Puente completo con cambio de fase (PSFB) vs. LLC de puente completo (FB-LLC) para una conversión CC/CC de alta potencia

Hello, my name is Colin Gillmor, and I am an applications engineer with Texas Instruments based in Cork in Ireland. So welcome to this talk in which I will compare the phase-shifted full bridge and full bridge-LLC topologies, especially for high-power DC/DC conversion applications.

The conventional wisdom is that the phase-shifted full bridge topology is the best choice for high-power DC/DC converter applications. However, this conventional wisdom has been challenged recently by some customers who are choosing the full bridge-LLC topology for applications where we would normally suggest the PSFB.

What I want to do here is provide a clear comparison between the topologies and to try to indicate which is best for a given application. Of course, TI is well placed to offer advice here because we have both phase-shifted full bridge and LLC controllers in our portfolios.

This talk is intended to be a comparison between the phase-shifted full bridge and full bridge-LCC topologies. These two topologies operate in very different ways, but both have some desirable features. They are also similar in some regards, component count, for example.

Both of these topologies are intended for high-power applications, really anything greater than about 1 kilowatt. The main applications we encounter in this power range are battery chargers for electrical vehicle operations. The difficulty, of course, lies in deciding which of them is best for a given application.

Now, the first step in deciding which topology is best suited is to look at the application. And that is what I want to do in this first section. Once we understand the application, the next step is to look at how the topologies operate. This is covered in part 2 of this talk, which will be quite detailed. This will allow us to make sense of the strengths and weaknesses of the two topologies and make comparisons and take conclusions, which are the subject of part 3.

This final section will also include some reference material, some suggestions for further reading, links for TI components and reference designs. As an aside, please note that much of what I say about the full bridge-LLC is also applicable to the lower-power half bridge-LLC.

The first thing I want to do is to describe the background against which these comparisons will be made because the suitability, or otherwise, of any topology depends heavily on the application. For example, a designer isn't going to use a flyback topology for a 1 kilowatt battery charger, nor are they going to use full-bridge topology for a 10-watt phone charger.

So here is a high-level block diagram of an electrical vehicle charging system. This takes energy from the AC grid and uses it to charge an onboard battery. We're going to consider two kinds of battery here-- low voltage lead-acid batteries operating at between 12 or 48 volts and lithium-ion batteries operating at around 400 volts.

There are various charger power levels. In the US, these are described as level 1, 2, and 3 by the Society of Automotive Engineers, the SAE. Their specifications refer, of course, to charges connected to the US grid. The grid source for level 1 charges will normally be a domestic socket in the user's home, for example. The power drawn from the socket will be limited to about 1.92 kilowatts from a normal US grounded receptacle. And this works pretty well for overnight charging.

Level 2 refers to a connection to a single-phase dedicated 208/240 volts AC connection. Normally, current is limited at 32 amps. This means that level 2 chargers operate at up to about 19.2 kilowatts. Charging time varies, of course. But it would normally be in the range of one to four hours.

Level 1 and level 2 have an AC connection from the charging station to the vehicle being charged. And the power level is relatively modest. This is the power level where the choice of topology between the PSFB and FB-LLC has to be made.

Now, shorter charging times require higher-power charging. And to do this, level 3 uses a high-voltage DC link to achieve charging times in the range of 10 to 30 minutes. The PFC stage is now offboard. That is, it's not in the vehicle. And the DC link is up to about 600 volts. But really, this power level is outside the scope of this discussion.

The EVSE-- this is Electric Vehicle Service Equipment-- block is a full implementation of a J1772-compliant service equipment specification. It is physically outside the vehicle, and it's the electrical equivalent of the petrol pump. The link is just the cable from the charging station to the vehicle, fixed at the charging station end with a plug at the vehicle end.

It's not simply a cable, of course. The plug is specially designed to be safe and, in addition to the power transfer cables, contains proximity sensors and signaling cable.

The link is just a cable from the charging station to the vehicle, fixed at the charging station and with a plug at the vehicle end. It's not simply a cable, of course. The plug is specially designed to be safe and, in addition to the power transfer cables, contains a proximity sensor and signaling cable.

The link ensures that power is not applied until the plug is correctly inserted in the socket. It also provides ground fault current interruption, or also known as earth-leakage current breaker, and other safety functions. The proximity detector is used to prevent the car driving away while the cable is still connected. And the signaling connection allows communication between the external EVSE and the vehicle for building power transfer level, et cetera, et cetera.

Inside the vehicle is an AC/DC onboard charger. I am not going to mention it again, but this stage provides a power factor correction and a high output voltage, typically around 385 volts, which I'm going to round up to 400 volts for convenience. The higher-power charging systems use a DC link. And the AC to DC conversion is, of course, replaced by a DC/DC conversion.

The battery being charged depends on the type of vehicle. In my old hybrid vehicles where there are no traction motors, there may be a high-capacity 12-volts or 48-volts lead-acid battery. Or if electric traction motors are used, then a 400-volts lithium-ion battery is more usual. Energy storage varies from about 3-kilowatt hours to 135-kilowatt hours, depending on the vehicle and whether it's a hybrid or full electric vehicle. The link tidub87 will bring you to a reference design for level 1 and level 2 EVSE.

Lithium-ion batteries were first proposed in the 1970s and developed into practical form in the 1980s. Today, they are in widespread use in consumer electronics, laptops, mobile phones, et cetera. Their main use in electric vehicles is as an energy source for the traction motors. In traction applications, several thousand individual 4.2-volt cells are arranged in a series of parallel circuits. The resulting battery pack has a typical nominal output voltage of around 400 volts and a capacity which varies from 18-kilowatts hours up to around 100-kilowatt hours.

I should note that management of the state of charge of individual cells in the battery to equalize charging among them and to work around so-called weak cells is necessary and complex. However, it's outside the scope of this presentation. So I'm not going to mention it further.

Lithium-ion batteries are much more sensitive to charging voltage than lead-acid types. And the battery charges must be able to hold very tight tolerances during the saturation and top-up charge phases. There is no float-charging phase associated with this chemistry. Instead, a periodic top-up charge is applied when the terminal voltage drops. Overcharging can, of course, damage the battery. And these batteries can be charged more quickly than lead-acid types, typically in two to three hours.

Now, there is a complex trade-off to be made between the charge rate, the charge time, and battery life. In fact, a fast charge can charge the battery up to about 70% in a short time although the time to 100% charge is extended so that the total charge time remains more or less constant.

In fact, charging to 100% can overstress the battery and reduce its lifetime. And the optimum charge level seems to be between about 70% and 85%. The price for this, of course, is less available energy in the battery and reduced driving range.

Lead-acid batteries have been in existence for over 150 years. And they've been the main type of rechargeable battery in use for most of that time. Despite its age, the technology remains popular, and it has been much improved since it was originally invented in 1859. It's still in widespread use in cars, for example, 12-volts and 48-volts types are common in EV applications. And they offer a reasonable energy density in the range of 33 to 42 watt-hours per kilogram, and they offer a good cost-performance ratio.

Charges for lead-acid batteries must operate in constant current and constant voltage modes. Initial charging is at a constant current, and the battery terminal voltage increases steadily to almost full voltage. At this point, the battery is about 70% charged. This part of the cycle takes about 40% of the total cycle time.

The charger then changes to constant voltage to supply the remaining 30% or so of charge. Once it's fully charged, the voltage is dropped to the so-called float level. And this compensates for the battery's tendency to self-discharge, and the voltage can be maintained indefinitely.

The constant voltage applied during the absorption-charge phase is a function of temperature and the precise battery chemistry involved. This means that the battery temperature should be monitored and the charger voltage adjusted accordingly. Lead-acid batteries should be checked fully charged when not in use. They cannot be fast charged, and a full charge takes about 12 to 16 hours.

Despite the fact that their chemistries are different, there are some common features in the charging profiles of lead-acid and lithium-ion batteries. For a start, the chargers must operate in both constant current and constant voltage modes, depending on the state of charge of the battery. Initial charging is done in constant current mode. Of course, the current must be set to match the battery capacity and the chemistry, but that's outside the scope of this discussion.

Normally, we would expect constant current operation in the range from a V-max, at which point the charger switches to constant voltage mode down to about 50% of V-max. If the battery is at voltages less than that, then, depending on the chemistry, it may be permanently damaged or need a special low-current initial charging current. In any case, we don't consider that operating region any further because in order to extend their operating lifetime, that's the number of charge and discharge cycles. Traction batteries in particular are not normally discharged fully.

During the CI charging interval, the voltage increases more or less linearly. This means, of course, that the power delivered by the charger increases during this interval. The CI charging interval is long compared to the thermal time constants in the charger, which means that it will operate in thermal equilibrium at all times.

Temperature rises in the charger depend on the power dissipated rather than the power delivered to the load. For example, a 3-kilowatt charger operating at 1.5 kilowatts and 90% efficiency will dissipate 166 watts. If it operates at 97% efficiency at 3 kilowatts, it will dissipate 92 watts and run significantly cooler at 3 kilowatts than at 1.5 kilowatts.

This is, of course, an extreme example. But it does emphasize the need to be mindful of efficiency of the entire load range. As we will see later on, this is one parameter where the PSFB has an advantage over the full bridge-LLC. But I don't want to get ahead of myself, so now let's turn our attention to the sort of typical specification that I'm going to use as a baseline for this comparison.

So having looked at the system level requirements, battery chemistries, and charging profiles, we can outline typical specifications for a high-powered battery charger. The regulation of the battery charger, whether for lead-acid or lithium-ion batteries, is quite a bit more complex than that of a fixed output voltage source. So it's worthwhile taking a quick look at how CI/CV regulation is achieved.

Charges vary in their detail design, of course. But all must operate in both constant current and constant voltage modes depending on the state of charge of the battery. Initial charging is done in constant current mode as a current set to match the battery capacity and chemistry.

Normally, we would expect CI operation until the battery voltage reaches its fully charged voltage, at which point the charger switches to CV mode. So constant voltage operation must be [INAUDIBLE] at all current levels, from the maximum where it enters CC operation, down to zero current where it is operating in float mode, or in top-up mode in lithium-ion.

The charger has to switch automatically between CV and CI regulation, like the typical characteristics shown at the bottom left here. There are two error amplifiers, one comparing the output current against the constant current regulation set point and the other comparing the output voltage against the constant voltage regulation set point. If the system is in CI mode, the voltage error amplifier is saturated high because VO is below the voltage regulation set point.

If the system is in CV mode, the current error amplifier saturates high because the current is below the current regulation set point. The outputs of the two error amplifiers are diode [INAUDIBLE], and the lower of the two errors is fed into the controller. Note that the error amplifier in the UCC28951-Q1 is configured as a voltage follower. And other than this, it plays no part in the control loop.

The onboard reference of the UCC28951-Q1 is not accurate enough for a lithium-ion battery charging in particular. So an external LM4132-Q1 reference is shown. Output voltage control for temperature compensation of the battery voltage and for setting the float voltage in the lead-acid battery charger is accomplished by a digital potentiometer controlled by an external micro-controller over an I-squared-C bus.

So here's a list of the main characteristics that we need to consider in the design of an onboard charger. Cost, hm, well, it goes without saying that cost is very important, especially for high-volume automotive applications. Now having said that, the most important aspect is the overall systems cost. For example, if we can spend some more money on the charger to make it efficient enough to eliminate the need for liquid cooling, then the systems-level cost saving can be significant.

Reliability and lifetime, well, the automotive operating environment is a very harsh one. And users, that is you and I, expect our cars to run reliably for years. Component count is often taken as a proxy for reliability. It isn't necessarily so, of course, but adding components unnecessarily is never a good idea. Adding components to protect the system against infrequent transient events, lightning strikes for example, will of course improve reliability and extend the system's lifetime.

Efficiency is often taken to be the most important characteristic of a power supply. Efficiency or, more precisely, as we shall see, the power dissipation in the power supply has a strong effect on the thermal design of a product. It affects the size, weight, and cost of the heat sinks needed and may dictate the need for liquid rather than air cooling.

Switching losses in the power switches have a direct effect on system efficiency, of course. For systems operating from a high-voltage input, this means that it is important to maintain zero-voltage switching across the entire operating range.

Component stresses feed into reliability and lifetime calculations. Generally, higher currents require more expensive components to carry them. Further, components are normally derated to provide a margin between the stresses experienced by the component and the ratings provided by the manufacturer.

Light load operation is important because many systems spend most of their lifetimes idling at no load or at very light loads. The CI/CV operation is required for all battery charging applications. We care about the voltage and current regulation and that the system has enough control authority or compliance to operate over the full CV and CI regulation range.

The decision of whether constant or variable frequency operation is preferred is to an extent a system-level property. Concerns include the generation of beat frequencies in the audio range and the effects of asynchronous noise on other equipment in the system.

For very high-power systems, it is not unusual to split the power handling part into multiple stages, each of which carries a proportion of the full system power. The benefits include flexibility to add extra power stages to increase system power and the possibility to include some redundancy to allow for the failure of one or more of the power stages. Of course, if we have multiple power stages, it's usual to have equal current sharing between them.

Finally, isolation is normally needed for safety reasons. It will be required irrespective of which topology we choose for the power stage. One important consideration is that providing the isolation barrier must not affect the overall performance of the system. We need to look for topologies that are tolerant of the increased leakage inductance, which is an unavoidable byproduct of the primary-secondary spacing needed for the isolation barrier. As we shall see, both the PSFB and FB-LLC perform well in this regard.

I just want to talk a little bit about efficiency and why we pay so much attention to getting from a 98% to a 99% solution or, perhaps more realistically, from a 96% solution to a 97% solution, like the 480-watt, 97%-efficient unit shown here. Now, at first glance, an efficiency of 99% isn't so much different from 98%. But this isn't the correct way to think about things. And it's better to consider it in terms of power loss.

So if we have a product giving 2 kilowatts at 99% efficiency, we have to design for 20 watts of loss. But the same product operating at 98% efficiency has 40 watts of loss, and we have to design for twice as much heat. At constant power, as the product gets smaller, surface area reduces and the temperature rises.

This is inevitable. And the best solution to reducing the temperature rise is to reduce losses because higher efficiency eases thermal design. And as I alluded to earlier, it may allow the replacement of expensive liquid cooling with lower-cost air cooling. Thank you very much.