Modern instrumentation can help reveal sources of power dissipation problems where electric motors are tightly integrated with their controls and test points are hard to find.
Large integral-horsepower motors account for about 90% of the electricity consumed by all motors. But large motors make up only 10% of all motor sales. The other 90% are fractional horsepower motors as found in power tools, industrial automation, elevators, and vehicles. Many of these smaller motors are characterized by a tight integration of the control system, motor drive, and motor itself. This integration forces engineers to take similarly integrated measurements of the control system, motor drive, and motor when debugging and validating the system.
It can be useful to see how modern instrumentation provides facilities for making these integrated measurements and for sniffing out energy efficiency problems. One instrumentation system designed for making these kinds of measurements is the Teledyne LeCroy MDA810 Motor Drive Analyzer. The following measurement examples demonstrate how one might use this analyzer to test a small hand-held tool containing a sine-modulated permanent-magnet synchronous motor operating at high speed.
The tool’s operation involves reversal of the motor direction once per second. The measurements examine how drive signals correlate to the behavior of the tool and dynamic power consumption. Measurements include the power consumption during the transition in rotational direction and the difference in power consumption while rotating in each direction. The goal: to understand and mitigate power losses during these periods, which, if too high, could annoy users and/or cause reliability issues.
A typical measurement scenario might use five 12-bit, 1-GHz analog acquisition channels. They would monitor two control signals, the encoder position of the rotor, the actual velocity of the motor, and the commanded velocity. An external controller board designed by the test engineer can process the encoder position and velocity signals for input to the MDA810. Because this is a sensorless motor, the encoder is added for test purposes only – it’s not part of the motor in normal operation.
The control signals on channels 1 and 2 tell the motor to change rotational direction. The rising edge of channel 1 initiates the reversal in motor rotation, while the falling edge of channel 2 represents the point in time when the motor completes its reversal. Capturing data over a longer period (five seconds, in this example) enables the viewing of many cycles of the transition. Zoom traces display details of one of these transitions, clearly showing the timing of the control signals and the motor’s response. We also monitor speed change at this transition point. From these signals, we see that the motor rotational reversal behaves well and operates as expected.
OUTPUT POWER ANALYSIS
In this example, we use a two-wattmeter method to analyze the motor drive output and calculate three-phase power values. The two-wattmeter method permits measurement of three-phase system power with only four signals, leaving other channels available to acquire control or power behaviors. The MDA810 also supports three-wattmeter methods.
In the two-wattmeter method, two high-voltage differential probes and two current probes connect to the motor drive output. A single acquisition of the line-to-line voltage and line current waveforms would show both to be 120° out of phase, normal in a three-phase system. While the un-zoomed waveform appears to be noisy, a zoomed view reveals that the appearance of noise is really the switching characteristics of the devices in the drive output. These details could not be seen with a traditional eight-bit oscilloscope, but the MDA810’s 12-bit acquisition system has enough resolution to make this observation possible.
A longer acquisition captures the complete motor rotational direction change and enables calculation of power values before, during, and after the change. Of interest is the amount of power consumed during the transition from one direction to another. Ideally, we do not want a sharp increase in power at this transition point. This longer acquisition contains two transitions of the motor direction.
To determine the cyclic period for making all voltage, current, and power calculations, we must choose a signal to contain the “reference period.” In the MDA810, we refer to this as the “sync” signal. The sync determines the measurement interval for computing the per-cycle voltage, current, power, efficiency, mechanical measurements, and other values. It is usually necessary to filter the sync signal to remove high-frequency content to get better periodicity, and this task is simple in the MDA810.
One may use the line-to-line voltage, filtered with a 500-Hz low-pass filter, as the sync signal. We examine the sync signal to verify it identifies the right time periods, thereby ensuring power calculations are correct. After verifying the measurement periods, we turn off the sync signal.
In this example, the values of greatest interest are the motor VRMS, IRMS, real power, apparent power, reactive power, power factor, and phase angle. The mean values of these measurement parameters for the full acquisition appear in a Numerics table display in the accompanying figure, similar to what a power analyzer would display. The P(∑rst) and S(∑rst) waveforms (overlaid on top of each other in the bottom right grid) are per-cycle “synthesized” waveforms that plot the per-cycle values (shown in the Statistics table) versus time. These per-cycle values are time-correlated to the original acquisition waveforms. They are created by touching or clicking a Numerics table cell value.
These per-cycle waveforms clearly indicate the dynamic power behaviors of the motor drive output and motor, something that would not be obvious by only viewing the mean value in a Numerics table. Viewing the real power and apparent power per-cycle waveforms at the directional transitions of the motor provides valuable insight into the power consumption during each direction change, important in this application because this motor is part of a hand-held tool. Minimizing power consumption keeps the tool from becoming too hot to hold.
For a closer look at the area of interest during one of the transitions, we use the MDA810 Zoom+Gate feature. It provides a simple means of simultaneously zooming all input sources, detailed waveforms, and sync signals. It can also position the zoom window on any portion of the trace. The common zoom window then acts as a measurement gate for the Numerics and Statistics tables.
We focus Zoom+Gate on a given area of interest, such as the transition from one rotational direction to the other. In the example, this is one complete cyclic period as shown by the DrvOutSyncZ sync signal. The Statistics table shows a power consumption of 3.894 W during the transition. This value is reasonable for the motor under test.
Determination of heat loss during operation provides a still deeper analysis of motor power consumption. We analyze the loss via the MDA810 by measuring power values using the Harmonic Filter settings at both Full-Spectrum and Fundamental simultaneously. After comparing these results, we calculate the heat loss in the windings as the difference between the Full Spectrum and Fundamental calculated real power results. The Harmonic Filter setup defines the filter applied to the input waveforms for power calculations. We may define this filter in both the ac input and drive output.
For this example, the ac input harmonic filter is set for Full Spectrum while the Drive Output is set for Fundamental only. We set the voltage and current inputs to be on the same channels in both setup dialogs. As a result, the only difference in the power calculations is the measured harmonics.
The accompanying figure shows Zoom+Gate set for one motor rotational operating cycle. The ∑abc parameters in the Numerics table represent the full-spectrum power values, while the ∑rst parameters represent the fundamental only. Using cursors, we can measure the time of the motor operating cycle and then calculate Joules from the power parameters. The equation for converting watts to Joules is:
E(J) = P(W) × t(s)
For this example, the power value shown in the Numerics table for ∑abc is 3.668 W. The length of time measured using cursors is 907.68 msec. This parameter displays as the ΔX value located in the bottom right corner of the accompanying figure. Converting to Joules, the energy consumed for ∑abc is 3.33 J. Using the same method to find ∑rst, we calculate 2.51 J was consumed. By subtracting these numbers, we arrive at 0.82 J. The 0.82 J value represents the heat loss in the winding for this tool while it rotates in one direction before switching directions.
Let’s look at an example where the power plots indicate an issue with the motor operation. In this next example, we test the same type of motor with the same two-wattmeter wiring configuration. The display also shows motor position (C3: encoder position, C7: sensorless position), control (C4), speed (C8), and power (P(∑rst) and S(∑rst)) signals.
Note that in the clockwise rotation, the per-cycle P(∑rst) and S(∑rst) power waveforms are smooth and consistent. When rotating counterclockwise, however, the power waveforms in the upper right corner reveal an oscillation in the signal.
After observing this vibration, further investigation discovered an issue with the motor commutation, which stemmed from a sensorless control issue.