One of the most productive new approaches to power quality is immunity testing — making sure that loads can tolerate common disturbances without additional filtering or support. This kind of testing is often done with a piece of equipment called a voltage sag generator. At first glance, sag generators seem simple — just a variable transformer of some sort, coupled to electronic switches. In reality, though, they are complex devices that are becoming increasingly critical with the adoption of standards such as SEMI F47 (see sidebar, on page 26).

In every broad-based power quality survey, voltage sags are given as the most common disturbance. Introduced on single-phase and 3-phase systems by faults, momentary overloads, and undersized wiring and transformers, voltage sags disrupt sensitive loads in almost every environment — commercial, residential, and industrial.

Traditionally, you mitigate voltage sags with devices (e.g., ferroresonant transformers, tap switchers, dynamic sag restorers, and uninterruptible power supply (UPS) systems) installed between the utility grid and the sensitive load. These solutions are expensive, but effective.

It makes more economic sense to adjust the design of a sensitive load so it can tolerate ordinary voltage sags — especially if these adjustments don't cost anything. To do this, engineers need a reliable, repeatable source of poor-quality power — the sag generator (Photo 1).

A typical sag generator (Photo 2) supplies rectangular sags, which are of constant depth and varying duration. The user-selected depth typically falls between 40% and 90% remaining voltage, or it may be set at 0%. User-selected durations normally range from 0.5 cycles to several hundred cycles.

Class Distinctions

There are three common classes of sag generators: amplifier-based, transformer-based, and switching-impedance-based.

Amplifier-based generators create output voltages and currents from an arbitrary wave source. This means they can generate controlled harmonics and varying fundamental frequencies as well as voltage sags. It also makes amplifier-based sag generators more expensive, per kVA, than transformer-based sag models.

Consequently, most high-power sag generators are based on some form of electronically switched variable transformer. The simplest switches for the variable transformer are, of course, contactors. But contactors cannot provide precise switching for very short sags and do not permit phase-angle control of the sags. Therefore, most sag generators use electronic switches.

To minimize the switching transient, the generator must switch “off” at a current zero-crossing and “on” at a voltage zero-crossing (Fig. 1).

Depending on the load's displacement power factor, which the sag generator's designer cannot control, this can introduce up to a half-cycle switching delay.

Clever techniques, such as using the sag generator to analyze the load's current waveform before it switches, can minimize, or even eliminate, this delay for certain types of loads.

Some sag generator operators just ignore this problem and hope the resulting switch transient doesn't create an unexpected response, which it usually doesn't.

Aside from switching transients, there is another challenge with transformer-based generators. Power-frequency transformers are heavy, and sag generators are usually portable. The practical limit for airline shipment is roughly 150 lb, which sets an upper limit, even with toroidal transformers, of about 25kVA per phase.

To address this limitation, a new type of sag generator uses a high-frequency, switched impedance to create controlled sags. This technology promises to reduce the weight and increase the maximum power of portable sag generators. Although this type of generator is still a laboratory tool, it may become commercially available in the next year or so.

Floating the Components

You connect sag generators between the utility grid and the sensitive load. This is a simple process for single-phase loads, but a complicated one for 3-phase loads. For example, you may want to connect a sag generator to generate sags between two legs of a 3-phase delta load, or between one leg and neutral of a 3-phase wye load. That's why it's important for the sag generator's design to float the input connectors, fuses, transformer(s), and electronic switches with respect to earth or neutral.

While this approach allows for greater connecting flexibility, it also necessitates strict adherence to insulation requirements. Usually, the insulation must be tested for ten times the application voltage, or 5000V on a 480V system. And great care must be taken to ensure that the electronic switch-control signals, which are always referenced to ground, are sufficiently isolated from high-voltage sections of the switches.

Of course, the sag generator itself must be grounded for safety, and all exposed conductors and the enclosure must be bonded to the protective earth conductor.

On the Safe Side

One of the challenges in sag generator design is meeting safety standards that are not written with sag generator applications in mind. For example, a sag generator must have high-voltage, high-current connectors that you can readily attach and disconnect. This is not a common requirement around high-power devices.

In addition, sag generators must protect themselves from a wide variety of load faults because loads can behave in unexpected ways when hit with sags. Nevertheless, domestic safety standards intentionally disable any electronic protection in the sag generator during safety qualification on the grounds that the protection is not inherently reliable. Even the definitions in the standards may require special interpretation: for example, given the intentionally varying output of a sag generator, what exactly is the “rated output voltage?”

Sag Generator Protection

It is not unusual for a well-behaved, 15A electronic load to suddenly decide to draw 200A or more immediately following a sag (see Fig. 2, on page 26).

Even loads with carefully designed inrush limiting may be designed to disable their current limiting once they are fully powered up. And the power factor and harmonic characteristics of the load are unknown. Many 3-phase electronic loads will draw more than their maximum specified current on a non-sagged phase during a sag. All of these characteristics make it a challenge to protect the sag generator from the load's behavior without altering the required characteristics of the sag. A well-designed sag generator will consider these characteristics and will protect itself from damage, regardless of the load's behavior.

Data Acquisition Systems

During voltage sag testing, loads either work or they don't. If they work, there's no problem with voltage sags. If they don't, you need to determine exactly what (in the load) failed so you can adjust or fix it. For this reason, every sag generator comes equipped with a data acquisition system.

Simple data acquisition systems have a few channels, each measuring voltage relative to ground. But often the interesting voltages — transformer outputs, DC supply outputs, and even floating logic signals — are not near ground. Therefore, differential floating channels with safe connectors are almost a necessity.

Common data acquisition systems accept ±10V input signals. But with sag generators, channels rated for ±1000 V are useful for power-system voltages and high-voltage power supplies. Other channels rated for ±25V cover common low-voltage supplies and signals, including 24VDC.

As a practical matter, you should consider attaching readily viewable meters to the data acquisition channels (see Fig. 3, on page 28). Often, there is some minor confusion about which channels are connected where, and meters can quickly answer any questions.

Finally, it is useful to have BNC connectors that provide scaled, isolated-output voltages that track all floating channels. These connectors provide inputs to equipment such as oscilloscopes, or external strip-chart recorders for backup recording.


Today, sag generators are readily available. If you use them, you can often achieve quick, inexpensive increases in your equipment's tolerance to voltage sags. But don't forget that these devices also are more complex than they initially appear. Before you use a sag generator, consider such things as size, switching techniques, safety standards, and acquisition systems. When you set out to conduct your own immunity tests, you'll be glad you did.

Alex McEachern is the president of Power Standards Lab and the official liaison between the IEEE and the IEC on all power quality issues. He represents the United States on IEC 77A WG9 and also serves on the board of directors for Dranetz-BMI and Electrotek Concepts. You can contact Mr. McEachern through his Web site,

SEMI F47 - A New Type of PQ Standard

SEMI F47, a new standard from Semiconductor Equipment Materials International, is less than a year old. Unlike most power quality standards, which specify how good the power must be, this standard takes the opposite approach: it specifies how tough the loads must be.

Based on a two-year, worldwide monitoring project, SEMI F47 stipulates that semiconductor manufacturing equipment must tolerate voltage sags to 50% remaining voltage for 200 ms, 70% remaining voltage for 500 ms, and 80% remaining voltage for 1 sec. It also recommends, but does not require, some additional sag immunity points. An associated standard, SEMI F42, specifies test methods for complying with SEMI F47.

There are signs that this real-world approach will expand to other types of manufacturing equipment. For example, IEC 61000-4-11 is undergoing revisions in a similar direction. After all, it makes sense to ensure that equipment can tolerate the normal bumps that exist on every power line. And it' far less expensive to slightly adjust a piece of the equipment's design up front, compared to the alternative of purchasing voltage regulation for every installation.

For further information about SEMI F47, including application notes, visit