Find a downloadable version of this story in pdf format at the end of the story.
Introduced in the late eighties, The IGBT quickly displaced high-voltage power Darlingtons in motor drive modules and gradually worked its way down into all kinds of off-line motor drives in consumer products, particularly air conditioners and washers. From there it migrated into a diverse range of applications, from plasma displays to welders, from induction cookers to piezoelectric ignition and hybrid powertrains.
IGBTs have a number of advantages, stemming from its high power output per dollar (see sidebar: The Success of IGBTs). IGBTs rated up to 3kV have been commercially available for a number of years and they are widely used in traction applications. At these voltages, the IGBT does not have credible contenders, even though carrier recombination limits its frequency of operation to a few kHz. The only competing technology is silicon carbide. Transistors implemented on silicon carbide structures are majority carrier devices and can operate at very high frequency. They will become a very serious threat once a cost-effective manufacturing process comes online.
But choosing the right IGBT for the application is no simple task. The equipment designer's foremost concern is to find which device is the most cost-effective in his application. He cannot deal with technology curves; he has to deal with part numbers from several suppliers.
For a given voltage, package and short-circuit capability, there will be many possible candidates. Some will be more efficient than others and, possibly, more expensive. In those applications where power density is important, a smaller heat sink is a desirable feature that may justify the extra cost of a more efficient IGBT.
Years ago, International Rectifier pioneered the Current vs. Frequency curve to help designers compare different IGBTs in an application-related environment (Fig. 1). It provided valuable information at a time when the equipment designer had a limited number of choices and available IGBTs shared similar technologies. Today, the choices are many, characteristics are different, BOM (bill of materials) cost has become critical, and time-to-market has shortened.
This boils down to a massive headache for the designer who has to pick up the right device for the application through a laborious iterative process. This process cannot be short-circuited but can be streamlined and rationalized with some helpful tools. International Rectifier has just introduced one such tool: The IGBT Selector tool.
The purpose of the tool is to trim the list of candidates to a handful, sorted out by price or power dissipation. Some key data need to be entered, like voltage, current and frequency, as well as package type and short circuit rating, as shown in Fig. 2. The tool uses this data to calculate power losses in the specified application conditions and only those devices that would operate at a junction temperature at least 25°C below maximum rating are returned as potential candidates. Enough heat sinking is assumed to keep case temperature at 100°C.
Continue on next page
Figs. 3 and 4 show how intricate the choice of an IGBT can be. With the data entered in Fig. 2 only three devices meet the stated conditions. However, if the short-circuit requirement is reduced from 10 to 5µs the list of candidates expands with the addition of less expensive devices, on one end, and more efficient ones, on the other.
Conduction and switching losses are calculated with the help of a statistical device model obtained with a large number of measurements at different currents and temperatures. These models can be profitably used to cross a competitive part to an IGBT from International Rectifier. The IGBT Cross-reference tool, available in the IR website, matches key data sheet parameters from a large number of commonly available IGBTs to one or more part numbers from IR.
In the end, despite the availability of tools, there is one device characteristic that is still left to oscilloscope investigation: switching waveforms. They are the result of the interaction between device characteristics, board layout and stray parameters. They have significant implications for one of the most difficult design challenges: EMI conformance to regulations. Available tools are still unable to bridge this gap.
AFTER DEVICE SELECTION: INVERTER DESIGN
The equipment designer has to insure that the power stage operates reliably within the thermal constraints of the application. The fact that he has a handful of IGBT candidates is a step in the right direction, but much work remains to be done to calculate power losses and heat sink size.
A number of design tools are commonly available to perform various design functions, including thermal simulation. Some are FEA based, capable of simulating almost any circuit, some are very narrow in scope, targeted to a specific topology and operated with a specific control strategy.
IR has deployed several such tools, some with FEA engines and some with extensive math engines. Of particular interest to IGBT users is a tool that calculates the operating condition of a power module in an induction motor drive, known as the IGBT IPM evaluation tool. It compares performance and efficiency of three different modules (Fig. 5) and provides a wealth of information, in graphic form and in tabular form, including transient temperature rise during overloads of short duration.
One limitation of this tool is that it assumes the backside of the module at a constant temperature. The next generation of tools overcomes this limitation by suggesting a heatsink suitable for the application and calculates the current capability of the inverter as a function of the heatsink size.
CONNECTING THE DOTS AND EXTRAPOLATING THEM
Looking back to 20 years of IGBT history a pattern emerges: as the silicon technology advances and increases in complexity, the product offerings become more numerous and more tailored to one application or the other. This, in turn, requires either a closer relationship between the IGBT manufacturer and the design engineer or specialized tools to support the designer. In a parallel development, a larger share of the design resources is being re-directed towards compliance issues, at the expenses of the power stage.
Continue on next page
THE SUCCESS OF IGBTs
THE MAGIC INGREDIENT for the success of IGBTs is a simple figure of merit: Amps per Dollar. Technological improvements and learning curve have conspired to drive this figure up to the point that it became an enabling factor for higher levels of functionalities, the most conspicuous example being the new generation of front-loading washers.
The 200-250V applications remain the hotly contested boundary between IGBTs and traditional FETs. The IGBT can deliver more Amps per dollar at lower frequencies, while the FET remains the device of choice over 100kHz. Technology has improved this key figure of merit for both devices and the choice is frequently determined by second order considerations, like packaging costs and efficiency at low current. IGBTs operate at higher current densities than FETs and they have a significant advantage in high current applications where the packaging component of cost is significant.
FETs, on the other hand, have an advantage in some battery-operated equipment where “overall” efficiency is important. Overall efficiency takes into account the fact that most of the time the equipment operates at reduced currents where FETs, with their characteristic square law of losses vs. current, have an edge.
Handling the minority carriers has, over the years, spawned a few technological innovations and a multitude of buzz-words that are peculiar to IGBTs - “punch-through”, “field-stop”, “NPT' and the like. Much has been written on this topic but, in the end, only two games can be played to deal with the issue of minority carriers: lifetime control and charge control. Long lifetime and high charge density lower the voltage drop across the IGBT but prolong recombination times - and vice versa.
Minority carriers are injected at the bottom of the die and recombine in the n- layer, where the stored charge resides (Fig. 6). One intuitive way to reduce stored charge without affecting their density is to reduce the volume where these charges reside. “Field-stop” and “punch-through” designs achieve this with a thinner n-layer. Since the n- must deplete to sustain the blocking voltage, it must be complemented by a “field stop” or “buffer” layer to insure blocking capability.
“Trench gate” designs reduce the stored charge by reducing the area of the die, without decreasing the charge density that is important to achieve a low voltage drop. A vertical - or “trench” gate increases the packing density of the cells and the injection efficiency of the majority carriers.
Once the structure of the device is settled, the lifetime control game can be played, with radiation or other techniques. Lifetime killing speeds up charge recombination in the n-layer and, in so doing, increases voltage drop while reducing the turn-off time.
All these different devices will behave in different ways in a specific application. Not only in terms of conduction vs. switching losses, but also in terms of switching waveforms, short-circuit capability, EMI, etc.
Download the story in pdf format here.