Inductive Coupling Coefficient

how long?

audio ferrites? low perm ferrites?

I don't know. What would be best? ~100kHz sinusoidal drive ...Jim Thompson

obviously, best orientation depends on length.

from memory rods usually are 4:1 up to 10:1 aspect ratio rarely longer, they break pretty easily. surprisingly the aspect ratio determines more the 'effective' permeability much more than the basic permeability, like with a 4:1 anything with more than 100 rel perm makes NO difference.

do you have any way to 'close' the magnetic path just a bit? Tuning makes a BIG difference! on getting a huge voltage out of it.

I can make some 'guesses' or spend some time to look at it in a bit more detail over night and get back to you with justifications for why do this versus do that.

Cross talk you can get some good estimates. If this is for a 'communication' at all; we should discuss that a bit too.

To see if what's feasible? Signal transfer, power transfer, mechanical reaction (are they mechanically fixed WRT each other)?

(Pretend you're a newbie that's been asked to less inadequately specify your problem).

Cores parallel? In line with each other? At right angles?

Is there anything else ferromagnetic nearby, for values of "nearby" ~5x coil separation?

SWAG between 0.1 and 0.25 or so.

Don't take this badly, but methinks you've spent too much time in SPICE simulations, and not enough just playing with hardware.

Careful, look what happened to the Galloping Gourmet... ;>)

Mark L. Fergerson

This is an RFQ for a slow speed (~100bps) data link "magnetically", with ~5mm gap.

I'm just wondering feasibility before sticking my neck out. ...Jim Thompson

IIRC, the Galloping Gourmet, after a bad car wreck, got religion, which ruined his image >:-} ...Jim Thompson

Data? That can't miss at 5 mm. I've used similar geometries to transfer enough power to run a microprocessor-based electric meter

Get an assortment of small unshielded drum-core inductors from Digikey and measure things. Drums have a nice geometry for axial coupling. Resonating both ends will help keep the power requirement down. 10 to

That's pretty much what I expected. But I also have a 5mm "above the PCB" maximum... do drum/pot cores come that small a diameter? ...Jim Thompson

Full-duplex? piece of cake! I've done that using air core PCB coils [no winding necessary] and operate over a distance of 8 inches using less than

no 5mm, 100bps full-duplex, piece of cake.

Big hint: communication between two entities should NEVER be treated like a radio transmitter/radio receiver pair. leave THAT for the broadcast industry where ONE talks to many. Rather THINK SYSTEM! where BOTH must communicate hand in hand, not simply throwing information out there hoping the other stumbles over it. Thinking system will get reliable cheap communication that even instantly flags you as the link starts to weaken!

Sorry, have to get offline for a bit, painting here.

Thanks! ...Jim Thompson

Me, too! IKEA bookcases ;-) ...Jim Thompson

Take a look at the unshielded ones, ME2200 or DO1608C. Coilcraft is great for sampling.

Stop it! you're going to optimize a weak design following this path. If you want to avoid FCC scrutiny stay below 10KHz. Don't even use ferrites, no real need. You aren't transferring energy, you're transferring 'information', where almost NO power still will tell you stuff.

ARRRGGG!

I've got a system in mind that will cost less than $1 each in volumes of

ferrites?! barbaric! plus weight, mounting, wrapping, vibration etc etc .... stay away from antique concepts, PLEASE!

I did something like that once. The two inductors were two pairs of three P CB-linked U-shaped lumps of wire, staking either side of a nylon-coated fer rite ring to a board. The broad-side spacing between the three-turn coils w as more than 5mm - the smallest ring I could get then was bigger than that.

might work - the EPCOS ring offered is epoxy coated.

Coupling was very good, but with three-turn windings the inductance wasn't high. I got enough volt.seconds to very reliably flip a comparator with hys terisis and could push arbitrarily slow square waves through it, up the the limits of the LM393 comparators involved.

The ferrite ring did bridge the isolation barrier, but it's insulation was good enough that it didn't matter

I'm guessing 50% or less.

That means the receiver has a zero input impedance preamp, by convention. 'No load' means no resistance to induced current, in the magnetism model.

Obviously, with 90 degree orientation, you could achieve zero coupling, so let's look at a maximum-couping case. I'll assume the core lengths are 1 cm. Axially aligned, field is inverse-square on two point monopoles, is the transmitter model. Positions 0 and 1cm , for the transmitter monopoles. The 'receiver' area is 5mm diameter at position 2cm (or you can model the receiver as two monopoles, too; you need that active area, and the ferrite defines that, for this approximation)

So, it captures flux into pi* (5/2)**2 area, of the total flux into (nearest monopole) sphere area (4*pi*(10**2) minus (farthest monopole) that into sphere area (4*pi*20**2)

So, coupling would be (3/4)( (5/2)**2)/400 ) = 0.012

You'll get way over one percent! No chance of losing the signal in noise.

Drat, I looked at the wrong distance squared: that "400" should be "100", and the coupling is more like five percent. Either way, more than good enough to get data through.