Haha, that paper is just dripping with inexperience; but it's exactly what I would expect given its origin: senior design project, undergrad level. You pick a topic, you're barely told anything about it, and told to get to work solving it!
The frequency range is entirely unspecified, and neither analysis not discussion of results is present. I suspect their results are equivalent to a helical transmission line through air -- those core memory cores (wow, blast from the past!) are specifically made with substantial hysteresis to yield the logic function necessary. Meaning they have very low initial permeability. Similar results I bet would've been had on an air core. GHz is also above the range where ferrite does much; they could've just looped transmission line through the air with similar results. Likewise, if they had considered which frequency range to prioritize, they could've chosen a more appropriate Microcircuits part; or at least tested several and seen how they respond, not just one and given up on the idea! (Granted, you don't get any budget either, so, buying from
Microcircuits is rather hard to justify.)
Even basic stuff, like the permeability test fixture, wasn't explored, or justified. Either it works or it doesn't; test a range of parts and compare results to the datasheet. Is it a systematic error or just not representative at all? How could it be changed to be more representative? How do professionals do it? (Well, obviously they just buy the fixture:
https://www.keysight.com/us/en/product/16454A/magnetic-material-test-fixture.html But, one might use that as a jumping-off point for appnotes and papers, and develop an understanding of how it's done.)
(I say this, having done a senior design project myself, looking back on it. So, part hindsight, but also anyone with experience can, I think, identify these traits and shortcomings of the paper.)
But I get the similarity; you're in a similar position, I suppose. Beware that you're balancing the familiar feeling of the search for knowledge, against the certainty of knowledge attained (but knowledge in the abstract, that is often ever so hard to communicate to those outside of that state of knowledge). That is, you identify with the author's inexperience, but you may not have much to learn from the author, or with the correct conclusions. (These are not peer-reviewed, remember; at best they're graded by a professor!) Conversely, it's difficult learning from direct sources, because you lack the stepping-stones of knowledge that they followed to attain that knowledge (or, even if available, those stepping-stones just don't produce the same understanding for you).
As for basic parts availability -- RS operates in .au, don't they? Well, maybe their shipping or minimum order is still >= $15, I don't know, but surely something domestic couple-days delivery is available. Offhand I see a half-dozen electronics suppliers that target .au (not necessarily
are in .au, like Digikey), though I don't know what of them are professional/wholesale versus individuals/one-offs. And if so, which ones have any selection of ferrite beads. But it seems like there ought to be something.
As for salvage: pulse transformers, common mode chokes, and EMI cores (including beads on cables) are usable here. CMCs and EMI cores kind of less so, because they tend to have either low-frequency priority (CMCs are generally more about getting high impedance for SMPS fundamental -- 100s kHz), or damping (EMI cores are lossy in the 10s to 100s MHz, and may not have all that much impedance or inductance overall for transformer purposes). But given more size, or number of turns, that's still doable.
Power cores (ferrite) are usable as well, but watch out for air gaps. Small gaps can be sanded down flat with a lapping plate (e.g. stick a sheet of SiC sandpaper to a sheet of glass or other flat surface, and grind with that; or if you happen to have a diamond sharpening stone, those are very nice), but large gaps, you'll end up running out of space around the bobbin, unless you want to build your own rather than salvage the one it came with. (Or you broke it in the process of salvage...)
Ferrite cores can be unglued by heating. Baking to ~200C is a good idea, softens the varnish. If you break the core, it can still be superglued back together (but beware the pieces might not fit exactly anymore, adding air gap; lapping the faces flat again helps maximize effective permeability).
Powder cores (usually bicolor toroids, but also blue or black, or other colors) have low mu and cutoff frequency, so generally aren't usable for pulse transformer duty.
Ethernet transformers are indeed in the pulse/CMC/EMI category so are fine here; as long as you don't need more winding area or flux capacity, heh.
Tim
Home
Help
Search
Login
Register
