13
\$\begingroup\$

This question is related my my previous question: What to do about very hot planar inductor?

What I'm trying to do

I am trying to make a planar inductor (made from the tracks of the PCB, and surrounded by a 2 part a ferrite core). According to the datasheet of the ferrite core, the AL value is 1700nH, which means that with 12 turns around the core, I should get 12x12x1.7 = 244uH.

The problem

However, when I measure the inductance on my LC meter, it reads only 1.8uH. What's stranger is that if I create an inductor with the same core, but using stranded wire, and only 10 turns, I get 46uH!

I measured the resistance of the PCB tracks, and it's 0.25R, just as it should be, so I don't think there's a short in there.

Planar vs wound inductor

Planar inductor layers

My questions

What's going on? Is the AL value not sufficient for calculating inductance? How could it be that fewer turns gives a much higher inductance? Is the stranded wire massively better than PCB trace?

\$\endgroup\$
  • 1
    \$\begingroup\$ @Samuel they look OK to me! \$\endgroup\$ – Andy aka Nov 20 '15 at 19:10
  • 1
    \$\begingroup\$ the slightest gap between mating faces of the core will collapse the inductance value. Are you sure the faces are meeting, and not being fouled by the PCB? \$\endgroup\$ – Neil_UK Nov 20 '15 at 20:22
  • 1
    \$\begingroup\$ Yeah they look cool and exactly how I did it on a regular transformer. Shorted turn is my guess. \$\endgroup\$ – Andy aka Nov 20 '15 at 20:23
  • 5
    \$\begingroup\$ If the inside of the hole is plated, there's your shorted turn. Get the scalpel out! Doesn't matter if it's electrically isolated, it's still tightly coupled via the core. So you're measuring the leakage inductance at 1.8 uH. \$\endgroup\$ – Brian Drummond Nov 20 '15 at 20:58
  • 2
    \$\begingroup\$ I think @BrianDrummond nailed it. Whether internal holes are plated or not seems to be a bit of a crap shoot. Some use copper on either side of the edges as a marker for 'should be plated', but you have a definite gap there. All you need to do is to break the plating at one spot so it isn't a complete turn. \$\endgroup\$ – Spehro Pefhany Nov 20 '15 at 21:09
7
\$\begingroup\$

Brian has nailed it. Good shout. @Brian please take this picture and make your own answer. I'll delete this answer as soon as: -

enter image description here

NO VOTES PLEASE

\$\endgroup\$
  • \$\begingroup\$ No, shorted turn was your suggestion in the first place. (Upvoted) And I'm expecting there'll still be some discrepancy as with the hand winding, (gap?) so we may not be quite done yet. \$\endgroup\$ – Brian Drummond Nov 20 '15 at 21:38
  • 1
    \$\begingroup\$ @BrianDrummond You are a gentleman. Looking at the data sheet tells us that N97 material in that core size is indeed 1800nH for 1 turn so that's looking suspiciously significant. Using the wire should produce 180uH but it all depends on the measurement method - if it's being overdriven then saturation could easily decimate that inductance. \$\endgroup\$ – Andy aka Nov 20 '15 at 21:43
  • 1
    \$\begingroup\$ Fits well with his original question where he wondered why that thing got hot \$\endgroup\$ – PlasmaHH Nov 20 '15 at 22:15
  • 1
    \$\begingroup\$ You could always mark it as community wiki instead of "no votes please"... that way it can be upvoted (as it should because it's the correct answer)... but since you want Brian to [rightfully] get the credit. \$\endgroup\$ – Fizz Nov 20 '15 at 23:07
  • \$\begingroup\$ @RespawnedFluff how does one do that? \$\endgroup\$ – Andy aka Nov 20 '15 at 23:09
0
\$\begingroup\$

If the stack height of your pcbs is such that the core is not completely mated then this will dramatically lower your inductance.

\$\endgroup\$
  • \$\begingroup\$ The single PCB here is 1.6mm, whereas there's 3.5mm of space inside the cores. \$\endgroup\$ – Rocketmagnet Nov 20 '15 at 20:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.