By Dr. Ing. Hans-Peter Hones
Transportation represents about 25% of the worldwide CO2 emissions. What was considered the visionary dreams of “green mavericks” a few years ago is now in the headlines. Hybrid and electric vehicles have entered the market.
The clamor for hybrid electric vehicles (HEVs) continues to rise. By 2013, forecasts show that HEVs will represent about 6% of the worldwide vehicle demand (excluding trucks).
On the other hand, this also means that 94% of all light vehicles will still run on “classical” combustion engines. Even hybrid cars run on (downsized) gasoline engines most of the time. This easily justifies the huge effort underway to improve the fuel efficiency of engines, including support systems like cooling and fuel supply.
The introduction of high pressure direct injection systems using a “common rail” for fuel distribution at more than 1000 bar was a major milestone in the evolution of diesel engines. Engine performance had improved, emissions were reduced, and the noise level dropped considerably.
Solenoid injectors allow three to five injections per cycle, which contributes to a controlled pressure distribution in the cylinder to achieve a smoother engine run. Figure 1 shows a simplified power stage of a solenoid type “common rail direct injection” (CRDI) system using “two banks” of injectors. While switches QH1A to QH2B regulate the injector current on the high side, lowside switches QL1 to QL6 select the individual cylinder.
The elevated booster voltage VS2 enforces the fast injector opening. After activating the injector with QH1A/2A, QH1B/2B take over the current switching to VS1 at batetery level, which is just high enough to hold the selected injector open.
To get higher performance, two “banks” of two independent power stages are used to create higher flexibility with regard to independent injection timing between the cylinders (Fig. 1, again). In compact and mid-size cars, single-bank systems are often utilised to help reduce costs.
However, the evolution continues. Replacing solenoid injectors with piezo technology does bring several significant advantages. For instance, piezo injectors are much faster, resulting in reduced dead times and enabling more accurate control of the injected fuel amount. The fast reaction also allows more injections per cycle (up to 15 and more), diminishing rapid pressure changes in the cylinder and nearly eliminating typical diesel noise.
At the begin of their volume production, piezo injectors experienced problems with injector aging and reliability issues of the ceramic stack. However, material optimization solved those issues, leading to today’s more mature piezo injectors.
As seen in a simplified semi-resonant architecture of a piezo direct injection module, the capacitive injector load and the serial choke are building a resonant tank (Fig. 2). For years now, this architecture has been in high-volume production.
Many systems on the market today also use the CRDI architecture for piezo injectors. These versions need more components, but are easier to control and less complex in terms of the software.
On just the component level, the major difference between solenoid and piezo direct injection systems is the significantly higher operating voltage. While solenoid systems run at up to 120V boost voltage (VS2), piezo systems need a boost level of 250V to 350V. For both architectures, Fairchild Semiconductor offers optimised power switches in trench and planar technology as well as rectifiers.
From an architectural point of view, gasoline and diesel engine control units (ECUs) are very similar. In gasoline engines, highpressure direct injection is rapidly replacing the old-fashioned port injector system.
The major difference between the ECUs for diesel and gasoline engines is the required output power of the injector driver stage and the operating software. The very precise control of the injected amount of fuel using a piezo direct injection allows a defined stratification of the fuel-air mixture in the cylinder. The result is a very lean burn at significantly reduced fuel consumption.
On the other hand it’s rather difficult to ignite a very lean air-fuel mixture. In addition, one runs the risk of incomplete, non-homogeneous combustion.
To overcome this challenge, the multi-spark ignition needs to be exploited, especially at low rpm or when the engine is cold. To address these challenges, Fairchild developed the EcoSPARK family of ignition IGBTs.
INTERFACING POWER: BEWARE OF TRANSIENTS
You can have the best MOSFET or IGBT, but what good is it without a proper interface between the control circuit and the switch? To address that issue, Fairchild created a family of high side and halfbridge drivers targeted for direct injection applications.
The FAN708x drivers are built in HDG4, a dedicated high-voltage process with a self-isolating structure. Using a proprietary patentpending noise-cancellation technique, these drivers are less noise sensitive than competing devices and allow negative transients of more than 10V.
In addition, delay times and thresholds are nearly temperature independent. This eliminates dutycycle variations in the application, which cause variations in the injected amount of fuel.
To avoid switching overlap of the power stages (“shoot-through”), the FAN7080 includes a programmable dead-time function. Shootthrough not only reduces efficiency and is a potential risk for the power switches, but can cause EMI.
However, the dead-time controls just the switching of the power stage in forward current direction. In addition, the reverse current through the body diodes of the MOSFETs must be taken into account.
Special care should be given to PCB layout. Experience shows that in many cases, when customers complain about unsatisfying reliability of the power stage and non-explainable driver failures, an improper PCB layout is the root cause.
Figure 3 shows an example of a H-bridge circuit driving an inductive load. The load current rises as long as Q2 and Q3 are turned on. The current decreases when Q1 and Q4 are turned on. The inductance forces the direction of the current flow.
During the dead time, current (IOFF) is flowing from ground through the body diodes of Q1 and Q4 into the supply. At that instant, when Q2 and Q3 turn on again after the dead time, a high current peak flows through the body diode of Q1 and the channel of Q3. This reverse-recovery current significantly stresses Q3, because it has to absorb the reverse-recovery transient at potential near VBIAS.
Assuming Q3 can tolerate these transients, still another risk emerges: If the reverse-recovery current snaps off, node A slams down to ground potential and may ring well below ground as a result of circuit parasitics. Such ringing can destroy the driver’s output stage, if it’s allowed to exceed the device’s absolute maximum ratings.
Observation of a proper layout with low-impedance ground planes, separate traces for driver and power ground, as well as the addition of a resistor in the gate return line will usually solve this problem. Good designs place the gate driver very close to the power switches with the gate traces shorter than 1 inch. When using separate boards for power stage and control circuit, gate lines NEVER should go through the connectors. In extreme cases, a snubber circuit may be required to prevent device overstress.
A significant contribution to fuel consumption is also related to engine support functions. By replacing mechanical components with electrical systems, efficiency could be further improved.
By using a brushless motor for engine cooling, the precise control and increased airflow make it possible to reduce the volume of the cooling fluid. Thus, the combustion engine’s operating temp is reached much more quickly during warm up, and it reduces the amplitude of temperature excursions. Shortened warm-up time also lowers emissions and leads to longer engine lifetime.
Another contribution comes from electrification of the cooling pump. Simply introducing optimised, speed-controlled fans can reduce the fuel consumption of a mid-size car by up to 3%.