Unshielded drum cores in close proximity

I don't suppose you can 'point' them in three orthogonal directions?

George h.

You could quickly set up an experiment to measure coupling as a function of distance. I've used unshielded drums to transfer isolated power and send data.

Unless you trim things, drum core oscilators will be pretty inaccurate in frequency, so even uncoupled it would be tough to enforce a 1% difference. At 1%, they would probably want to couple and lock. They will also couple into PCB copper, which will change L and Q. And pick up ambient H fields. Unshielded drums are awful.

Those CD75s will couple pretty well a couple of cm apart. Coupling does fall off radically with distance, cube or something, so an experiment is suggested.

Keep the two 1% guys as far apart as possible.

I suppose you could add parts to the circuit to null out the couplings. It might even be done with pcb trace loops. That would be interesting but labor intensive to work out.

A drum core on a little PCB with a battery, an amp, and a headphone jack would make a fun AC mag field snooper. Add a photodiode for more fun.

Pimpom wrote in news:BIzCD.273598$E81.55490 @fx13.ams1:

Absolutely should NOT be a problem. Those cores should fully contain their respective transformations/fluxes. One only sees those huge shields and shield clips on power switchers where the core is being pushed near saturation. And even thse likely leak less than the person adding the shield gave it credit for.

I doubt you would even have a lot of success even trying to sniff signals near them.

It should not be a problem.

Are you thinking of pot cores? I don't think that is what pimpom is using.

GH

I used an axial inductor as part of a power supply filter on a low noise circuit. Wrong! (made things worse.)

Are pot cores and toroids topologically equivalent?

George H.

George Herold wrote in news:709cb718-684e-4a68- snipped-for-privacy@googlegroups.com:

person

oh... my bad. I did assume he was talking about little pot cores.

Yeah, it could probably be done with two inductors in series, wired so their axial field goes in opposite directions. Arranged common-centroid, you could probably get the coupling to cancel, at least if the inductors in each pair were identical. For real parts, that might not help enough to matter.

Keeping those 1% oscillators from locking together will probably take heroic supply decoupling (2-pole capacitance multipliers suggested) and separate steel shields for each one, maybe with steel on the bottom of the board as well, connected with many magnetic steel screws to the top shield.

PCB copper doesn't shield magnetic fields in that frequency range very well.

Cheers

Phil Hobbs

Nearly. Toroids are a bit better because there's no flux crowding as there is in the centre of a pot core. You also have to be super careful with gapped pot cores that you don't have an odd half-turn, whereas with a toroid you can put the wire any way you like. The half-turn produces a field that goes up the side of the core, across the top and down the other side, avoiding the (gapped) centre post. It thus sees the full permeability of the ferrite, which can lead to all sorts of entertaining saturation effects.

Cheers

Phil Hobbs

Topology is tricky. A coffee cup is equivalent to a donut.

A pot core has a simple solenoid winding, and some ambient flux passes through it, especially in an ungapped core. A toroid wraps the windings around in a circle, so voltages induced into the various loops (mostly) cancel.

I've mounted power transformers such that their emitted 60 Hz fields cancelled at some important nearby location. A pair of drum cores, in series or in parallel, could be made fairly insensitive to an external low-gradient field.

Pimpom could use three drum cores such that one pair has zero magnetic coupling to a nearby single one.

Right it's not quite the same... but you can imagine a toroid with all the windings crowded to one small area. That looks like the central pillar of the pot core. Then the rest of the toroid is just returning the flux.. like the outside of a pot core.. but a little different.

George H.

That size of core is about 5mm pole spacing, vertical; if you have to place them near each other, consider shielding (2mm of aluminum, cup form, would be good) or perhaps doing PC board 'hinges' (really separate two sections so that they can be snapped off, and reassembled tab=in-slot fashion) that puts all three cores mutually at right angles.

The 'oscillator' circuit is self-energizing, not a PWM thing? And, these are really handling power? For a small-signal oscillator, there are inexpensive SM inductors with other geometries (and even ferrite shielding overall) that might be preferable.