FB-LLC vs PSFB for converter with variable load (up to 1.5kW)

[PowerStarter_]

Hello,
I was wondering if FB-LLC (FullBridge LLC) or PSFB (Phase Shift FullBridge) is a better suited topology for a up 1.5kW converter (DC-bus 400V to DC-out 100V). The important part is that it should work over a wide load range (10W to 1.5W).

For my understanding is there the PSFB a better choice, but it radiates more EMI. The FB-LLC, which has lower EMI, is more complicated to develop for a wide load range. Based on what I've read I would say that the efficiency is about equal, when having a wide load range.

I'd like to hear your comments regarding the better choice.

[PowerStarter_]

I'd like to note that GaN based transistors are intended to be used for this project in order to reduce the losses during switching.

[Anshika_G · Accepted solution]

Hello,

Thank you for posting on Infineon Community.

Regarding your query, both topologies FB-LLC and PSFB are suitable for high-power applications. The deciding factor between these two depends on your application requirements.  Please have look at brief information about their differences which might help you to choose between the two:

1. Operating Frequency -  PSFB supports constant frequency operation while FB-LLC LLC can be made to operate very efficiently if it is operated at or close to its resonant frequency. The input to output voltage conversion ratio of the LLC is conventionally called the "gain," and the gain is a function of the switching frequency. This means that output voltage regulation is achieved by varying the switching frequency. This characteristic has important system-level consequences for synchronization and current sharing.

2. Synchronisation-  The motivation to synchronize is that with two switching converters and some nonlinearity between the two, you get mixing which gives energy at the sum and difference between the frequencies. The difference in the frequencies can also appear in the audible range.
Because phase-shifted full-bridge uses a constant switching frequency, it is relatively simple to implement synchronization. The same cannot be said for full-bridge LLC, as the switching frequency is variable.

3. Current Sharing - A digital controller is needed for parallel operation for current sharing in LLC but easy for PSFB.

4. ZVS - Both topologies operate at ZVS which reduce switching losses. 

5. Synchronous Rectification - Because of the fixed frequency and duty cycle control paradigm, synchronous rectification is quite simple with the phase-shifted full-bridge. This allows for a synchronous rectification solution to be implemented using simple gate drivers. The full-bridge LLC's secondary current form makes synchronous rectification a little more difficult. And a dedicated SR controller is typically needed.

6. EMI -  While dv/dt rates are similar between the two topologies , di/dt rates are higher for the phase-shifted full-bridge. The LLC transformer will have lower primary and secondary capacitance and should have lower common mode currents as a result. The phase-shifted full-bridge has a fixed frequency, making the EMI filter easier to design. The phase-shifted full-bridge is inherently more noisy. But no topology is noise-free.

7. Cost - A resonant capacitor is needed in the LLC topology for proper operation. A DC-blocking capacitor is needed for voltage mode control in the phase-shifted full-bridge but not needed for current mode control. The DC blocking capacitor in the phase-shifted full-bridge is typically larger and more expensive than the resonant capacitor of the full-bridge LLC.

To summarise: 

Phase-shifted full-bridge provides higher system-level benefits including synchronization and current sharing. A full-bridge LLC transformer will typically be larger than a phase-shifted full-bridge transformer. LLC produces less noise but has a variable frequency. LLC has lower noise generation but has a variable frequency. And the phase-shifted full-bridge has a wider operating range.

The phase-shifted full-bridge offers easier design and more system friendly features, while the full-bridge LLC offers best efficiency at resonance and generally better EMI performance. 
Point to be noted is design process for LLC is an iterative process, as you have also mentioned design optimization is difficult for LLC.

Hope it helps. Please let me know if you have further queries.

Regards,
Anshika

[PowerStarter_]

Thank you for this insight.
So it seems like the PSFB would be the better fit, due to the benefit of the wider operating range which is needed for a variable load. But I've read is that ZVS can only be achieved at high load condition. The  FB-LLC can achieve easy ZVS for all loads but is limited in operating range by the gain/frequency.
Conducted EMI shouldn't be a big problem at the primary side because it's using the dc-bus. So only the increased radiated EMI provides a challange.
Please correct me if I'm wrong.

[Anshika_G]

Hi,
Zero Voltage Switching is widely used in medium to high power applications. Sorry, but I haven't come across that it is achieved only in high load conditions.  But one thing to mention is that energy savings are significantly more under high operating conditions(such as input voltage, switching frequency).
On the basis of EMI source detection, EMI reduction measures can be taken.  Filtering is commonly used to solve conducted emissions problems, while shielding may be used to solve radiated problems. 

Regards,
Anshika

[PowerStarter_]

---

@Anshika_G wrote:

Zero Voltage Switching is widely used in medium to high power applications. Sorry, but I haven't come across that it is achieved only in high load conditions.

---

ZVS is not impossible but it requires a large enough inductor to achieve it at light load. So it's way easier at heavy load, that's why often shim inductors are being used in the design. (see below)
For LLC ZVS can be achieved by simply having a switching frequency higher than the resonance frequency.

---

Key points in a ZVS FS FB design are:

- dimensioning of the resonant inductance with reference to the minimum load at which ZVS is required and the output capacitance of the used MOSFETs
- setting of delay or “dead” time between the conduction of the switches on the same leg.
- dimensioning of proper external RG(11) for each switch, taking into account its impact on the commutation behavior
- setting of delay between primary and synchronous rectification MOSFETs conduction, in order to minimize body diode conduction on secondary side
- dimensioning the main transformer in order to guarantee duty cycle availability at any load and input voltage condition, taking into account the actual duty cycle window determined by the combination of the total primary inductance (and so the slope of primary current) with the chosen values of delay times and RG,ext(12)
- setting the transformer turn ratio in a way to minimize the reverse voltage peak on synchronous rectification MOSFETs, which allows to use the lowest possible VDS(13) range and so the lowest possible RDS(on)

(copied from Infineon-Application Note AN2013-03)

---

Regarding the energy savings and there for the efficiency  are you referring to the PSFB topology?

---

@Anshika_G wrote:

But one thing to mention is that energy savings are significantly more under high operating conditions(such as input voltage, switching frequency).

---

Best Regards

[Anshika_G]

Hi,
Yes, I was referring to ZVS-PSFB topology.  

Regards,
Anshika

[PowerStarter_]

So have I understood it correct, that for the LLC the load is bounded by the chosen Qmax? Meaning we only define a maximum heavy load (minimum resistance R=Z/Q, output resistance is proportional to equivalent resistance) and the LLC than operates between zero load (maximum resistance) and defined maximum load. The values of the LLC converter must be designed in a way that maximum load achieves than the maximum load defined by the desired output power R=U^2/P.

The PSFB on the other hand defines the maximum load only by the chosen output power R=U^2/P.

The problem with LLC is that it needs to use burst mode at light load to keep the frequency in range. But so does PSFB to keep ZVS.

Therefor both can tackle the variable load problem. And the load range is basically the same with zero load up to max specified load. Or is there a difference in the load range? It seems that it's easier to get this wide load range with PSFB.

The operating range only refers to the input voltage, which is for the LLC dependent on the gain it can provide which offers only a narrow range. The PSFB offers a wide range of input voltage, due to the gain depending on the phase shift. Meaning the LLC is a better choice for a single phase SMPS with a narrow dc-link voltage range.

Please let me know if my understanding is correct. @Anshika_G 

Ps.: I've found this neat table, where it states that the PSFB is more suited for high power applications.

[Anshika_G]

Hello,

Sorry for the delay in response.
I have attached the application notes for the Evaluation board comprising the topologies LLC and PSFB. Please have a look. They both can operate at range varying from medium to high power,  but efficiency results are better at high load conditions. These app notes are rated for same power and using same MOSFETs.  This is for an overview, choosing between the two topologies mainly depends on what you need in your application, and which features are on priority.

EVAL_3K3W_BIDI_PSFB 
EVAL_3K3W_LLC_HB 

Best Regards,
Anshika