Testing DC Power Systems

DC is not always DC. When systems are developed to operate on DC power, the DC electrical environment will have noise and may experience other disturbances. This article will explore considerations for generating low frequency disturbances (glitches, sags, surges, spikes) that occur in the 100 microsecond and longer region with bandwidths of 10 kHz or less. In contrast, noise, which is outside of the scope of this article, can be low frequency (like 50/60 Hz hum) but becomes a challenge to generate in range from kHz, to MHz or up to the GHz range.

The Agilent N6705A DC Power Analyzer is tuned for DC dis- turbance generation. Built-in power supplies can generate transients up to 600W, with rise/falltimes of less than 1 microsecond and bandwidths of 5 kHz. A built-in arbitrary waveform generator makes programming of disturbances easy.

Disturbances can be caused by a variety of factors. Source side disturbances can be due to events that cause a variation of the source, such as changing source output power due to environmental conditions (like a cloud passing over a solar panel or a car alternator spinning at a different speed). Load side disturbances can be due to changing loads being placed on the DC power, such as a DC rail being pulled down by new loads (devices) plugged onto it, or the powering up/down of subsystems that are turned on/off to conserve power.

When designing systems that run on DC, engineers need to consider the effects of DC disturbances on the overall operation of the system. Certainly, normal DC disturbance events, such as plugging in a new device, should not cause the DC rail to collapse and cause other devices on the same common DC power bus to lock up, reset, lose data or fail. Thus, design validation of DC powered systems and devices will require simulation of DC disturbances for testing to ensure proper operation of the system.

DC Systems and Possible Disturbances

Many systems and devices operate from a DC bias or rail. The system will operate properly if the DC rail stays at the desired DC operating voltage with some operating margin. Here are some examples of DC systems and how DC disturbances can occur, causing undesirable results.

We typically think of an automotive electrical environment as a 12V DC system. However, the voltage bus in a car has many deviations and disturbances. Changes in engine speed can cause voltage fluctuations and changing loads, such as activating window motors, can pull down the 12V DC bus. Significant transients are generated on the 12V DC bus during engine crank and startup. Electronics in the car, ranging from engine control units to entertainment systems to instrumentation, must be tolerant to these transients, because you would not want your car’s engine computer to lockup when you put on the brakes or race the engine.

When LED lighting runs from a DC power distribution system, disturbances on the DC system can impact the visible light output of the LED lighting. Disturbances and transients can produce unwanted and annoying flicker on the LED lights.

When multiple devices or subsystems are run from a common DC bus, adding or removing devices from the bus can cause transients that impact all of the devices on the bus. When a USB device is first plugged in, inrush currents can exceed the nominal DC operating current and even exceed the current rating of the USB 5V line. This is particularly true for high-powered devices, such as USB hard drives, that draw large initial inrush currents as the motor spins up the drive. This inrush current pulls down the voltage on the USB 5V line and if the voltage goes low enough, it could cause other devices on the USB to reset, leading to possible loss of data.

Like in the aforementioned USB example, startup currents can cause trouble even when the system is a fixed configuration and does not have devices added or plugged in. When FPGAs are first powered up, they can draw significant inrush or startup currents. These startup spikes, which can be 10x larger than steady state current, can easily draw down the DC rail within a system, causing a momentary sag that can adversely affect other components and subsystems running off that rail, causing resets or shutdown. While the effects of startup and inrush can be minimized by using a larger power source to power the rail, this may not always be practical, so the various subsystems may need to be tested to see how they tolerate disturbances induced by inrush current.

Generating DC Disturbances During Test

Summary table of key power supply requirements for DC disturbance generation.

System designs must be made tolerant to a range of DC voltage disturbances. This is particularly true of systems that depend on interoperability of plug-n-play devices. The average user just wants to be able to plug in a new gadget and expect it to work properly without impact on other gadgets running on the same DC bus. The only way to ensure this is to have standards or design specifications for DC bus stability and power quality, and then to design devices to be tolerant to those specified deviations. In R&D and design validation, engineers need a way to create a DC power bus with controlled, repeatable disturbances to test their designs to ensure they adhere to the standard or meet the design specification.

In order to test how tolerant a design is to voltage disturbances on the DC power bus, a special kind of power supply is needed. Not only does it need to be a DC source, but it must be able to generate non-DC output in the form of a time varying DC voltage transient that can be programmed to the shape of the DC disturbance.

Engineers have tried, sometimes unsuccessfully, to create a power supply for test that can generate DC transients. Some designs are based on an arbitrary waveform generator, but unless the device under test has very low power requirements, most arbitrary waveform generators cannot source enough current to power a device. Some engineers have attempted to enhance the arbitrary waveform generator by adding some form of power amplifier stage. The amplifier could be a custom design or an off-the-shelf amplifier, such as a car audio amplifier, which can generate high power with good bandwidth. But these homemade solutions cannot be counted on as a general solution, as these designs are difficult to make stable over a wide range of load impedances. Thus, the design of the source would need to be adjusted for each load condition.

When selecting an off-the-shelf DC power supply that can generate DC transients, dynamic specifications need to be considered. First, the power stage of the DC power supply needs to have sufficient speed to generate the designed rate of change of voltage. This specification is often called the risetime of the power supply or the programming response time of the power supply.

Next, the power supply needs a way to be programmed to quickly change its output to create the desired disturbance. One method is to use an analog programmed supply. In this scheme, an arbitrary waveform generator creates the shape of the disturbance and its output drives the analog input to the power supply, which acts like a power amplifier. Another method is to use a power supply that is fast enough to respond to a series of individual computer commands and rapidly reprogram/change its output to generate the desired disturbance shape. A third method is to select a DC power supply that has some kind of internal sequencing ability, where you can download the DC transient waveform into the power supply and have it execute the sequence to generate the waveform.

Another consideration is the capacitance of the device being powered, which appears as a load on the power supply’s output. The current needed to drive the device is defined by the equation I=C(dv/dt). So, as the rate of voltage change goes up, the drive current needed goes up. And if the device has a high input capacitance, the currents can be huge if you want to quickly change the voltage across the input of the device under test. So, you need to consider the current that is needed during the transient, not just the steady state DC current, and size the DC power supply accordingly.

A final consideration is how fast the power supply changes its voltage downward. This specification is the falltime of the power supply. Many DC supplies on the market cannot change their voltage downward at the same speed as they can change their voltage upward; in other words, the falltime is much longer than the risetime. It is common to see a rise-time specification of 20ms and a falltime specification of 200ms. When the device under test has a large capacitance on its input, once charged up, the capacitor stores the charge and holds the voltage steady across the capacitor. If you need to reduce the voltage into the device under test, the capacitor must be discharged for the voltage to drop. Some power supplies have a special circuit called a downprogrammer or active pull-down that engages a small electronic load in the power supply to pull current out of the capacitor. For these supplies, the falltime specification will be equal to or similar in size to the risetime, allowing for fast transients in either direction.

Off-the-Shelf Power Supply for DC Disturbances

With the understanding of the practical considerations for generating disturbances, let’s turn to possible off-the-shelf power supply solutions. To address high currents needed during disturbances, there are many power supplies available, but most high current supplies have slow rise/falltimes, making them unsuitable for disturbance generation.

Some power supply vendors make higher speed power supplies with fast risetimes and with downprogrammers, or even full 2-quadrant operation, that allow for fast changes down in voltage. These will be better suited for disturbance generation, but they will still need a way to be programmed fast enough to generate the transient. Again, analog programming, fast output programming via computer control or internal output voltage sequencing will be necessary.

Some engineers turn to 4-quadrant power supplies or bipolar amplifier power supplies. These tend to be available in low power (< 50W) with very high accuracy (sometimes called an SMU) or high power (1000 W), which are very heavy and bulky. In either case, a 4-quadrant supply is much more expensive than a regular DC power supply at the same power level. Although much faster than a regular DC supply, the output speed still needs to be considered when selecting a 4-quadrant supply. Furthermore, due to the limited number of choices, finding a 4-quadrant supply with enough speed, voltage, and current can be challenging.

This article was written by Bob Zollo, Product Planner, Agilent Technologies, Inc. (Santa Clara, CA). For more information, contact Mr. Zollo at This email address is being protected from spambots. You need JavaScript enabled to view it., or visit http://info.hotims.com/28057-403 .