Internet Explorer is not a supported browser for TI.com. For the best experience, please use a different browser.
Video Player is loading.
Current Time 0:00
Duration 16:08
Loaded: 0.00%
Stream Type LIVE
Remaining Time 16:08
 
1x
  • Chapters
  • descriptions off, selected
  • en (Main), selected

This plot shows the effects of variation of resonant parameters LN, which is the ratio between the magnetizing inductance and leakage inductance, LN over LR, and the QE factor, which is the ratio between the square root of LR over CR over Lo. Over higher beginnings of them using some more [INAUDIBLE] LN, as shown here. When LN increase, the peak gain reduces. The gain at the selected operating conditions also reduce. With LN fixed, the resonant peak value is a function of QE, as QE increase, the [INAUDIBLE] corresponding resonant peak variable decreases, and again, at a selected operating condition also decreases.

Let us now look at some details, which the full bridge ARC operates. The nodes x and y are switched between plus v8 and minus v8 in any phase during two signals, using two signals OUTA and OUTB, each of which has a duty cycle 50% minus the daytime. This result is square root voltage appear across the resonant tank and transformer primary. Then as the voltage is applied to a resonant tank, its current is essentially [INAUDIBLE] wave.

A resonant-- the capacitor voltage across CR and the inductive voltage have the same amplitude and are interfaced. So the entire voltage appear across the transformer primary. This voltage also appears at the transformer secondary, scaled up by the 10 ratio and is redefined. The secondary current I_SEC is a redefined scaled version of the primary current I_PRIME. Energy has been transferred from the primary to the secondary during the whole interval.

During this time, the transformer magnetizing current is increasing and reaches a maximum at the end of this interval. At the end of the energy transfer cycle, OUTA turn off. while OUTB is still remaining off. The magnetizing current continues to flow, which charges the capacitance at nodes x and y, so that the voltage at x changes from plus vn to minus vn, and [INAUDIBLE] to minus vn as the body diode of QB turns on. And at the same time, voltage at y changes from minus vn to plus vn, where it, too, is [INAUDIBLE] by the body diode QB prime.

This ZVS transition happens during the daytime between OUTA going low and OUTB going high. When OUTB goes high, QB and QB prime are turned on. Current transfer from the body diode into the channel of the mosfet. And the next energies, its transfer intervals starts.

There are two energy transfer intervals, one which OUTA is high and the other when OUTB is high are identical, except that the voltage of the transformer primary is reversed, and the secondary current is in the other half of the center tapped finding. Similarly, the two ZVS transitions, MAG, ZVS_1, and ZVS_2 are the same, except that the direction of the voltage swings and nodes x and node y is reversed. When the switching frequency is lower than the resonant frequency, the resonant current falls down to zero before the end of the interval, causing the energy transfer to cease even though the magnetizing current continues.

Operating below the resonant can still achieve ZVS driven by the magnetizing current in the same manner. The secondary current has become discontinuous and requires more secondary current in the resonance circuits to deliver the same amount of power to the load. One characteristic that should be noted is that the primary ZVS could be lost if the switching frequency becomes too low. When the switching frequency is higher than the resonant frequency, then the primary resonant current is interrupted before it drops to zero.

The input voltage plus the reflected upper voltage appear across the resonant inductor results in the step increase in the inductive voltage. The primary current [INAUDIBLE], and once it reaches the magnetizing current level, the voltage across the inductor drops back. The waveforms are slightly different, although the ZVS transition are still driven by the magnetizing current in the same manner. The primary and secondary current waveforms at the right hand side are the summary of the one showed in the previous three slides. What I want to point out here is that ZVS could be achieved both at below, at, and above resonant frequency.

Of course, the designer has to ensure that there is sufficient magnetizing currents to drive the ZVS transition, especially at a higher frequency where the peak magnetizing current is naturally lower than it will be at its lowest frequency. The second thing to note is that the rectifier and switch off at zero current when the switching frequency is at or below resonant. Zero current switching is lost at its higher frequency. Finally, I just wanted to point out, again, that the switching frequency is almost independent of load at resonance.

Some other features of the full-bridge LLC are listed here. First, we know that the time needed for the ZVS transition is a function of the magnetizing current. This current is an inverse function of the switching frequency, so we can expect that the ZVS transition would take longer as the switching frequency increases. By now, I hope that it is beginning to become clear that both phase-shift full-bridge and full-bridge LLC are well suited to uni-directional battery charger application. There are some references for the bi-directional operation for full-bridge LL. However, I don't think either of these topologies are very well suited to bi-direction operation.

So far I have covered the high-power battery charger applications, the characteristics of the batteries being charged, and the operation and characteristics of both the phase-shift full-bridge and full-bridge LLC topologies. I am now going to gather all this information together and move on to a side-by-side comparison of the two topologies. Here are some comparisons.

First, we have noted previously, both topologies can achieve ZVS with some constraint. ZVS is at low current is difficult on the passive to active leg of the phase-shift full-bridge and much more difficult as the switching frequency of the full-bridge LLC increases. Second, the full-bridge LLC has ZVS off the operating rectifiers if it is operating at, or at, below, resonant frequency. Finally, both topologies normally use a Burst Mode and light loads. This allows operation down to zero load current while preventing significant increases in switching losses.

The full-bridge LLC has an advantage here if it is operating at its resonant frequency. This is because the power drain is delivering energy to the secondary almost 100% of the time. The only time when it is not delivering energy to the secondary is during the short date-time intervals. The challenge to the designer is that operation at resonant is possible only at a single input to output voltage ratio, but the output voltage of a battery charger varies widely. Both topologies have ZVS characteristics, and the phase-shift full-bridge has a good overall efficiency as a result. The net result is that there is little difference between the two topologies in an actual application.

Parallel operation is a very useful system-level technique, and it is almost trivially easy in the phase-shift full-bridge. Paralleling is used to increase system-level power in manageable steps. For example, a 5kW system may be built from three individual modules of 1.7 kilowatt each, which is easier to handle. Or n+1 redundancy may be built in if the same 5 kW system were to build up from four modules each to a total capability of 6.7 kilowatts. If one of the four units fails, the remaining three units can take over the full load. Downtime is prevented, and system availability is increased as a result.

Paralleling also increases the system-level reliability, availability, and lifetime. Current sharing improves transient performance in the case of a module failure and also increases system-level reliability and lifetime, because the individual modules in the current shared system are not run at full load. It is possible to parallel multiple full-bridge LLC converters. The problem is that it is very difficult to prevent parallel current shared if the modules are to synchronize.

Parallel operation synchronization can be provided if the switching waveform of one of the modules-- let's call it a slave-- is pulse width modulated. This allows the slave to operate at the same frequency as the master while its upper voltage is regulated down by the PWM modulation. It can be appreciated it is difficult to implement this behavior at a system-level. And for instance, the tiduct9 reference design uses a C2000 digital processor to control the system.

Synchronization of phase-shift full-bridge power stage is simply a matter of running them off the same clock. Some controllers like the UCT28590 provide a dedicated sync pin for this purpose. The full-bridge LLC regulates its output by varying the switching frequency. Synchronizing multiple full-bridge LLC power states forces them all to run at the same frequency and prevents individual regulation of the multiple output voltages. The problem is that small differences in component endurances of the power state means that they will all generate slightly different voltages when run at the same frequency. [INAUDIBLE] multiple output voltages with small differences between them make it impossible to connect them in parallel. The multiple outputs may, of course, be regulated, as a group of-- by varying their common switching frequency.

I think there are some technical issues. I'll repeat this slide. Synchronization of the phase-shift full-bridge power stages is simply a matter of running off the same clock. Some controllers like the UCC28950-- the full-bridge LLC regulates its output by varying the switching frequency. Synchronizing multiple full-bridge LLC power states force the system all to run at the same frequency and prevents individual regulation. The problem is that small differences in component [INAUDIBLE] of the power stage, means that they all generate slightly different voltages when running at the same frequency.

Providing multiple output voltages with small differences between them makes it impossible to connect them in parallel. The multiple outputs may, of course, be regulated as a group by varying their common switching frequency. Synchronizing to a fixed switching frequency external clock is not possible, of course, because you wouldn't be able to regulate the outputs at all. In summary, synchronization and parallel operation with current sharing is easily achieved with the phase-shift full-bridge, but it's more difficult in the full-bridge LLC, especially if synchronization is required.