Power Tips: How much can a LLC series resonant converter do?

After reading my last blog you may have considered placing the LLC series resonant converter (LLC-SRC) into your next power supply. LLC-SRCs are usually hiding behind another well regulated power stage, such as the example in Figure1. This isn’t ideal in terms of circuit cost because the BOM cost of a two-stage PSU is always higher than a single stage PSU. What you want is to replace the Flyback converter in your old designs with an alternative topology that provides higher efficiency and lower EMI and avoid increasing BOM cost. Is the LLC-SRC truly ruled out by the single stage and low cost topology selection criterion? 

Figure 1. Power stages of  a generic two-stage off-line power supply unit.

The LLC-SRC needs  another power stage in front of LLC-SRC due to its sensitivity to input voltage variation. Designers generally insert another power stage to provide a well-regulated input voltage for the LLC-SRC as it optimizes LLC-SRC efficiency. In general, LLC-SRC efficiency can be >94% in a two-stage PSU.

If you take a look at the PMP5141 design example in my last blog, you find that the LLC-SRC in PMP5141 can only operate within 280VDC to 400VDC input range. If we use PMP5141 LLC-SRC stage in a single stage AC-DC off-line PSU, we might only accept 240VAC+/-10% input. At this point, the LLC-SRC operational input voltage range now needs to be wider.

Higher Lm/Lr ratio leads to a narrower regulation range. Hence, if we lower the Lm/Lr ratio, the operational input voltage range has the capacity to be wider. PMP5141's ratio close to 5. If we reduce its Lm to make the Lm/Lr ratio 3, the minimum input voltage can be as low as 220VDC, as shown in Figure 2.  The LLC-SRC parameter is now optimized for a single stage PSU with 176VAC to 276VAC input range.

Figure 2. Voltage gain of an LLC-SRC with Lr=72µH, Cr=0.033µF, n=8.333, and Pout=240W.

The example above illustrates how lowering the Lm/Lr ratio making a single stage LLC-SRC PSU possible. A concern of reducing Lm is the smaller Lm makes the circulating current at primary side higher, which is a reduction of the converter efficiency. Keep in mind we are now a single stage PSU, we do not have the efficiency reduction on the power stage in front LLC-SRC as we had before.

PMP10283 is a single stage LLC-SRC PSU designed for 88VAC-132VAC or 176VAC-264VAC Input, +/-30V/200W output. When comparing the total efficiency of the two-stage PMP5141 and single stage PMP10283 PSUs in Figure 3, the performance of a single stage LLC-SRC PSU is significantly greater. Single stage LLC-SRC efficiency can be as good as a two-stage PSU and provide much lower Bom cost.

Figure 3. Total efficiency of PMP5141 and PMP10283.

PMP8762 – 100VAC-132VAC input, 12V/9A output is another design example of single stage LLC-SRC for US line. Keep an eye out for more single stage LLC-SRC TIDesigns in the upcoming months

Anonymous
  • Hi Rick,

    LLC-SRC is most beneficial for step-down applications because of its series connected resonant tank. I.e. your load is in series with the resonant inductor.

    You can still have LLC-SRC operate at n<1, one concern of LLC-SRC operates with n~1 turn ratio is that you might have higher parasitic capacitance on your transformer winding. With higher parasitic capacitance on transformer winding, it is easy to lose soft-switching at light load due to not enough energy to discharge MOSFET Coss. Especially in your application, your resonant inductances will be quite small, that will also make your converter easy to lose soft-switching at light load.

    Actually, double or triple your input voltage should be a good idea if you want to use LLC-SRC in your application, because you are now greatly reduced the current in the resonant tank and you can have higher inductances on your resonant inductors (including the transformer primary inductance). I have done two single stage LLC-SRC designs: one with full wave rectifier (PMP9739) and the other one with voltage doubler (PMP10283). I got high efficiency on the one with voltage doubler with smaller transformer.

    Regards,

    Sheng-Yang Yu

  • Hi,

    For typical application where input is 53VAC to 90AVC (60Vpk to 132Vpk) and output is 43VDC (300W design) , i am trying to design single stage LLC HF Bridge topology using UCC25600.

    After reading LLC related blogs and looking at input voltage variation I need to consider Lm/Lr low 3 ~ 4 (?). But input to output voltage conversion at low line input gives me transformer turns ratio n < 1 and at high line n > 1.

    What could be the situation if n<1 and input is 90VAC (132Vpk)?

    If I use voltage double input voltage range will be very high (144V) and need to compromise with efficiency.

    Can any one suggest a possible way ?

    Thanks

    Rick

  • Hi Antonio,

    Thanks for your comment. This is a good question and I believe many people have the same question in mind.

    If you go to IEEE Xplore and search "48V LLC", you can find some 48Vin LLC-SRC designs.

    However, LLC-SRC is benefit for its high efficiecny under high voltage transfer ratio condition. For the telecom 48V to 12V intermediate conversion, it is not always benefical in efficiency.

    In addition, the BOM cost of using LLC-SRC is likely to be much higher than Flyback.

    That might be the reason why you couldn't find many 48Vin LLC reference designs.

    Hi Yuri,

    Thanks for your comment. I think Active-clamp Forward might be a better choice for your application. Check http://www.ti.com/tool/PMP9127.

    Regards,

    Sheng-Yang Yu

  • Hi Antonio, last year I designed a 100W LED-driver with input: 24Vdc-50Vdc, using LLC converter, because isolation was required. Efficiency about 92% if Vin=36V and Vout=33V. Much lower power will not be cost efficient and very high power is difficult because of the resonant capacitors. Need to use metalized polypropylene caps, which can stand high enough ac current/voltage at 100kHz.  Ceramic is usually out of the question because of too high tolerance (if X5R or X7R).  So if you can find the components and not too big or expensive, it can be done !

  • Hi

    I'm interested to understand if such topology is also suitable for flyback replacement in telecom applications, where the nominal input voltage is a 48Vdc.

    I spent some time looking for LLC converter with such input,  but I was not able to find a single ref design.

    I suspect there will be a reason for that.

    Is there someone that could explain me why ?

    thanks

    ar