two drums

It's got to be right two inductors. Put two unrelated inductors close together, and you've got an embarrassing amount of cross-talk.

Looks like they are cheak unshielded types. But anyway, nice to get more indtance for free.

When you put them in series, the SRF increases a little bit also.

I would suggest you measure with a Bode 100 or equivalent, that crappy LCR does not give you much info.

Ushielded drums are an unfortunate necessity here. Their specs greatly exceed any shielded inductors that I can fit into the equivalent volume. These Murata parts are tall and skinny, further optimizing my use of PCB area.

There just aren't many inductors in the milliHenry range. All the cool shielded surface mount things seem to stop at 1 mH. Audio transformers are terrible as inductors.

Not an issue in this application. I'm simulating r+l loads like motors and solenoids, pretty slow stuff.

A jet engine FADEC might drive over 200 relays, soldenoids, and torque motors, DC on-off and PWM from maybe 5 to 30 KHz.

That's a classic, original AADE lc meter. I believe it in this range. No L meter that reports a single number is really right, because inductors are complex.

I cross-check L meters with a 50 ohm signal generator and a scope. That finds L, ESR, SRF. Or I TDR the small stuff.

Inductors are by far the worse parts we use.

Seems like a problem you're having is a problem other inductor users are having and thus a market opportunity for component manufacturers.

Have you talked to your Visahy rep about this? They claim to be able to modify designs to help customers like you.

Check ripple and transient load response (load one - measure the other).

If chanels not synchronized at the switching frequency, odd results might be seen at mixing frequency.

RL

The two inductors will be in the two legs of, basically, a floating power supply, to decouple it from whatever customer impedance. So I'll be running the exact same current through both inductors. If I get the polarities right, I get the bonus inductance and the far-field magnetic cancellation.

I don't want signal coupling to other channels on the same board, so the field cancellation benefit is squared. That overcomes some peoples' objections to unshielded drum cores.

The physics is kinda weird. The 5 mH per inductor uses the universe for its energy storage, but the bonus L is energy stored in ferrite, I guess. I just invented the semi-shielded inductor.

The far field won't cancel, unfortunately. Coupled inductors in series have a total inductance

Lseries = L1 + L2 +- 2M.

Since your inductance is going up, the current is going the same direction in both cores, so the in the low frequency limit, the B field contributions add everywhere.

Cheers

Phil Hobbs

If one core has a north pole up while the other is south, the flux makes a tight local loop. That makes L go up. It's approaching an all-ferrite gapless inductor.

I tested the theory with a probe coil. Along the centerline of the two vertical inductors, induced field is zero. The two 5 mH inductors in series add up to 16 mH in this configuration.

If the two series cores have current in the same direction, I get a bunch of field radiated and only about 8 mH net. It's kinda like winding all that wire onto one big drum core.

Ah, okay, I forgot that your inductors are actually vertical not horizontal. My bad.

Cheers

Phil Hobbs

Isn't this the difference between a dipole field and a quadrupole field, regardless of vertical or horizontal?

Joe

My electromagnetics physics skills are greatly enhanced by experiment.

My Fields teacher at Tulane was Japanese and nobody could understand anything he said, but he graded on the curve. Class average scores on quizzes ran in the low 20's.

It doesn't use the whole universe - just the bit in the immediate vicinity of the inductor. The bonus L comes from the field lines that go through the adjacent ferrite. A inductor wound onto a toroid can be completely shielded - but only if the winding is non-progressive. Something wound around a pot core is less perfectly shielded, so the semi-shielded inductor was invented a long time before people realised that toroidal cores were a good thing.

Which is to say that your grasp of the theory is poor. I didn't get remotely comfortable with it until I got exposed to the transformer equations in the Siemens ferrite data book, and years of hands-on understanding gelled with that. I wasn't able to persuade Win Hill to put them into the 3rd edition of "The Art of Electronics", so the years of hands-on work may be a crucial precursor.

Tulane isn't famous for the quality of its instruction, but I do get the impression that transformer and inductor theory is pretty badly taught everywhere.

Thermodynamics is another famously difficult subject to teach, but a least it is recognised to be swine to get across to the students

On 2025-03-13 17:30, Joe Gwinn wrote:> On Thu, 13 Mar 2025 17:12:08

-0400, Phil Hobbs > snipped-for-privacy@electrooptical.net wrote: >

A quadrupole is basically a current distribution whose total dipole moment is near zero, e.g. two closely-spaced antiparallel dipoles like John's.

Cheers

Phil Hobbs