Hi gang, We are using Coilcraft conical inductors for bias tees.
formatting link
There are a lot of them on the board, right next to each other. A rule of thumb I've used for inductors is to place them at 90 degrees from each other, this is supposed to reduce mutual inductance, assuming the inductors are the cylindrical coil type.
How do conical-wound coils interact? I think these are air-core although the units I have have some sort of plug that the coil is wound on, I don't know what it is.
I'm also guessing that coils placed on top and bottom of a standard 60 mils thick FR-4 PCB with two ground planes will still couple magnetically, right?
The conicals (originally designed by Piconics, who incidentally are cheaper) are filled with ferrite epoxy stuff. They are so tiny that I'd guess that coupling is minimal unless they are very close. Copper planes should block through-pcb coupling above, very rough guess, 50 KHz.
But why don't you just test a couple?
Bias tees, especially super-wideband ones, are interesting. The conicals are a good first inductor, albeit expensive, fragile, and hard to handle.
To minimize magnetic coupling, you want to end up with cancellation of fields going one way with those going the other way. The field around a cylindrical coil is symmetrical about any plane through its axis and also about a plane perpendicular to the axis through the center of the coil (except that the field on either side of that plane will be in opposite directions, which is where the cancellation comes in...). The conical coil will have the symmetry about planes through its axis, but not about a plane perpendicular to the axis.
To get a better "feel" for how the cancellation works, I connected a couple identical coils, one to each port on a network analyzer, one in a fixed position and the other on an insulating "wand" so I could move it around and see just how much cancellation I could get, and how sensitive it was to error in position versus separation. That was worthwhile for me. I suggest you do the same with a couple of the conical coils. One thing I especially got a better appreciation for was that, though you can get good magnetic cancellation, when the coils are close, you still have a capacitive coupling.
As for coils coupling magnetically when on opposite sides of a board with ground planes: consider Faraday's law of magnetic induction. There can be no time-varying magnetic field passing through a perfectly conducting plane. Any field that gets from one side to the other must be because it leaked around the edges, or because the copper is less than a perfect conductor. I have several home-made RF filters with coils in adjacent cavities constructed with copper-clad board on 5 of the 6 sides, but the 6th side is just open. These filters show high attenuation in the stop bands, as predicted by a model with no coupling between the coils. In those that are designed for it, I get 120dB and more stopband attenuation. I'm quite sure that would not be possible if there was significant coupling through the partitions between the cavities.
I see. I will try to set up an "appointment" with our high-speed test instruments.
Thanks. The problem is that I don't know (yet) how low the frequency is on the "DC" end of the coil. It's supposed to be DC that changes once in a while to another DC value. But it's entirely possible someone will want to modulate the "DC" end with sine waves of varying frequencies. Could be anything from the audio band upwards. Either way I'm already going into full paranoia mode by placing them all at 90 degrees of each other. Of course every other one will be in- line but hopefully the distance will be great enough. The board is quite small and these coils will go all the way to the edges. There is no plan for an enclosure yet.
Eh, tight schedule, no time for "play". I'm not very informed on magnetics. I'll have to finish the job at hand first and then go back to freshman-level physics...
Remember our consultant that recommended the staggered bypass decoupling caps to stagger the "resonances" in the caps? I asked him how come the AC coupling caps on bias tees, which are just plain ol crappy X7R 100 nano jobs, are good enough to pass DC-6 gigabits?
I'm still waiting for an answer as to why we aren't putting in 5 different cap values for that bias tee cap...
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.