Finished LLC Resonant Converter Project

[sean0118]

Hi everyone,

I've been meaning to post my final year university thesis project for a while now so here it is. This was the first and only prototype.






It's an LLC resonant converter (SMPS) for high voltage operation. It was designed to step an input voltage of 370-410VDC down to 12VDC. The main focus of this project was to prove it's compatibility with the current doubler rectifier (CDR). The CDR is an interesting circuit as it allows twice the current rating (at half the voltage). Under normal operation the transformer secondary winding is only conducting half of the output current. Here's a schematic:



The LLC resonant converter topology is particularly well suited to high voltage, high frequency applications. This is due to the current circulating in the resonant tank causing zero voltage switching of the main power MOSFETs. This drastically reduces the switching losses of each switch as the voltage Vds has fallen to 0 by the time the MOSFET begins to conduct. However this does not reduce the conduction (I^2*Rds) of the MOSFETs.

One difference with the LLC resonant converter is that it's a variable frequency controlled SMPS with a constant 0.5 duty cycle. Regulation can be achieved because the gain (Mg) of the resonant tank changes with the switching frequency. See the plot below for the relationship between switching frequency and the gain of this converter (Note: fn is normalised against an 80kHz frequency, so fn = 1.1 = 88kHz etc)



The plot above also shows one of the main dangers of the LLC resonant topology. If the switching frequency reduces too far the converter can cross over to a capacitive mode of operation. This region needs to be avoided at all costs as ZVS is lost and there's a very high chance of the converter being destroyed.

Also, the resonant tank requires Lr and Lm to have very specific inductances to achieve the required gains. The eagle eyed forum members might also notice that Lm has been implemented using the transformer magnetising inductance. So this project required hand winding inductor/transformer goodness:



And of course a PCB layout, not a single bodge wire was required. 



Here's a scope screen shot from testing that shows the halfbridge switching at 200V (blue), resonant tank current (purple) and output voltage (aqua). Of course 200V is not in the regulation range which is why the output is only 5V (fs is at the lowest switching limit of 50kHz which causes max gain).



Here's two more that demonstrate the frequency changing to achieve output voltage regulation for an input voltage of 400V and 450V. The halfbridge switching voltage is shown in purple while the current of the two CDR output inductors are shown in aqua/green. I only designed it to regulate for an 370-410V input range, however testing showed that in practise it regulated for an input of ~390-450V. I'm pretty confident that this was due to the inductor airgap being slightly off, however I never got to test this theory and I don't have a 400VDC power supply at home.




Thanks to nctnico, Richard Head, T3sl4co1l and other forum members for their advice last year regarding this project. 


Questions comments? 

[TerraHertz]

Nice! {applause}

Yes, a few questions.
So, what's the max output current at 12V, and efficiency?

If you meant to remove your name and student # (?) from the silkscreen in the PCB layout, do you realize it's still visible in the photos?

Who'd you get the board manufactured by, and what was the cost?

I don't suppose you have a list of all the texts and online references you used while designing this?

Are you going to publish the complete schematic, parts list and suppliers details?

Switchmodes are something I've never designed apart from a few extremely trivial cases, but I'm very interested in doing so in the future. So, much need for learning. I find resonant converters hard to visualize, and this would be fun to build and play with.

[sean0118]

Thanks TerraHertz 

I wasn't planning on removing my name from the pics but now that you mention it I guess it wouldn't hurt.

So, what's the max output current at 12V, and efficiency?

Well the official documentation shows it's a 96W converter with 8A output. However, in practise the output rectification diodes would be getting mighty toasty at ~50W. So I guess I'll always remember to double check my diode power dissipation equations before sending the PCB design away for fabrication. 

Also, it was only tested at 12W output and the efficiency wasn't tested. Due to the high voltages I could only test it under supervision and I couldn't organise a supervisor for long enough to run every test I wanted. It does really bug me though not knowing it's efficiency, it's the first test on the list for when I get a high voltage supply.   

Who'd you get the board manufactured by, and what was the cost?

I have no idea actually, the uni took care of it all. I submitted the PCB layout to the uni by a deadline and then I got it back some weeks later. It was sort of annoying actually as I wasn't allowed to just fabricate the PCb at uni or send away to a manufacturer of my choice.

I don't suppose you have a list of all the texts and online references you used while designing this?

Yeah there were a fair few, if you want I can PM you my thesis with the full list, I think the main ones were:

LLC:
1. (SLUP263) Texas Instruments, “Designing an LLC Resonant Half-Bridge Power Converter,” design review, in Texas Instruments Power Supply Design Seminar SEM1900, 2010. *If you can't find this online send me a PM as I have a copy saved*

2.Yang, F.C. Lee, A.J. Zhang, G. Huang, "LLC Resonant Converter for Front End DC/DC Conversion," in Seventeenth IEEE Applied Power Electronics Conference and Exposition, 2002, pp. 1108-1112.

3. G. Hsieh et al, “Design Considerations for LLC Series-Resonant Converter in Two-Resonant Regions,” in IEEE Power Electronics Specialists Conference, 2007, pp. 731-736.

4. B. Lu, W. Liu, Y. Liang, F.C. Lee, J.D. Van Wyk, "Optimal design methodology for LLC resonant converter," in Twenty-First IEEE Applied Power Electronics Conference and Exposition, 2006, pp. 533-
538.

Current Doubler Rectifier:
1. A. Ioinovici, “Derived Structures of DC-DC Converters,” in Power electronics and Energy Conversion Systems, Vol.1, John Wiley & Sons, 2013, pp. 705-721

Are you going to publish the complete schematic, parts list and suppliers details?

Yeah good point, I'll try and post that tomorrow along with some Matlab scripts.

Switchmodes are something I've never designed apart from a few extremely trivial cases, but I'm very interested in doing so in the future. So, much need for learning. I find resonant converters hard to visualize, and this would be fun to build and play with.

Yeah, I didn't go too far into the maths behind the resonant converter, I stopped once I could plot the gain vs frequency graph I showed earlier. I think deriving all the maths behind it would have taken me more than a year. 

[TerraHertz]

Well the official documentation shows it's a 96W converter with 8A output. However, in practise the output rectification diodes would be getting mighty toasty at ~50W. So I guess I'll always remember to double check my diode power dissipation equations before sending the PCB design away for fabrication. 

Heh. I wondered about those diode packs sitting there with no heatsinks. If you screwed even fairly small finned sinks to them, it would make a big difference.

Also, it was only tested at 12W output and the efficiency wasn't tested. Due to the high voltages I could only test it under supervision and I couldn't organise a supervisor for long enough to run every test I wanted. It does really bug me though not knowing it's efficiency, it's the first test on the list for when I get a high voltage supply.   

Well now you're free. To electrocute yourself on your own time.
Hmm. Now I come to think of it, I don't have any supplies in the 300 to 1000V range either.  I do have a couple of 0-300V 1.2A linear supplies, but I don't think it would be a good idea to run them in series.

I have no idea actually, the uni took care of it all. I submitted the PCB layout to the uni by a deadline and then I got it back some weeks later. It was sort of annoying actually as I wasn't allowed to just fabricate the PCb at uni or send away to a manufacturer of my choice.

Wow, that is annoying. I bet they had it done by some very high cost local company. Weren't you tempted to do one on your own in parallel?

I don't suppose you have a list of all the texts and online references you used while designing this?

Yeah there were a fair few, if you want I can PM you my thesis with the full list, I think the main ones were:

LLC:
1. (SLUP263) Texas Instruments, “Designing an LLC Resonant Half-Bridge Power Converter,” design review, in Texas Instruments Power Supply Design Seminar SEM1900, 2010. *If you can't find this online send me a PM as I have a copy saved*

This?
Designing an LLC Resonant Half-Bridge Power Converter - Hong Huang
http://www.21dianyuan.com/home/download.php?action=download&id=64104

2.Yang, F.C. Lee, A.J. Zhang, G. Huang, "LLC Resonant Converter for Front End DC/DC Conversion," in Seventeenth IEEE Applied Power Electronics Conference and Exposition, 2002, pp. 1108-1112.

3. G. Hsieh et al, “Design Considerations for LLC Series-Resonant Converter in Two-Resonant Regions,” in IEEE Power Electronics Specialists Conference, 2007, pp. 731-736.

4. B. Lu, W. Liu, Y. Liang, F.C. Lee, J.D. Van Wyk, "Optimal design methodology for LLC resonant converter," in Twenty-First IEEE Applied Power Electronics Conference and Exposition, 2006, pp. 533-
538.

Hmmm. Without actually going and looking, I expect those are all paywalled?

Current Doubler Rectifier:
1. A. Ioinovici, “Derived Structures of DC-DC Converters,” in Power electronics and Energy Conversion Systems, Vol.1, John Wiley & Sons, 2013, pp. 705-721

That's definitely going on my books to buy list. Alas, $158 is a bit expensive for me just now. Later.
http://www.bookfinder.com/search/?ac=sl&st=sl&ref=bf_s2_a1_t1_1&qi=vgWWfA9bwdGzlMWyDUSrDUKIX0Q_1433412560_1:210:958&bq=author%3Dadrian%2520ioinovici%26title%3Dpower%2520electronics%2520and%2520energy%2520conversion%2520systems%252C%2520fundamentals%2520and%2520hard%2Dswitching%2520converters

Amusing that the second hand copies are more expensive than new.

Yeah, I didn't go too far into the maths behind the resonant converter, I stopped once I could plot the gain vs frequency graph I sowed earlier. I think deriving all the maths behind it would have taken me more than a year. 

Or in my case, forever, ie not possible. Oh well.

[sean0118]

Heh. I wondered about those diode packs sitting there with no heatsinks. If you screwed even fairly small finned sinks to them, it would make a big difference.

Yeah, I've got some in the parts draw now, I should chuck them on.

Wow, that is annoying. I bet they had it done by some very high cost local company. Weren't you tempted to do one on your own in parallel?

Yeah, probably pretty expensive, I did hear it has some silver content and stuff. Really overkill for a SMPS board. I was tempted to make my own PCB so I could test it sooner but I also needed to design and wind the resonant inductor and transformer. The deadline was the real problem, I had to submit the PCB design hoping I had chosen the correct sized bobbins, turned out ok though.

This?
Designing an LLC Resonant Half-Bridge Power Converter - Hong Huang
http://www.21dianyuan.com/home/download.php?action=download&id=64104

That's the one, it didn't download properly when I tried before, but it's working now.

Hmmm. Without actually going and looking, I expect those are all paywalled?

Yeah, you can probably find the same info on free sites online, but I needed papers to reference. 

Current Doubler Rectifier:
1. A. Ioinovici, “Derived Structures of DC-DC Converters,” in Power electronics and Energy Conversion Systems, Vol.1, John Wiley & Sons, 2013, pp. 705-721

That's definitely going on my books to buy list. Alas, $158 is a bit expensive for me just now. Later.

Yeah I think it was pretty good, but I can't remember using it a heap. It's about the only book that has the CDR in it though which was nice.

[T3sl4co1l]

Hmm, neet.

Looks awfully huge for all of 12W, unless that was just what you were able to test it at.

I'd love to see transient response, supply/load range, and efficiency stats.

Tim

[sean0118]

Yeah, it should do up to 50W as is, or 100W with heatsinks on the rectification diodes. Really only tested it at 12W to check that it worked due to time constraints (But it didn't seem to heat up at all).

Also, I'm not sure I mentioned that it's switching frequency range is roughly 50kHz- 110kHz, so the magnetics are kind of large. I've seen LLCs that operate up to the 1MHz region which reduces the size a lot. Also, the inductor bobbin is way oversized, but my uni did have a whole box of that type so it made sense to use them. Oh, and I guess I could have used SMD, but I'm glad I didn't as I had to try a few different sizes of gate drive resistors to drive the MOSFETs a little slower.

 

Edit: Sorry I don't have test data for the transient response, supply/load range, and efficiency. However I did run PSIM simulations of line and load changes and it seemed very stable. Similarly, I designed the current doubler to work down to an output current of 0.4A (5%) and tested during simulation, but didn't have time to test in the prototype.