General > General Technical Chat
Paper converter - φ2 inverter operated in 27.12 MHz.
(1/1)
WaveletSea:
https://pelab.imass.nagoya-u.ac.jp/research-english/#study-7
https://twitter.com/YamamotoPENU/status/1394502809156866048/photo/2
https://twitter.com/YamamotoPENU/status/1395172565665132547

Did someone knows or understands how to calculate those PCB air inductor and how to cancel their one turn axial field?

Thanks.
Weston:
I work in a lab that does similar work.

I don't think those designs cancel the one turn axial field. If you create a two PCB inductors, stack them, and connect them in series, you can get the one turn axial field to cancel. But you cant really do that on a single PCB.

For the inductors shown the axial field probably contributes to a significant portion of the inductance. You can get an ok approximation of the total inductance as calculating it as the sum of the inductance of a single turn with the mean radius and the inductance of the smaller spiral.

To get accurate approximations of the inductance you need to simulate the inductors in some FEM software. Typical design flow would be to create a script to generate the inductor structure and then simulate it, iterating on the parameters until you get the values you need.

COMSOL is a common software package for this, but costs a lot. FEMM is a free magnetics modeling tool but its only 2D, so not very accurate for these structures. Fast Henry is a free tool I have used a bit that can be used for this: https://www.fastfieldsolvers.com/fasthenry2.htm

You may also want to look at Elmer, which is a free FEM tool that supports magnetics modeling. Its open source and I have seen some people do impressive stuff with it, but the user interface is pretty bad.
harerod:
I had a look at the uni website's English and Japanese versions. I couldn't find any information that might help you.
Why not contact them directly, I thought? Worst that could happen is that they are using a version of some Japanese design software with no English user interface. :)

Funny that the tweet would contain more information than the uni website (the schematic).

Things that I would by curious about: actual radiated emission and efficiency.

Searching for "ペーパ・コンバータ" (paper converter) told me that the designer is also active here: https://garage-nagoya.or.jp/

Reading his profile told us that his name is 山本 真義 (Masayoshi Yamamoto).
https://garage-nagoya.or.jp/wp/wp-content/uploads/2020/09/116de0ec0093af5300919bdab03e695f.pdf


His university profile:
https://profs.provost.nagoya-u.ac.jp/html/100009841_en.html

And one more:
https://www.researchgate.net/scientific-contributions/Masayoshi-Yamamoto-2055636409

There must have been an easier way to find that stuff. Hope I could help. Say "yoroshiku" from Germany, should you decide to contact him.
Edit: One could have just replied to his Tweet, right? Maybe I should get acquainted with that new-fangled social media stuff and start to think along those lines...

 :)
T3sl4co1l:
Note that the Q factor of those will be quite poor.  There's a lot of choke points (inner turn edges, vias) and the faces are broad, no stranding or whatever to avoid eddy currents (not that litz works very well at 27MHz, anyway).  That's on top of the highly unfavorable geometry -- the narrow aspect ratio leaves very little field volume, for all that resistance (a square cross section is preferable).

It could be improved simply by using a thicker board (1/8" say), which could even lead to some other optimizations like putting those relatively bulky ceramic caps down into the board -- probably not as directly embedded components (some things can actually be fabbed in like this, but I doubt components of that size can?), but it would be fine to rout pockets for them, and do something goofy with castellated holes to allow wave soldering after gluing them in, perhaps.  Maybe with some strain relief so they don't get torn apart / snapped by board flex, at even more expense to stray inductance of course (which is already lower due to not having a ground plane under the body -- connection can only be made edgewise off the side of the component).  Such a build would have the unique distinction of being essentially flat; well, the transistors and support components will still stick out, but those can be almost arbitrarily small if we allow die bonding say.

On the upside, Q generally goes as sqrt(F), so for the frequency being 100 times the usual switching range, it might go from a uselessly low Q, to a passable if inefficient (say 30s?) level.

On a related note, Murata makes a line of DC-DC converters with embedded ferrite core transformers, "wound" in much the same way.  They're made as a sandwich I think, gluing together three panels, or perhaps two with a milled pocket in one.  So there's no problem with prepreg squidging out around the embedded component, it's dead air space inside rigid materials.  Then the stack is drilled and plated, and there's your vias.

Tim
Navigation
Message Index
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod