[SOLVED] Resonant inductor for Phase-Shifted Full-Bridge SMPS

Hello All

I'm working on a new power supply design project. This is a Phase-Shifted Full-Bridge dc dc converter 2.1 kW , 400 V input (from PFC) ,170V output and 200kHz switching frequency seen by output inductor i.e. 100Khz seen by primary . I calculated the needed resonant inductance to be around 17uH. I wanna use a ETD49 with 33 to 24 turns ratio to implement the main transformer. I do not have experience designing transformers before, So my question is
1-Can I expect the transformer to provide 17uH needed resonant inductance for me? I mean can I expect the transformer manufacturer to build it for me so that it comes with 17uh leakage inductance seen from primary? If not how much leakage inductance typically I can expect the transformer to have? Is the any downside to design the transformer to comes with customized leakage inductance (cost/efficiency?)
2-Let's say I have to use an additional inductor as the resonant inductance. Can I find such inductor Off-the-shelf (17uH, 9 amp peak) Or should I consider custom design for this inductor? If both options are applicable what are the trade-offs?

Thanks

First of all, you are super ambitious to to have never designed a transformer before and now you are designing a phase modulated full bridge converter at 2.1kW!

1. In a FB phase shifted topology you don't want the resonant inductor in the transformer itself or the secondary ringing voltage will kill your output diodes. You want the resonant inductor to be external and be clamped to the rails via a pair of fast recovery diodes. This is commonly done with this topology and for good reason. This is a quasi resonant topology, not fully resonant.

2. I doubt you will find the required inductor off the shelf (and why anyway?). If you wind it yourself you have much more freedom with its value.

An ETD49 will probably be a little small at 2.1kW and 100kHz, but it's not impossible.
Try to reduce your leakage inductance as much as possible by interleaving the windings. Even then I would expect you to have serious ringing on the output diodes, so cater for some snubbers now.
Also, opt for current mode control as it gives you much more control, particularly under fault conditions.
Dick

RichardHead said:

An ETD49 will probably be a little small at 2.1kW and 100kHz, but it's not impossible.

Dick

Click to expand...

Great info. Thanks. yes. you're right. the EDT49 will give lots of copper loss at 2.1KW. However, we are designing this transformer for an audio application. In audio application the average power is 1/3 of full power. This converter will see 2.1kW only for 200ms once a while. That 's why we believe we can get way with smaller size wire.
By the way I'm not very ambitious. I do not make the transformer by myself. I just do the basic paper design like turns and losses and the third party will build it for us with more detailed consideration.

The 33% duty cycle load makes a huge difference to the design. You can basically design a 700W converter with a decent short term overload capability. Unfortunately Phase shifted FB converters normally cease soft switching at about 1/3 load so you may well find that average efficiency is not great. Anyway, if your audio amp is class AB then it's only 78% (from memory!) efficient anyway, at best.
Your transient response is going to have to be fast so don't just aim for 1ms when 100us can be achieved.

with 170v output i reckon you are better off with LLC converter....there is no overvoltage ringing on the output diodes of an LLC.....and LLC is not said to have the problems of the PSFB with reverse recovery of the pri fet intrisic diodes.

treez said:

with 170v output i reckon you are better off with LLC converter....there is no overvoltage ringing on the output diodes of an LLC.....and LLC is not said to have the problems of the PSFB with reverse recovery of the pri fet intrisic diodes.

Click to expand...

LLS at 2.1 kW power level means either lots of circulation current or weak output voltage regulation. I have very tight voltage regulation requirement that I believe at least in theory would be hard to achieve with LLC at 2.1kW power level. On the other hand making LLC transformer with Lm around 100uH or less means huge air gap and other side effect which comes with it. I haven't build any PSFB yet but it look like a common sense solution's at 2.1kW power level. It looks like I can get way with 400V ultrafast diodes if I add diode sunbber on primary (Db/Dc diodes in UCC28950 datasheet) plus RCD clamp at the output.

LLC will regulate according to your feedback network. You can do LLC with external "leakage" inductor".
With the LLC you can use 250V diodes.

When designed transformer for AHB topology, we tried to increase leakage inductance of transformer on ETD39. In construction N1 minimum leakage, that we can achieve is 30uH (with minimum separator width). In construction N2 maximum leakage, that we can achieve is 10uH (with 20 layers of tape). My advice - use external inductance for FSFB topology. Transformer with integrated resonant inductance (leakage inductance) has greater power dissipation (heat of trans and resonant inductor in one device)

I would seriously consider an LLC design. I haven't seen the price of controllers recently but they were always expensive. An LLC uses a simple VCO.
LLC can use lower voltage output diodes so lower Vf and dissipation.

ICE2HSO1G is i think a good LLC controller

I'm referring to the high cost of PSFB controllers. My last post was a little ambiguous.

treez said:

ICE2HSO1G is i think a good LLC controller

Click to expand...

My teammate is working on 1.2 kW LLC and he has lots of trouble to regulate the output voltage response and reacting to fast load change and at the same time bringing down the circulation current. Yes I believe LLC is the best option for below 1 Kw. And also I believe a full bridge LLC is a good option for 2.1 kW too but if fast response to load change is not required and regulation is not fist concern. In our case we have to deign the power supply that maintain regulation while load changes from zero to full load. On the other hand at 2.1 kW you need a magnetizing inductance not below 100uH and that means huge gap in core which comes with fringing effect and thermal issues.