Solid ground with selecting servos
To make your work more fun while selecting drives for precision motion control applications, here are some of the typical goofs that you can avoid. These common oversights are offered by three engineers who receive those challenging calls from other engineers pleading for assistance because “things didn’t go as planned.”
Know thy load
In addition to knowing the load profile before even beginning to select a drive, the experts tell us, you must know — not just perceive but know — load inertia changes, not just the worstcase conditions. Charles Geraldi, division manager of Parvex Servo Systems, emphasizes that many drive specifiers completely neglect to consider how the load inertia changes as a machine moves or a process evolves. Such changes are intuitive, but they are commonly overlooked.
Consider this case: While selecting a drive for a robot, a conservative designer sized a servo for the maximum torque needed. He wisely calculated the inertia with the arm at maximum extension and holding the largest load. So far, so good, but he stopped there.
Result, he has a large enough servo, but it cannot be tuned to control an empty arm accurately at the shortest extension. It is unstable — hunts before finally reaching the starting point. He would have avoided the quicksand if he had given his supplier the ranges of the inertias rather than just the worst-case condition. The supplier’s application engineer could have then selected a different motor inertia and speed-changing ratio to optimize the performance at all operating conditions.
Synchronized conveyors are another sleeper. Some designers specify drives for a fully loaded unit, but neglect to consider how it responds when the conveyor is carrying only one or two parts. Slow conveyors, no problem, but high-speed units with precision matching requirements are another story.
The effects of torque ripple are frequently a costly oversight, reports Mr. Geraldi. Often, significant dollars are spent to change a drive motor with a conventional armature to one with a skewed armature, because the specifications omitted the maximum allowable torque ripple at low speeds. Applications such as laser imaging and profile inspection machines are more demanding than, say, machining applications, even for some aircraft parts.
Equally common, according to Robert Kish, an application engineer at Pittman, are the cases where well-meaning engineers use preliminary data for servo drive selection, without updating it when the final parameters are available. Granted, this sounds too basic to be a goof, but he believes it happens with increasing regularity, because of personnel reductions in many engineering departments. During the haste to get a product shipped or a machine into operation, that last important step is often omitted.
Determining the load
The inertia and friction losses of many machine components can be calculated. However, many others are so complex such calculations are extremely time consuming, even with computers. One way through this maze is to operate a machine with a servomotor for which you know or can obtain the torque per ampere, frequently called Kt or torque constant. Then you can ratio this test performance to the desired performance to establish the needed servo rating.
In addition to knowing the torque constant for a specific servo motor, you need the change in this value as the motor loads up. This information is typically published as two factors — “change of torque/deg C” and “temperature rise in deg C/watt.” This temperature rise value is the rise above the ambient temperature.
In general, most brushless dc (BLDC) motors produce nearly constant torque per unit of current, others may not. In any case, the linearity should be established. If a manufacturer omits this information in the published data, it can usually be obtained for the asking.
When approaching the final selection, be advised that some servo manufacturers give their continuous torque ratings at 40 C ambient and others at 25 C. Therefore, you should derate the values at 25 C by 12 to15% to get the operational torque rating, because motors seldom operate at 25 C, which is near room temperature.
Temperature and space
A recent application had two gotchas. The first failure was caused by the designer neglecting to consider the ambient temperature for a complex machine that was installed near a furnace. Soon after startup, the servo performance deteriorated, then it became a smoke generator. The obvious solution was to install a larger motor that could survive the heat. With minor machine modifications to accommodate the larger motor, the new motor went through its paces — for awhile.
This time no smoke, but the performance again rapidly deteriorated. The second gotcha had landed. Although the new motor was larger, it too was equipped with an encoder. Generally, these aren’t designed for high ambient temperature. Solution: replace the encoder with a resolver, which can handle the heat.
Continue on Page 2
Controller mismatch
Another common problem, says Motion Science’s Dan McFarland, is mismatching a feedback device, its operating speed, and a controller.
Typical case: To satisfy requirements for high linear speed and accurate resolution, a designer chose a lead screw that required 20 revolutions per inch of linear travel, a motor-mounted encoder with over 4,000 pulses per revolution (ppr), and a specific motion controller. However, during startup, they learned that the controller couldn’t keep up with the feedback signals when the motor was delivering the maximum linear speed. The pulse rate from the encoder (several hundred thousand pulses per second) was above the controller’s capability. He had essentially four choices:
• Install a new controller capable of handling the higher pulse rate.
• Change the lead screw to a coarser thread.
• Use a new encoder with fewer pulses per revolution.
• Combine some or all of the above.
This situation taught several people two lessons — more fully understand all the components and test the machine before shipping it to the customer.
This application also raisies another design consideration — the tradeoffs between position accuracy and repeatability and how these criteria influence the bottom line.
Response or repeatability
With the awesome capabilities of microprocessors and digital signal processors (DSPs), it is often a wise step to forego high resolution and depend on repeatability, comments Mr. McFarland. For example, some applications need the ability to move something from point A to point B by programming the motion controller. Other applications only need to repeat specific move sets that are taught to the motion controller by running the tracer through the steps. Then the machine will follow the instructions it was taught. This is just one way a precision system can compensate for machine variables, such as lead-screw thread variations, yet deliver highly repeatable moves.
Budget
One of the biggest time-wasters is neglecting budgetary considerations early in the design phase. It’s the old story of engineers designing a much better product than the market will buy. As Mr. Mc- Farland notes, a lot of time is often spent by both the equipment builder and the drive supplier before budgetary constraints enter the picture. This oversight frequently sends the project back to the drawing board and generally delays the project.
We gratefully acknowledge the important contributions to this article by engineers from the following companies.
Motion Science, Parvex Servo Systems , Pittman Servo Motors.