Skip to content
Aerospace & Defense, Gimbals & Optical Mounts, Optics & Photonics, White Paper
White Paper

Motion Actuator Designs for Electro-Optic Test Systems

Tom Markel
Aerospace Strategic Accounts Manager

This article discusses motorized angular drive mechanisms and performance differences between direct-drive gimbals, indirect-driven gimbals and articulated-arm robots for electro-optic (EO)testing applications. From this article, readers will gain an appreciation for the pros and cons of each drive technology and be able to make better choices when deciding which automated testing system would be best for their EO testing applications.

A gimbal is a two- or three-rotary axis device that has the axes of rotation at right angles to each other and intersecting. Gimbals are typically constructed with either direct-drive or indirect-drive(gear or tangent-arm drive) rotation axes. Figure 1 shows the conceptual architecture of a two-axis and three-axis gimbal, respectively

Figure 1. (a) 2-axis gimbal conceptual architecture and (b) 3-axis gimbal conceptual architecture.

Both gimbals and articulated-arm robots can be used to manipulate various sensors, instruments or objects for EO testing applications. It is important to be familiar with the various motion architectures and configurations to determine which will suit your application’s particular requirements. Deciding on the required actuator precision for EO testing is key to making sure the system performs well enough to characterize the sensor being tested. Accuracy is certainly important, but there are a number of other important parameters engineers must weigh when selecting the motion systems for their EO test applications. For example, motion system performance versus size and cost should be evaluated. Competing motion system solutions should also be compared based on size, weight, integration complexity, speed, duty cycle, maintenance, initial purchase price and total cost of ownership.

Direct-Drive Gimbals

Direct-drive gimbals use motors attached directly to the rotation axes without mechanical couplings or gearboxes between the motor and the output rotation shaft. For precision applications, the actuator typically uses an optical encoder, resolver or inductive feedback to sense the position of the shaft relative to the stationary housing. Since the feedback and drive mechanism are directly coupled to the output shaft, this design lends itself to high precision and dynamics. In addition, the reliability and life of the actuator is very high due to the lack of mechanical wear and friction in the drive mechanism.

Both mechanical and air bearing axes are used in these designs. Mechanical bearings are more tolerant to impact loading and typically have higher stiffness. However, there is some small amount of friction and higher rotational error motions when compared to air bearings. Using a high-resolution encoder and low-noise power electronics, both bearing designs can achieve sub-microradian motion with a direct-drive mechanism.

Figure 2 shows an example of a direct-drive gimbal design.

Figure 2. Aerotech’s AMG200 direct-drive gimbal.

Direct-drive gimbals and rotary positioners are the highest performance EO test solutions in terms of accuracy and repeatability, and they offer high acceleration, low-jitter tracking and fast direction reversal capability. They also have the very small motion step capabilities needed for long-range pointing and are often used for pointing targeting systems at fast moving objects, simulating motion for inertial equipment testing and pointing telescopes at satellites or other distant objects for laser communications. Direct-drive gimbals can be configured for two or three axes of motion and combined with linear stages to cover hemisphere test regions as discussed in the article Electro-Optic Sensor and Systems Performance Verification with Motion Systems. One-sided gimbals, like the one shown in Figure 3, provide a good configuration for EO sensor azimuth and roll test capability for products such as missile seeker or tracking sensors.

Figure 3. One-sided gimbal used for azimuth and roll testing. The large aperture on the roll axis allows the payload (missile or tracking sensor) to extend into the aperture.

A mirror can be mounted on a one-sided gimbal, like the example shown in Figure 4. This approach, which eliminates the secondary rotary axis support, can be used to simplify the design and reduce some costs. However, care must be taken to ensure the design meets the required dynamics and load capacity requirements.

Figure 4. 300 mm cell diameter, one-sided mirror gimbal.

Direct-drive gimbals can make incremental motion steps down to 0.01 μrad and readily achieve bidirectional accuracies better than 8 μrad over 360 degrees of travel.

High-precision designs using air bearings and special feedback configurations can easily achieve better than 1 μradaccuracy over 360 degrees of travel. In-position jitter can reach nanoradian levels with the proper control tuning, power electronics and vibration isolation. Rotational error motions at the sub-microradian and sub-micrometer levels are readily achievable depending on the design and bearing configuration.

Direct-drive gimbals are ideal for high-performance and high duty-cycle applications in both lab and production environments for EO sensor testing.

Gear-Driven Gimbals

Gear-driven gimbals are a type of indirect drive system with gearing between the drive motor and the output shaft. Worm gears, planetary gears and harmonic (strain-wave) gears are the most common ways of accomplishing gearing between the input drive and output shaft. An example of a gear-driven gimbal that uses worm gears is shown in Figure 5.

Figure 5. Aerotech’s AMG-GR gear-driven gimbal.

One major advantage of gear-driven rotary axes and gimbals is that they allow for a smaller input motor since the output torque is multiplied through the gear train. This approach reduces power consumption and can save costs. Also, the gearing can be designed to prevent backdriving, which can be advantageous with large, unbalanced loads.

However, gear-driven designs have drawbacks. The gearing system typically has some small amount of backlash or mechanical play to function properly. Designs, like harmonic gearing, can be made to minimize – or even effectively eliminate – this backlash when initially designed. However, it is nearly impossible to make the backlash truly zero and maintain that zero backlash over the lifetime of operation due to wear and friction. Because of this backlash and mechanical imperfections in the gear teeth, accuracy and repeatability aren’t as good as direct-drive designs. Adding a direct-encoder feedback to the output shaft can improve accuracy and repeatability. However, because mechanical play still exists between the gearing and no servo system is able to respond instantaneously to disturbances, the motion performance can never achieve performance levels that are capable in direct-drive designs.

In addition, while initial costs may be low, gear-driven designs typically end up having a higher total cost of ownership due to the gear wear and the resulting maintenance and rebuild cycles that are required to keep them performing at “like-new” levels. Other drawbacks of gear-driven designs include lower acceleration, lower rotational speeds and lower duty-cycle capability.

In general, gear-driven designs are great for medium-performance applications that have low power consumption requirements or position large imbalance loads. Like direct-drive systems, they can be assembled in two- and three-axis gimbal designs.

Gear-driven gimbals can make incremental motion steps at the single-digit micro-radian level. Accuracies on the order of 10’s or 100’s of microradians can be achieved. Similar to direct-drive designs, in-position jitter can reach nanoradian levels with the proper control tuning, power electronics and vibration isolation. Bidirectional repeatability is usually on the order of the achievable accuracy (10’s to 100’s of microradians) due to the aforementioned backlash and mechanical play. Rotational error motions typically reach the microradian and micrometer levels due to added parasitic drive forces that are present on the rotational output shaft.

Although not as precise as direct-drive designs, geared rotary system performance is often adequate for lab testing, low-volume manufacturing and medium accuracy testing of EO sensor systems.

Tangent-Arm Drive Gimbals

A tangent-arm driven rotary stage uses a lever, or tangent arm, to convert linear motion to rotary motion. In these designs, the lever arm is attached to a linear motion mechanism, typically a lead or ballscrew. A rotary servo motor is then coupled to the screw to produce linear motion at the end of the tangent arm. The tangent arm couples to the output rotary shaft and as a result, acts in a similar way as a gear-driven design.

In this design, the tangent-arm mechanism must include added mechanics such as flexures or mechanical bearings to eliminate the kinematic over constraint that exists and permit angular motion of the output axis. In these designs, rotary output motion is typically limited to a few degrees of output.

A conceptual model of a tangent-arm drive rotary is shown in Figure 6.

Figure 6. Conceptual model of a tangent-arm drive rotary.

Tangent-arm driven rotaries can be designed to make incremental motion steps to well below 0.5microradians with careful design. Similar to gear-driven designs, error sources and backlash in the linear drive can limit bidirectional motion performance. Adding a direct encoder on the output shaft will improve positioning accuracy and repeatability as discussed with gear-driven designs. Similar to other designs, in-position jitter can reach nanoradian levels.

One disadvantage of this type of design is that output angular motion is typically limited to a few degrees. Speed and accelerations are very low as well and generally, these designs are not suited for high-duty cycle applications.

In satellite sensor testing applications, high-speed positioning is typically not required. Because tangent arm gimbals have low in-position jitter, they are good choices for testing high resolution sensors or satellites where applications require lower speeds and duty cycles and don’t require large angular travel ranges.

Hybrid Solutions

It is often necessary to mix drive actuation strategies on gimbals for a specific application. For example, one axis might require high-speed and continuous travel where a different axis on the same system might need to hold a large unbalanced load and only rotate a few degrees.

Figure 7 shows an example of a hybrid solution where the azimuth axis is direct-drive and elevation axis is a tangent arm design.

Figure 7. Hybrid gimbal with a tangent-arm elevation axis and direct-drive azimuth axis mounted on a long-travel ballscrew linear stage.

Articulated Arm Robots

Articulated arm robots can be configured with a number of different degrees-of-freedom and drive mechanisms. Typically, the arms use geared-rotary axes at the joints and through a kinematic model of the robot (known lengths and positions of the arms), the end effector or tool can be calculated and controlled.

An example of an articulated arm robot is shown in Figure 8.

Figure 8. An example of an articulated arm robot.

Depending on the size, these robots can move over large volumes, carry thousands of pounds and have a large range of angular motion at the tool point. Because of the complex possible poses and varying moment loads on the robot, tool point accuracy can vary as the robot moves through space or uses different poses for the same tool location.

Most articulated arm robot manufacturers only specify repeatability and resolution. Positioning accuracy is generally not specified due to the difficulty in performing the measurement as well as the generally poor positioning accuracy available with this type of motion architecture. As robot arms extend further from the rotational joint, positioning errors increase since the angular errors are multiplied over a distance. In addition, gravity and changing loads on the joints, arms and mounting base affect tool point and end effector positioning. Positioning accuracy can easily be on the order of millimeters and 10’s to 100’s of milliradians over the full travel range.

Using external feedback such as machine vision or laser trackers may seem like a way to improve the positioning accuracy of the robot. While some improvement may be made with external sensors, generally their overall complexity in implementation and limitations in measurement capability prevent them from improving the positioning accuracy and repeatability to levels that are necessary for stringent EO sensor testing. In nearly all cases, it is more affordable and higher performance is achieved by using combined cartesian and gimbal designs to accomplish the testing tasks. Example architectures are discussed in the article Electro-Optic Sensor and Systems Performance Verification with Motion Systems.


A summary of the advantages and disadvantages of the various motion actuation technologies used for EO testing is shown below in Table 1.

Table 1. Qualitative summary of the various motion actuation technologies used in EO testing (++++ = better, + = worse)

CharacteristicDirect DriveGear DriveTangent armArticulated arm
Accuracy++++ +++ +++ +
Repeatability ++++ +++ +++ +
Small step capability ++++ ++ ++++ +
Speed / acceleration ++++ ++ + +++
Good for unbalanced loads+ ++++ +++ ++
Travel / work volume+++ +++ + ++++
Initial cost + +++ ++ ++++
Total cost of ownership ++++ ++ ++ ++

As with most engineering decisions, there are tradeoffs that must be weighed during the system design process. Armed with this information, new testing applications can be assessed for the optimal motion actuation strategy and more informed decisions can be made when discussing the various approaches with motion control and automation vendors.