Can someone explain how this Telcom 5-10KW Resonant DC-DC Converter works?

[DCPreamp]

Hi There,

I ran across this: http://www.pes.ee.ethz.ch/uploads/tx_ethpublications/biela_DCDC_INTELEC07.pdf
and was fascinated with such amazing power levels from a small SMPS. But upon reading it, and looking at the pictures of the completed unit, I was lost as to what/where the primaries and secondaries were as well as where the core is located. It appears to be just a big block of copper, yet has all the components forming a huge transformer. And, this 5 to 10KW beast of a supply fits in a 1U height chassis!!! (That's just 1.75 inches high) Really? How?

I've built some pretty huge switching supplies before, but never anything that looks like this beast! So I defer to other's on this forum with much greater knowledge than I to see if you can explain this supply in layman's terms so that even humble-ol' me can understand.

Thanks for your help on this. And certainly, anybody is welcome to add their comments on this. Maybe if we pick-it-apart enough with each person adding a little piece of the puzzle, we can still figure it out!

Please throw in your comments and ideas! And thanks everyone for making this such a great place for SMPS knowledge and projects!

P.S. I've attached a copy in case something happens to the link down-the-road.

 

[wally7856]

I had to study it for awhile but I think i have figured out this transformer . It is mostly smoke and mirrors. The fact is it is just an E-core transformer with copper foil winding’s. Because of the huge amount of power they are getting out of the core they needed a good way to keep it cool. So they made a 2 piece clam shell of thick copper to conduct the heat away. They even use a thermal conductive grease to transfer the heat from the core to the copper shell (page 7 table IV, good pic of shell).

They give the outside dimensions of the E-core in the same table. After playing with the numbers for awhile, as well as the 1U rack dimension of 1.75“ i have come to the conclusion that these are outside dimensions of the transformer clam shell including the winding height. That would give about a 1/16 inch clearance or so internally to the 1U case. 1U max allowed height 1.719“ - thickness of the case top and bottom measured from a 1U power supply i have at .04“ each = 1.639“ minus the height of the transformer 1.58“ = .059“ clearance.

Width 6.4 CM = 2.52“
Breadth 6.4 CM = 2.52“
Height 4.0 CM = 1.58“

The real story here is their trick/goofy winding placement. Look again at the same picture and notice the 2 slots on the top for the copper foil winding’s. Notice both slots are the same width, but one slot is wider than the other. The secondary goes down the wide slot and out the other and goes round and round the middle leg of the E-core as the first coil on the transformer. Then the tricky part. The primary copper foil winding goes down the wide slot and out the bottom and around the front side and down the wide slot again. To see that, look at the next page fig 11. You can also see the connections to the copper foil being brought out.

On page 6 fig 7 the 2 winding’s and one of the transformer legs is called out. If you zoom in on the picture to full screen you can see the secondary winding. The end leg that the primary crosses over has a gap in it, something i have never seen before. Usually the middle leg is gapped or the 2 end legs are gapped or all 3 are gapped. But not just one end leg.

Speaking of sneakiness, on page 6 fig 7, zoom in on the middle of the copper clam shell. Notice they used an insulator between the half’s to prevent a shorted turn.

The coupling between the primary and the secondary must be fairly poor and i assume a necessity of the series parallel resonant converter used.

As a wrap up i am not very impressed with this cooling scheme and i do not believe it would work in the real world. In normal operation ferrite gets hot and expands. This is why bobbins have at least .015“ clearance from the ferrite, and also why you are not supposed to dip the whole transformer in varnish if you want to insulate the winding’s. For their cooling method to work the ferrite would have to remain in direct contact with the copper and their ridged clam shell will not allow this. Thermal grease was never meant to fill in such a large gap and if you pored in such a large amount it would drip out of the winding slots. So in my opinion this does not pass the smell test.

 

[DCPreamp]

Outstanding review and explanation! Very interesting design, and also a very crappy write-up. Obviously the author wanted to brag about their design and accomplishments, while not divulging enough specifics about the design, the core used (although, pretty sure it's an E65, N87), winding specifics, etc. to allow someone to build another. Still, that is a pretty impressive amount of power for a relatively small envelope. Granted, it probably couldn't produce those huge power levels on a continuous basis, especially with your comments about clamping of the heatsink & the core's thermal expansion, but it probably wouldn't take too much more to get it to perform way better.

Anyway, thank you for your knowledge and insight into this product. I think it's this type of exploration, analysis, and explanation, plus DIY exploration that helps us all advance on this site!