approximately one full range of motion cycle per minute. Maximum loading on a device should be equivalent to a 110 kg or larger adult male unless the device is specifically designed for children.

Current practice in the industry is to use conventional high-quality brush-type direct current (DC) motors as the power source. These motors are usually run at less that 50% of their rated speed and load capacity to increase their longevity and decrease noise. Alternatives such as stepper motors, brushless DC or alternating current (AC) motors are also feasible but offer less power to weight capability and lower electrical efficiency and hence are rarely used.


Electrical power consumption of CPM machines is within a range where battery operation is feasible. Typical average power required to move a joint at 1 cycle/min varies from a low of 0.005 W in the case of a single finger to 1.0 W for an entire lower limb. Electrical inefficiency and redundancy in power conversion result in a 3 to 20 fold increase (or even higher) in power actually delivered to a machine to deliver this electrical work to a joint. When a variable speed control is added the electrical efficiency decreases further.

In the smallest CPM machine designed to move a finger, the least amount of power required using the most efficient drive design without speed control is 0.005 x 3 = 0.015 W. Two AA alkaline batteries have 3 W-hr x 2 = 6 W-hr of power capacity, therefore the battery life for a hand device can be as high as 200 hours (battery life = 6 W-hr/O.O15 W = 200 hours). Devices for the hand and jaw shown in Figures A.3 and A.4 actually achieve this performance. Nickel cadmium cells, when used, have only 25% of the power capacity of an alkaline cell.

In the largest CPM devices for the lower limb, the minimum power requirement without compromising maximum speed is 3 W. The largest readily available alkaline batteries are D cells each with 15 W-hr of capacity. If four of these batteries were used, a machine would run 20 hours on a set of batteries or 5 hours using rechargeables. This performance is marginal and therefore almost all lower limb CPM devices are operated with AC/DC power supplies.


Certain basic controls including range of motion adjustments and an on-off switch are necessary on a CPM machine. Speed control is also a frequent addition. Additional features are frequently added but rarely improve the clinical effectiveness of the product. The possible exception to this is the inclusion of an interface for a neuromuscular stimulator. This feature allows synchronized use of both devices for patients where both treatments are separately indicated.


The necessity and difficulties associated with building a CPM device sufficiently durable to provide years of service is rarely appreciated or understood. Few electrically operated machines are operated for as many hours as those that provide CPM. In a single year a CPM machine may operate up to 6500 hours (over 3 million cycles) in a busy hospital. An automobile driven for the same length of time would travel almost 500,000 km (312,500 miles). Needless to say, an inordinate amount of care must go into the selection of materials and components to build these machines to obtain sufficient durability. Inevitably there is a tradeoff with weight and other performance features which are often compromised to achieve reliability. Most commercially available CPM machines are built for a service life of at least 5000 hours of use and a few are serviceable up to 15,000 hours.