LMG3410: Conventional PFC topology using GaN - Any advantages?

Hello Carlos,

I recommend you to check this online training video out at ti.com, which will answer most of your questions about TP PFC.

https://training.ti.com/designing-99-efficient-totem-pole-pfc-gan

I'll try to answer one by one.

"Both PFC applications for the LMG3410 I see on TI.com seem to be overly complex with either a bridgeless or freewheeling diode free approach."

"I understand these are to obtain the highest possible efficiency, but I was approached by a customer asking if there is anything wrong in using a single LMG3410 with a SiC rectifier and a traditional bridge rectifier for CCM operation."

There is nothing wrong using LMG3410 in a traditional boost PFC with SiC; however, as explained in the video above, you will not get the full benefits of GAN FETs. The efficiency will improve slightly but not a lot as mainly losses are on the diode bridge. At lower power levels <300W, it is definitely an option to use LMG3410 in a traditional PFC but as you go higher power, diode bridge losses become unbearable and decreases the efficiency. At this point, totem pole topology will surpass both efficiency and power density of any PFC topology.

"I can see the higher frequency being of benefit, with the option for phase interleaving at higher powers, but some questions arise:

• Given that the GaN device does not have a freewheeling diode, would this cause an issue? (I assume not, as the current is either circulated by the forward biased FET or the freewheeling diode, during reverse state

GAN does not have a freewheeling diode but can conduct similar to a diode in reverse direction (3rd quadrant operation) but with a large voltage drop. That's why we recommend synchronous switching to minimize the losses and improve the efficiency. But from operation point of view, it can operate just like a body diode - no problem.  

• Will there be any issues using a SiC rectifier instead of making a half bridge with the GaN device, for simplicity reasons?

I think I answered this above and video will show you some real numbers of the advantageous of GAN in totem pole structure. But as I said, it can be used with SiC rectifier at higher switching frequencies. The answer all depends on the power level.

• Why is the design based on a (rather standard) 100KHz switching frequency, and how high would we be able to go using the GaN device?"

The limiting factor for the switching speed is the total power losses. Higher fsw causes more losses, while increasing the power density. We have decided on the 100kHz as it provides a balanced efficiency and power density. Also, since EMI standards start from 150kHz, 100kHz helps with EMI stage as well. LMG3410 can operate up to 1MHz, and we do have a LLC design that operates at 950kHz. Please check PMP20637 for further info. 

"The customer concern is what would be the real world improvement made by switching to the GaN device on its own? The efficiency figures seem amazing, but they can easily be achievable by using standard bridgeless topologies, so they don't mean much."

The advantage of GAN w/ TP PFC at 1kW is numerically compared with classical PFC boost and dual boost semi-bridgless PFC topologies in the online training link given above. If you just replace Si MOSFET w/ GAN in a dual boost or classical boost PFC at high power levels (i.e. 1kW), you won't see a big benefit. But changing the topology from dual boost to totem pole, both the efficiency and power density can be greatly increased. Also, I have to say that totem pole topology has been proven to be the simplest (component count wise) and cost effective topology long time before, but because of half-bridge structure, it could not be used with Si MOSFETs due to large reverse recovery losses. GAN FETs become the enabling technology for this topology.

"Lastly, is there a reason why current sensing on PMP20873 is not implemented using our own CSA's, such as the AMC1301?"

There are a lot of TI reference designs that use AMC1301 for different applications; however, especially average current mode controlled PFC topologies in industry use hall effect sensors as high bandwidth feedback is not required. We did our reference design as close as possible to end application. But, it could absolutely have been used for this reference design as well.

Hope this answers all your questions and concerns.

Regards,

Serkan