DC Link Requirements for Grid Tie Inverters

A common question we field regarding Grid-Tie inverters goes something like "Can I interface to a 480V grid with a 680V DC Link?".  To answer this question, lets consider a typical Grid Tie Inverter or Active Front End application as illustrated in the figure below.

Generally, our application code uses center-aligned, space vector pulse width modulation (SVM) techniques for maximum DC Link voltage utilization.  Using SVM, the switch-mode AC voltage output of the inverter, before the filter components, is limited by the dead-time of the power stage and any minimum pulse requirements of the IGBT drivers.  For example, 3usec of dead-time when switching at 10kHz, results in a maximum achievable duty-cyle of 97%.   Depending on theswitching frequency, these hardware limitations will set the maximum, peak, line to line output voltage to ~95% to 97% of the DC Link.  Note however, that this is the point at which the duty-cycle is clamped, and the controls become non-linear.  In practice, you will want to allow several percent duty-cycle for control headroom in order to provide the controller linear operating range.  As such, 92-94% is a more practical design value.  

The previous discussion addresses the maximum obtainable voltage at the output of the IGBTs.  Considering a GTI or AFE converter can source current into the grid, one also has to allow for voltage drops across the filter components at maximum output current.  Let's consider a typical 100kW Grid Tie Inverter interfacing to a 480V AC line.  At 10% high line, the maximum peak line voltage is 480*1.1*SQRT(2) = 747Vpk.  Assumming 30Vpk is dropped across the filter components, the required minimum DC Link voltage is given by (747+30)/0.92 = 845V.

Common Mode Filtering Considerations for Grid Tie Inverters

In non-isolated, grid-tie inverter applications, it is common practice to connect the neutral point of the 3-phase AC grid to earth ground.  Unfortunately, this creates a problem for DC link common mode filtering.  Typical control methods utilize space vector modulation to control the power stage, which introduces a common mode voltage in order to maximize the DC link voltage utilization.  When you connect a capacitor between DC link and neutral, the space vector common mode voltage is impressed across it.  Unless the capacitor is very small, it will adversely effect operation, often leading to control loop instabilities.  In addition, this capacitor will tend to resonate with the grid connect filter, causing further behavioral problems.

As such, implementing a common mode filter for these converters requires a slightly different approach in order for the system to operate correctly.  First, if you are trying to control EMI, you should be able to get away with fairly small capacitors.  Very low ESL is generally much more important than capacitance value.  Use a number of small caps (in the range of 0.01uF) in parallel to drive ESL to an insignificant value.  Even though you grid filter is inductive, the large components tend to have a lot of parasitic capacitance.  Inserting a common mode inductance by passing the three AC lines through one or more large ferrite toroids will be much more effective at mitigating high frequency noise than the large inductors, and will reduce the EMI capacitor peak currents, and the resulting resonant frequency.  Keep in mind that you are still going to see mid-frequency oscillations with this configuration, but the higher frequency EMI (several MHZ+) will be much lower.

You may also want to include a similar common mode inductor on the source side to allow common mode filtering of the supply lines.  You may actually find that it’s OK for the DC link to have a moderate common mode voltage with respect to earth ground as long as you are able to attenuate most of the common mode signal from the AC and source lines.

Effects of Cable Length on SCR Drive Waveforms

A general rule of thumb when driving SCRs is to use as short a cable as possible in order to provide the most efficient turn-on characteristics.  As cable length increases, two important parameters are degraded: inductance and series resistance. 

In order to minimize inductance, gate drive cables should be constructed in a manner which tightly couples, preferably with a twisted pair, the gate and cathode connections.  However, regradless of the construction technique, increased cable length results in increased inductance, which in turn slows the rise time, di/dt, of the gate drive current pulses.  The second concern when lengthening cables is the assocated increase in series resistance which reduces the peak current, I_GM, delivered to the SCR gate. 

The magnitude of the gate drive current  I_GM, as well as it's di/dt rise time, directly effect key SCR performance characteristics, including:

• Turn-on Delay
• Turn-on fall time of teh anode voltage
• Turn-on switching losses
• di/dt of the anode current

High I_GM, typically >2Amps, and rise times of > 2Amp/sec help ensure reliable and efficient triggering of SCR power devices.  We recently tested OZSCR1000 and OZSCR1100 gate drive performance using a variety of cables and cable lengths.  The following figure illustrates the gate drive current pulse waveform when driven with a 13.5"cable, (500mA/div, 2usec/div).  The second figure illustrates the current pulse waveform when driven with a 150" cable, (500mA/div, 2usec/div).  From these scope plots, the effects of both inductance and resistance are obvious by the increased rise time and reduced current level in the second plot.  Please reference Oztek Application note AN-0001 for more detailed test results.

 

Grid Energy Storage Is a Growth Industry to Watch

The market for grid energy storage is robust and growing. According to a study published by GTM Research, the U.S. market for utility-scale grid storage is expected to rise from an annual value of about a quarter of a billion dollars in 2009 to about $2.5 billion by 2015.

The potential exists for continued rapid growth in this market in the United States and worldwide. Demand will be filled by companies pursuing a variety of technologies to fill the needs of the power grid.

What is Grid Energy Storage?

In an ideal energy world, demand for electric power would be matched exactly by electric power supplied, with no need for backup power sources or storage anywhere in the grid. But, in the real world, demand and supply are often mismatched.

Demand for electric power varies by time of day and changes with every season. Power supply varies based on many factors, including plant downtime, changes in fuel supply and prices, and the intermittent nature of renewable energy sources such as wind and solar.
 
The imbalance between demand and supply is evened out by various technologies that store energy and make it available to the grid as it is needed. With the rise of the "smart grid" and the increasing importance of renewable energy sources, the demand for grid energy storage and the complex technologies that supply it is large and growing.

Sources of Grid Energy Storage

There are numerous technologies already available or being developed for supplying reliable and economic grid storage. Batteries and pumped hydro systems are among the most mature grid storage technologies.

Batteries can store excess electricity from wind and solar plants for distribution to the grid when it is dark or the wind stops blowing. One strategy on the drawing board is to use electric cars as sources of grid energy while they are plugged in at night for recharging.

Pumped hydro systems use excess generating capacity during off-peak hours to pump water up hill for later release through turbines to generate electricity during peak demand periods.

Other technologies at varying stages of research and commercialization include flywheels, compressed air storage, super-conducting magnets and ultra-capacitors.

Interfacing with the Grid

With each of these energy storage technologies comes the need to be able to store and return the energy to and from the grid as needed.  Grid Tie Inverters (GTI) or Active Front End (AFE) power converters are typically used to provide this function.  Advanced digital control solutions such as the OZGTI3000 can be used to implement custom OEM solutions for this power conversion application.

Each grid storage technology has costs and benefits. The grid of the future will use a combination of several of them for maintaining the crucial balance between supply and demand.

Photo by kevin rigdon.

What is Smart Grid Technology?

Demand for electricity is expected to continue growing at historical rates for the foreseeable future. While utility companies strive to meet that demand by financing and constructing new power-generating plants, fueled by coal, atomic energy, wind, sunlight, and water, they’re also scrambling to create a “smart grid.”

The essence of the smart grid technology is using information technologies to better manage the generation, flow, and most important, the consumption of electricity over the course of each day and each season. By monitoring and managing the flow of current at a much finer scale than is possible today - metering its usage down to the level of individual office buildings and homes and even specific light sockets and appliances - it should be possible, experts say, to better match supply and demand on a minute-by-minute basis and thus conserve electricity and save everyone significant money.

As it is, the nation’s hundreds of electrical power-generating facilities pour their electricity into a vast network of long-distance and local power lines - the electrical grid. Much like water in a set of tanks connected by pipes, this electricity sloshes around within this grid and the grid’s managers do their best to move current to wherever demand is highest - to the time zone where it’s morning, for instance, and millions of people are switching on lights, showering, shaving, cooking breakfast, and starting their clothes washers. 

Unfortunately, all the grid’s managers can do is increase supply as best they can. They have no control over demand. They cannot, for instance, charge Mrs. Jones more for running her washing machine at breakfast time rather than in the middle of the night, when overall demand for electricity is particularly low. Instead, the power companies always have to be ready to supply power to every Mrs. Jones at any moment of the day. This has meant constantly building more and bigger generating stations.    

Smart grid technology would change this scenario by enabling market signals to be collected, distributed, and interpreted in more or less real-time. As demand rose in a particular area - due to unexpectedly hot weather, for instance - residential and office air-conditioners might be remotely adjusted by the power company to run a bit warmer in return for a break in the price of electricity. 

Ultimately, by providing more control over the consumption of electricity, the use of smart grid technology could save the U.S. billions of dollars.

Photo by Going Wimax.

Wind Power Projection Dependent On Improved Energy Storage Technology

Long believed by many to be one of the "purer" sources of alternative energy, any wind power projection has noted the rapid growth in the industry during recent years.  The World Wind Energy Association calculates that electricity from wind accounted for 2.5% of worldwide consumption in 2010 at 430 TWh, and that this output doubled what was available only three years earlier.  Many European nations, such as Spain and Denmark have begun intensive investments in wind energy and Germany is expected to be a major player following the forecasted shutdown of the nation's nuclear industry.  Outside of Europe, China has aggressively pushed wind energy in Asia and currently has the world's largest installed wind power capacity at almost 42,000 MW.  

Though many continue to advocate for wind energy as an alternative to a dependence on fossil fuels, numerous challenges still lie in the way of wind's widespread acceptance as an energy source.  Since wind is fortunately quite benign, especially when compared to the negative potential of nuclear power, safety hazards do not amount to much.  However, due to its fickle nature, wind faces an substantial economic challenge wherever it is installed. 

Even though there are a number of ideal "wind farm" sites worldwide, wind is always somewhat intermittent and possibly unreliable.  There is no guarantee that strong winds will blow during times of peak energy consumption, and since storage capacity (e.g. batteries) are very expensive and relatively primitive energy storage is a challenge.  This forces turbine operators to dump excess energy into the power grid, often at times when the price for electricity is very low.  This in turn makes the income from wind power unreliable and discourages investment in the industry, even though the technology has been proven and is now produced on a rapidly expanding scale.  However, wind power projections estimate yearly growth worldwide to be 28% annually (World Wind Energy Association), and this coupled with improvements in storage technology could make wind a vital component of the global communities future energy solution.

Alternative Energy Distribution Methods

The term alternative energy generally describes renewable sources such as solar, wind, and hydroelectric power. Hydroelectric energy accounts for the majority of U.S. renewable energy, at 70 percent, though the use of solar, wind and other renewable energy is continuing to grow. Methods of alternative energy distribution vary, and here I’ll provide a brief description of the three most prevalent methods. It’s important to keep in mind that digitally controlled power systems will help prevent wasted energy, no matter which distribution system is used. 

Large, Centralized Facilities  

Localized Alternative Energy  

Microgrids  

Photo by Fresh & Easy Neighborhood Market.

Utility-Scale Flywheel Energy Storage & Grid Tie Inverters

A few days ago Beacon Power announced that their installation of the world’s first 20MW Flywheel plant is currently operating at 18MW, and is expected to be running at full, 20MW capacity before the end of the month.  We’re pretty excited about this at Oztek, as we provide the controls for both the grid tie inverter that interfaces with the grid, and the sensorless permanent magnet motor drive  used for the flywheel in Beacon's system.  This program posed many design challenges, and it’s always extremely satisfying to see things finally come together, particularly on something as  large and complex as this. 

World's Largest Flywheel Plant in Stephentown, NY

This exciting news was picked up by quite a few sites, and as with any emerging technology, response seems to be as much confusion as it is appreciation and interest in what it offers.  The Beacon system is not a power generating plant.  Rather, it is a frequency regulation plant that stabilizes the grid by providing power during periods of high demand, and storing energy during periods of low demand.  The need for this stems from the fact that power plants cannot rapidly throttle power output to respond to sudden changes in demand.  Elaborate systems have been developed over the years to deal with this that, by their very nature, require a net surplus of power generating and distribution capacity to be built  in.  However, these systems still place demands on power generation plants that cause additional wear and tear, and greater production of CO₂ and other emissions than if the plants were allowed to operate under more steady state conditions.  The Beacon system is clean and very efficient, wasting only a small percentage of the energy it processes, and it’s able to respond to large demand changes in seconds.

For those interested in more detail, here’s an interesting article Why Frequency Regulation is Important, as well as a more comprehensive white paper on the subject  Frequency Regulation Basics - Oak Ridge NL .

MODBUS SCR Controller

Conventional SCR controllers typically use simple analog and discrete digital signals to control and supervise operation of the product, whether it be an SCR controller for heaters, an SCR controlled rectifier, or an SCR motor control.  However, as customers expect more and more out of our products, this simple user interface limits our ability to provide value add in the products we design.

The OZSCR1000 SCR Controller overcomes this limitation by providing a MODBUS serial interface in addition to the conventional control interfaces.  While the MODBUS interface provides a simple means to configure the product, it also opens the door to advanced features.  Using MODBUS, the user can:

• Implement sophisticated architectures employing multiple control boards
• Monitor system performance using fault and instrumentation data
• Provide improved maintenance & trouble shooting using detailed fault and performance data

MODBUS is one of the most popular industrial communication protocols in use today.  It's simple, inexpensive, and easy to use.  MODBUS was developed in 1979 by Modicon (now Schneider Electric) for use in communicating with multiple devices over a single pair of twisted wires.  While the original implementation operated over an RS-232 hardware layer, today's MODBUS devices run over virtually all communications media, including wireless, fiber, ethernet, etc.  Today, MODBUS-IDA (www.MODBUS.org), the largest organized group of MODBUS users and vendors, continues to support the MODBUS protocol worldwide.

The OZSCR1000 implements the MODBUS RTU protocol over an RS-485, multi-drop, serial interface, allowing many devices to operate on a single pair of twisted wire.  The following figure illustrates a typical system with a PLC serving as the MODBUS master controlling several slave devices:

Power Factor and Grid Tie Inverters

One reason engineers specify active rectifiers (a.k.a. active front-ends) for their systems is that they can operate with near unity power factor.  Being nearly the same system (see my recent post "Active Front-end or Grid Tie Inverter?"), grid tie inverters share this same beneficial characteristic.  However, this does not mean that an active rectifier or grid tie inverter must operate with near unity power factor, and in fact, we can use this to our advantage in certain applications.  This also implies that we may not want to specify power factor as a means of quantifying how well the system minimizes AC line current harmonics.

Let’s back up for a moment and first consider what we mean by Power Factor.  The purest definition is that power factor is the ratio of measured power to apparent power; that is PF = W/VA.  However, you may be more familiar with a couple of other equations that address two very different classes of applications.  PF = VA*COS(phi), where phi is the phase angle between voltage and current, is valid for linear loads that draw sinusoidal current and have negligible harmonic distortion.  This equation is well know by those working with AC induction motors, and until the last 20 or so years, is the one most used by electrical engineers. 

Typical inductive (AC motor) current and voltage

More recently, PF = 1/SQRT(1+THD²), has come into common use, as it addresses nonlinear rectifier loads used in a wide range of electronic equipment.  This equation ignores phase shift, however, the error this introduces is small in typical cases where THD dominates.

Typical 3-phase current and voltage with bridge rectifier and LC filter

So how does this relate to power factor and grid tie inverters?  Consider first how the grid tie inverter’s power stage and modulator are configured.  Most circuits (OZGTI3000 and others used by Oztek) are designed to inherently produce sinusoidal voltages.  This means that even without control loop intervention, to a first order, the inverter emulates a linear system.  We can then vary the AC voltage amplitude and phase angle in relation to the line voltage to produce currents with any phase relationship to the AC line that we want.  In fact, it may be obvious at this point that the resulting currents are simply a result of the differential voltage impressed across the grid interface filter impedance. 

With a properly designed filter, the differential voltage is relatively small in comparison the AC line voltage, making the precision afforded by closed loop control necessary for a practical implementation.  Closed loop control also helps compensate for non-ideal power stage behavior, and we determine how well this is all working by measuring THD.  Phase angle can actually be set to any practical angle.  If you want to move real power, it has to be in the vicinity of zero degrees, but it can be leading or lagging.  This can be used to advantage in larger installations to offset the lagging power factor produced by induction motor loads, or to compensate for transformer leakage and distribution inductances.   

Typical end-systems usually require compliance with more complex distortion specifications like IEEE 519, which specify current harmonic limits in absolute terms, place limits and both individual harmonics and the total, and limit even harmonics to much lower levels than odd harmonics.  In specifying grid tie inverters and active front-ends, you should therefore be primarily concerned with how well current harmonic distortion is controlled and whether current phase angle is readily adjustable, not power factor.