In Search of the Zero-Ohm Load
High-wattage resistors often are used as test loads for power transformers. Resistors don’t introduce phase shifts, and it’s easy to check that a transformer’s voltage regulation remains within spec at different rms currents. The effect of various primary tap and secondary loading combinations also can be measured. On the other hand, actual loads don’t always have a constant resistance (CR) characteristic.
Many switching power supplies present a constant power (CP) load to the AC input. DC power supply outputs are intended to provide a constant voltage (CV) or constant current (CC) depending on the mode selected. To simulate anything other than simple CR, you need an electronic load.
Both AC and DC electronic loads are commercially available with DC loads being more prevalent. Either type can emulate a short circuit as well as the CC, CV, CR, and CP modes of operation. According to Adrian Butoi, western regional manager at NH Research, AC loads are used for test applications that require linear or nonlinear AC loading with power and crest-factor control. In addition to the five basic modes of operation, the company’s Model 4600 AC Load also provides unity power-factor loading and a complex nonlinear waveform mode.
Built-in measurements include frequency, voltage, peak voltage, current, peak current, crest factor, apparent power, true power, peak power, reactive power, power factor, and resistance. DC load measurements compose a subset of this list that is not related to reactance.
Cliff Nazelli, the managing director of marketing and sales at PPM Instruments, described a couple of typical applications. In one example, vehicle electrical-system power-distribution module testing must simulate various load conditions. While the module’s durability and temperature rise are monitored, the load current is switched on and off. In another test, fuel-cell impedance is measured by modulating the DC load current. The impedance is determined from the current and voltage amplitudes and their phase relationship.
Isolation also is critical in many cases, especially those that involve off-ground differential voltage sources. Multitap battery load testing is an application that requires this capability.
DC Load Basics
Typically, MOSFETs are used as the dissipating devices in a DC load. These semiconductors have ON resistance much less than 100 mΩ, and several are operated in parallel to achieve the needed power rating. Of course, paralleling devices also reduces the resistance the load presents.
ON resistance is higher for high-voltage MOSFETS than for low-voltage devices, so DC loads rated for 500 V generally will have higher resistance than 50-V loads even if the same number of MOSFETS is used. Figure 1 shows this effect for a 5-kW PPM 600-V load (dark green curve) compared to a 5-kW 60-V load (red curve). The initial slope of the PPM Modular Electronic Load (Mel) 5000-200-600 is 30 mΩ compared to the 1-mΩ resistance of the Mel 5000-600-60 even though there is only a 3:1 ratio relating their maximum currents. The resistances of the lower wattage Chroma loads range from 5 mΩ to 25 mΩ.
Figure 1. Safe Operating Areas for Various DC Loads
Source: Chroma Systems Solutions and PPM Instruments
All electronic loads have a safe operating area (SOA) limited by voltage, current, and power. These areas are indicated in Figure 1 for the PPM Mel 5000-600-60. As the input current increases, the voltage across the load also increases because of the load’s finite resistance. At the maximum current, higher voltages can be supported but only to the maximum voltage and power ratings.
On a graph with linear X- and Y-axis scaling, the maximum power curve is a hyperbola. The load voltage can continue to increase to the maximum voltage limit as long as the current is low enough that the maximum power limit isn’t exceeded. As the figure shows, the SOAs of loads having the same power limit are bounded by the same hyperbolic curve although it may be intersected at different places by the voltage and current limits.
It’s also important to note the lack of standardization regarding load specification in datasheets. The NH Research Model 4750 DC Load and PPM Mel Series show the SOA on a graph with voltage plotted vertically and current horizontally on a log-log grid. Chroma’s 63600 Series datasheet plots this information with the same axes but on a linear grid.
Several manufacturers’ graphs showing low-voltage characteristics generally have linear grids with current plotted vertically and voltage horizontally. Of course, depending on the DUT, current or voltage could be the independent variable, so it really doesn’t matter how the graphs are drawn as long as you understand what they mean.
Finally, model numbers may include power, voltage, and current limits but not always in the same order. PPM’s Mel Series lists power, current, and voltage: Model 5000-200-600 is a 5-kW load with 200-A and 600-V maximum limits. Chroma’s Model 63630-80-60 is a 300-W unit with 80 V and 60-A capabilities.
DC Load Selection
You must choose a load with an SOA sufficient to handle the maximum current, voltage, and power expected from the DUT. The actual combination of current and voltage can lie anywhere within the SOA. Nevertheless, the trend in semiconductors is toward low voltage and high current so supporting this combination often is a major consideration.
Jim Dougherty, senior engineer at Chroma Systems Solutions, explained, “Up to a maximum current limit, a DC load presents a constant minimum resistance. For example, if a load can support a 10-A current and has a 10-mΩ resistance, the DUT output voltage must be at least 100 mV even if the connections and wiring were perfect. Taking into account the finite resistance of the connections and wiring further increases the minimum DUT output voltage required for the load to sink the full current. Of course, the load can be programmed to represent a higher resistance but you cannot get less than the minimum.
“Suppose you have designed and characterized your DUT interconnect and cable resistance and found that RSeries = 2 mΩ. Further, assume you need to support 0.5 V @ 60 A. To do this,” he continued, “the DC load must represent a resistance RLoad <(0.5/60)-0.002 or RLoad <~6 mΩ. The sloping lines in Figure 2 show the resistance associated with three models of Chroma 63600 Series Loads and correspond to the area circled in Figure 1. As expected, the higher the output current rating, the lower the resistance. And, loads can be operated in parallel to achieve higher current capacity as well as lower resistance.”
Figure 2. Low-Voltage Resistance Characteristics of DC Loads
Courtesy of Chroma Systems Solutions
What if you need to sink the full rated current but the voltage drop associated with the load resistance and wiring is large compared to the DUT output voltage? Adding a boost supply in series with the load effectively increases the DUT output voltage and makes possible even zero-volt full current loading.
However, Mr. Dougherty cautioned, “Boost supplies should be considered only as a last resort. The effective power of the load that otherwise would be available to the DUT is reduced; you no longer can perform transient tests; noise from the supply affects ripple and noise measurements; and complexity and cost are increased.”
Commenting on his company’s true zero-volt PLZ-4WA Series of DC loads, Takuya Takeda, vice president of Kikusui America, said, “The PLZ-4WA employs a bias supply installed directly in the electronic load. It supports true zero-volt operation by stepping up the input voltage and applying a load voltage adequate to allow the internal current source to operate correctly.
“A switched-mode power supply is small enough to fit inside the load,” he continued. “Typically, a switched-mode supply can create noise issues, but that is not the case with this bias supply. It has been designed using zero-volt switching technology and other special techniques to reduce noise.”
An extensive feature set has developed around the basic DC load function to address a wide range of applications. Four-wire Kelvin connections ensure that the DUT terminal voltage is used in power calculations, not the load voltage that is reduced by wiring and connection IR drops. Also, because many tests require switching the load on and off to stimulate DUT transient response, this aspect of DC load design has become very sophisticated.
A DC load’s internal wiring and terminations must present a low impedance, not just a low resistance. Mr. Nazelli explained that PPM uses a heavy-duty laminated copper bus structure internally in conjunction with a proprietary FET circuit board layout to ensure the lowest possible impedance. In particular, the laminated copper bus minimizes the impedance increase caused by skin effect that otherwise would occur at high frequencies.
The depth at which AC current density has been reduced to 37% of its value at the conductor surface is given by
where: δ = skin effect depth
ρ = conductor resistivity
ω = 2πf
µ = conductor absolute magnetic permeability
For a copper conductor, skin depth varies from more than 8 mm at 60 Hz to only 66 µm at 1 MHz. Engineers have used Litz wire for many years to minimize the resistance increase caused by skin effect. Litz wire is stranded, but each strand also is insulated. A laminated bus achieves a similar result in a form that may be more easily manufactured and terminated, especially for high current levels.
The PPM Mel units are specified with a 15-µs to 20-ms rise time, selectable in 36 discrete steps, and a DC to 10-kHz frequency response. A 600-A load has a maximum slew rate of 600/15 or 40 A/µs. When loads are connected in parallel, the slew rates add. On the other hand, the highest practical slew rate is limited by the inductance of the wiring needed to connect the loads in parallel.
Kikusui’s Model PLZ1004W is a 1-kW load with a maximum current rating of 200 A and a slew rate of 16 A/µs. For the PLZ-4W Series, the slew rate is variable over a 100:1 speed ratio and guaranteed to be accurate to within 10% for current within 2% to 100% of rated value. This series also supports frequency range selection.
According to Kikusui’s Mr. Takeda, “Dynamic response only is required for transient response tests of power supplies. A wide bandwidth isn’t necessary for static tests such as load variation tests and foldback characteristic tests. Excess bandwidth affects load stability so the PLZ-4W/4WA load includes selectable bandwidth, and it can be optimized to match the kind of test and test condition.”
NH Research’s Mr. Butoi explained that because the load manufacturer cannot control DUT and cable inductance, the most straightforward solution to mitigate voltage spikes, ringing, and oscillation is to allow load-current slew-rate programmability. Usually, slowing this slew rate eliminates the problems.
PPM provides two types of filtering as discussed by Mr. Nazelli, “The architecture of the Mel employs a control board with programmable loop response to vary the control-loop speed where you need to adjust rise/fall times for pulse tests and stimulus/response testing. This is one type of filtering. Separately, filtering is used on the FET circuit board assemblies to control the feedback loop response of the power-dissipating devices. The operator can adjust the loops to shape the response either to further control rise/fall times or eliminate oscillations induced from external reactive components.”
Power to the Load
Regardless of the other characteristics a DC load may have, it must dissipate power-sometimes a lot of power. The products in PPM’s Mel Series can handle 1 kW to 5 kW, and master-slave configurations up to 80 kW are standard. Eight models in NH Research’s 4700 Series range in capacity from 1 kW to 36 kW. Chroma’s Series 63200 High Power DC Loads are available in sizes from 2.6 kW to 15.6 kW. These three load series are air-cooled.
Most manufacturers of heavy-duty loads support paralleling for higher power handling. Loads feature individual device protection against over-temperature, over-voltage, and over-current conditions and further ensure performance through active current balancing. For larger power ratings, master-slave systems usually are rack mounted and up to 6 ft high.
One alternative to a big air-cooled unit is a water-cooled unit. AMREL’s PLW Series handles up to 250 kW, and several versions of the 36-kW model are available in a 4U-high x 27.5″ deep rack-mount size. As a comparison, a 5-kW AMREL Series PLA Air-Cooled Load is approximately the same size.
Another solution appropriate for high-power applications is Kikusui’s Model PLZ6000R Regenerative DC Electronic Load. The basic unit acts as a 6-kW load although only about 15% of this is dissipated. The rest of the power is regenerated as a synchronous AC current fed back into the AC mains. Up to five units can be combined in a master-slave system to provide a 30-kW capacity.
Several ranges of benchtop DC loads also are available with ratings to a few hundred watts. Three models from Chroma’s 63600 Series are shown in Figures 1 and 2. B&K Precision’s 150-W Model 8540 handles up to 60 V and 30 A in the CC, CR, and CV modes with current, voltage, and power measurements presented on an integral display. Model 8510 has a 600-W capacity with 120-V and 120-A limits. It, too, provides an integral display and includes a CP mode as well as battery test capability.
Kikusui’s PLZ-4W Series features 165-W, 300-W, 660-W, and 1-kW models. In addition to the basic products, the 165-WA and 660-WA models are available with a built-in bias supply and support true zero-volt operation.
The 300-W Model LD300 DC Load manufactured by Thurlby Thandar Instruments and available in the United States from Saelig has 80-V and 80-A maximum ratings. It supports the CC, CV, CR, and CP modes and provides a transient generator, a variable slew rate, soft start, and a current monitor output.
Controlling the Power
In addition to a load’s fundamental capabilities, the extra features it offers can be important depending on the types of tests you need to run. Kikusui’s PLZ-4W Series includes soft start, a variable slew rate, a switching function, a preset memory function, 100 setup memories, and a sequence function.
B&K Precision’s 8500 Series Loads also support battery testing by measuring total battery discharge in amp-hours. Jeremy Lo, an application engineer at the company, said, “Software is available to control the load for this test. It plots battery discharge curves in real time as well as gives you the option to export raw data in text or Excel format for further analysis. The software also can be used to monitor and plot power, voltage, and current levels at the load inputs.”
In NH Research’s Model 4700 DC Electronic Load, an auto mode provides glitchless switching among the CR, CC, CV, and CP limits. Further, you can programmatically control the mode of operation and its duration via a 100-step customizable macro with 10-µs timing resolution.
PPM’s Mel Loads have several means of control. RS-422 and USB 2.0 ports are standard with both GPIB and Ethernet optionally available. You can log into a load’s IP address to perform remote control and diagnostics. According to Mr. Nazelli, “This also enables PPM to send feature improvements without hardware intervention and without requiring you to return units to PPM. In addition, you can add modules in the field to upgrade a load. The load automatically reconfigures itself to its new capabilities.”
Chroma’s Model 63472 High Slew Rate DC Load incorporates Intel’s power test tool (PTT), which can simulate microprocessor load changes of up to 150 A at a 1,000-A/µs slew rate. Because the PTT is small enough to fit into a microprocessor socket, it cannot dissipate the required power without significantly changing its temperature and operating characteristics. The 63472 provides measurement hardware and over-current and over-voltage protection as well as the automatic calibration required to ensure test-result accuracy.
Unless you use a device such as the PTT, there’s no way to connect a high slew-rate load that will not introduce significant errors. The PTT mimics the load presented by a microprocessor, which may have 100 or more power and ground pins. The slew rate at each pin is only a few amps/µs, and most of the transient current is provided by local capacitors. With the PTT, you are testing the capability of the power supply in combination with local capacitors to cope with the overall 150-A changes and 1,000-A/µs slew rate.
Many models of DC electronic loads are available, some with very specialized capabilities. Determining the load that will best fit your test requirements starts with a list of the specifications you must have. Power dissipation, maximum current and voltage, and the minimum resistance the load presents are key to most applications. So, too, are the modes in which you will operate the load and how it changes from one to another.
(Source Tom Lecklider – Evaluation Engineering http://www.evaluationengineering.com/features/2009_september/0909_electronic.aspx )
Latest posts by Mike Novello (see all)
- GTEM Cells for Sale. - October 27, 2017
- Barely used lot of EMI / EMC test gear for sale – open to offers - August 24, 2017
- We are looking to buy two JDSU HCU-1500 chassis. - August 22, 2017