how to avoid mutual coupling between inductors

as

I've

0.28. 39nH

I can't move them further apart, as there are other components on the board in the way. I would have to do a complete redesign of the board, which may be necessary in the end! not sure what you mean by criss-cross methods?

Phil

they are about 2mm away from each other. it is a 9th order elliptic filter. i can't get a photo i'm afraid. i've doctored the circuit quite a bit so the layout I have is out of date. had to fix the circuit so that the inductors were orthogonal to each other.

not really sure what else could be causing the filter to fail then. if i'm looking at the delay S11 plot, there is a peak for the resonance of

1 resonant pair, and when I add the other resonant in, there is a second peak, which the simulation says shouldn't be there, because both resonances are very close together one hides the other. the main resonance is at 390MHz (correct) where the odd one is at 500MHz.

Just a side question:

If I have inductors in the same orientation (ie facing same way) but in different horizontal plane (i.e. one above the other, looking down on the circuit) will it still couple in the same way if they were just next to each other?

Just a side question:

If I have inductors in the same orientation (ie facing same way) but in different horizontal plane (i.e. one above the other, looking down on the circuit) will it still couple in the same way if they were just next to each other?

I may have come up with a way to move the inductor right out of the way, but want to make sure that it won't still couple.

Cheers,

Hi.

That little or no effect, together with that value of k after such efforts, makes me skeptical of your diagnosis. If you have unbroken ground plane surrounding the parts, that alone should get the coupling lower than you indicate.

So, before getting too excited about magnetic coupling, I would revisit the method you used to deduce that number.

I presume these are air core inductors from the picture published by Coilcraft.

If the coil axes are coplanar with the PCB, and you rotate one coil around another (in the PCB plane) while maintaining their relative orientations, you can observe 4 coupling nulls per rotation. [1] When the coil axes are orthogonal, those nulls occur when either part falls on the axis of the other. If your coupling is truly as high as you state, you likely have got the parts placed well between those nulls. So, one approach would be to do some lab work and find better placements. For example, if the parts are placed with their axes parallel, the nulls are about midway between the coil axis and its perpendicular.

[1. This holds for normally wound inductors, formed as a single solenoid. More complex winding patterns can be used to create more nulls and to cause the field to fall off faster with distance.]

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Can you post to a web page, to alt.binaries.schematics.electronic or email a photo of your layout? Are these inductors over a ground plane?

I must say that your coupling is amazing. If I had to place two surface mount inductors and there was a requirement that they have a K = 0.28, I would be sweating to make sure that happened. I suspect there is some other explanation for the performance of your filter.

--
John Popelish

Hmmm... switch to charcoal toroids? (Sorry, couldn't resist!) :)

I read in sci.electronics.design that Phil Newman wrote (in ) about 'how to avoid mutual coupling between inductors', on Mon, 10 Jan 2005:

Yes, assuming they are solenoidal, but the distance between them is the 'slant' distance, in the (non-horizontal) plane that contains both inductors.

--
Regards, John Woodgate, OOO - Own Opinions Only. 
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

At 500 MHz, they almost never do. Parasitic effects are too large to ignore.

As Larry says, I would revisit your measurements. One thing that caught me out once was to ignore the transmission lines between components at these frequencies.

Regards Ian

A nineth order elliptic calls for a compartemented build, IMO.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

I don't really have the necessary experience or knowledge to do that!

I've moved the inductor, which didn't help to solve the problem, so having tested magetic sheilding, minimising coupling by making inductors orthogonal, moving them, increasing ground vias, the next solution I think is to start removing some of the substrate (making holes) between the tracks, and seeing if there is any propagation between tracks. Er = 4.7 for the PCB which isn't that high, but I think this is the next logical step. Everything else has been tested!! I'll keep you updated!

Phil

What about the Q factor? these filters have quite a sharp transition so need to have a fairly high Q, particulalry as these inductors resonate at the first and second nulls.

I've only been fiddling about with these filters for a couple of months!

Can you not make up your own toroids to those (admittedly small) values? Micrometals make some very small doughnuts...

--

"What is now proved was once only imagin'd." - William Blake, 1793.

I find that admission very surprising, given what you stated in your original post. It's really very simple to make up a couple of sub-minature toroids; the formulae are widely available (probably just a handful of turns of fine-gauge wire) and the nature of the doughnut form of inductor means that there is no field radiated to speak of; it's self-contained so you vastly lessen the risk of any mutual inductance arising.

--

"What is now proved was once only imagin'd." - William Blake, 1793.

This should assist considerably:

--

"What is now proved was once only imagin'd." - William Blake, 1793.

OK, this may not come out as I wish, but this is how I had the inductors... _ _ | | _ | | _

so they were all in one line. they didn't share a common ground so coupling was still an issue.

I've removed some of the substrate to no effect. now drilling the gaps between the pads, to see if that makes a difference!

NO, the drawing didn't work (for me anyway)! I'm working through google groups (the new beta version is very pretty, and much quicker!) but still lacking in picture placements.

I'm thinking that if this coupling mechanism isn't inductive but capacitative? I've magnetically sheileded it, attempted to reduce coupling, reduced the effective permittivitty by drilling holes (therefore lessening propagation and forcing the inductive path). what to do....!