Air core inductor not changing value with introduction of steel?

[rs20]

Hi,

This is probably a dumb question but here goes -- I constructed a basic LC tank using a 68nF capacitor, and a hand-wound inductor of 5 turns with a diameter of 18 to 20mm or so. This works out to 0.2-0.7 uH or so based on online calculators (values rough because it's roughly made). So this should resonate at about 1 MHz or so.

I used a scalar network analyser to test the tank, with a 2.7k resistor from the tracking gen feeding the LC tank, and the LC tank connect directly to the signal analyzer.

I see a peak at 960 kHz, as expected, and I can reduce the frequency of the peak by a factor of sqrt(2) by doubling the capacitance, again as expected.

However, when I poke steel screwdrivers (which stick to magnets) and other steel thing inside or nearby the inductor, I see no change in the frequency -- the amplitude of the peak drops, presumably due to eddy current losses, but I was expecting an inductance change and therefore a frequency change.

Unfortunately I have no actual ferrite material to test with, but I thought steel would modify the inductance by at least a measureable amount; measurable even by my crude signal analyzer setup (crude compared to building an oscillator and using a frequency counter, of course). What am I missing?

[T3sl4co1l]

The permeability of solid metal drops off rapidly with frequency.  Likely, the increase in permeability is balanced by the decrease in volume of the magnetic field.

Namely, the metal shields the magnetic field away from the center of its mass; a more graphic illustration of this would be a copper cylinder the same size as the screwdriver, which will raise the frequency relatively strongly without reducing Q very much.

A nonmagnetic piece, of similar resistance and dimension (e.g., stainless steel), will raise the frequency, while reducing Q about as much as magnetic steel does.

Tim

[Ian.M]

Try packing the end of a wide drinking straw with fine iron filings, bonded together with superglue.  Stuff a small ball of cotton wool down the straw to pack them against.  The other end should be a copper slug (or copper foil soldered in a small cylinder).  Now you've got an old-fashioned tuning wand.

[rs20]

Thanks Tim, you're a legend as always!

I can confirm that a rolled-up piece of aluminium foil (which I presume is loosely substitutable for a copper cylinder) increases the frequency, and am fascinated by the explanation you gave of why all this is occuring. Kinda spooky how the lack of change of frequency with steel is due to two opposing effects cancelling out!

Ian -- thanks for the suggestion! What's the copper part for exactly?

[Ian.M]

The copper (or high conductivity non-ferrous metal) end decreases the coil inductance, and the iron dust end increases it.    Apply test frequency and see which end brings the tank circuit closer to resonance.  If it was the copper end you need to spread the coil or bend out the last turn.  If it was  the iron dust end, squeeze the coils turns closer.   If both ends make it worse even when you approach with them gradually, you are spot-on.

[T3sl4co1l]

I do the same thing with handy ferrite rods (you can get them from Amidon if you can't salvage any) and bits of copper tube.  Handy for pokings.

For more subtle tuning, I also have a bits of FR-4, stripped of copper except for one square at the end.  When placed across a coil, this shunts some flux, having the same effect; meanwhile, the capacitance isn't increased very much, because of the thin insulating strip.

Tim

[rs20]

Thanks folks, I should have googled "tuning wand" before asking a silly question 

[vk6zgo]

Hi,

This is probably a dumb question but here goes -- I constructed a basic LC tank using a 68nF capacitor, and a hand-wound inductor of 5 turns with a diameter of 18 to 20mm or so. This works out to 0.2-0.7 uH or so based on online calculators (values rough because it's roughly made). So this should resonate at about 1 MHz or so.

I used a scalar network analyser to test the tank, with a 2.7k resistor from the tracking gen feeding the LC tank, and the LC tank connect directly to the signal analyzer.

I see a peak at 960 kHz, as expected, and I can reduce the frequency of the peak by a factor of sqrt(2) by doubling the capacitance, again as expected.

However, when I poke steel screwdrivers (which stick to magnets) and other steel thing inside or nearby the inductor, I see no change in the frequency -- the amplitude of the peak drops, presumably due to eddy current losses, but I was expecting an inductance change and therefore a frequency change.

Unfortunately I have no actual ferrite material to test with, but I thought steel would modify the inductance by at least a measureable amount; measurable even by my crude signal analyzer setup (crude compared to building an oscillator and using a frequency counter, of course). What am I missing?

Your values of L & C are seriously out of proportion, with a huge capacitor, & a tiny inductor.
Unfortunately, the formula doesn't take this into account, hence your result.

I'm surprised you can see a peak, as with these values of L & C, your circuit "Q" will be horrible.
Think about why this is so (hint:- Reactance is important).


Try redesigning your circuit with  a capacitance value of around 150pf, & inductance of 170uH.

Re: the "steel things"--- solid mild or hardened steel is extremely lossy at frequencies much lower
than 1MHz, hence laminated cores for Mains transformers.

The extra loss should mainly affect the circuit "Q", but since that will already be abysmal, any change may not be apparent.

[rs20]

Your values of L & C are seriously out of proportion, with a huge capacitor, & a tiny inductor.
Unfortunately, the formula doesn't take this into account, hence your result.

Thanks for the advice -- I'm not actually intending to use the particular LC combination in a real circuit, the C will really be around 22 pF and not actually operating in resonance in the real application. The high C value was just what I had lying around, to test the inductor value.

[T3sl4co1l]

Your values of L & C are seriously out of proportion, with a huge capacitor, & a tiny inductor.
Unfortunately, the formula doesn't take this into account, hence your result.

I'm surprised you can see a peak, as with these values of L & C, your circuit "Q" will be horrible.
Think about why this is so (hint:- Reactance is important).

How so?

Empirical result must always trump ones' mental conception of a situation.  Are you doubting his measurement?

(Notice I say "mental conception", NOT "theory".  There is nothing in E&M that cannot be theoretically conceived of, analytically or computationally modeled, and verified in the laboratory, with high accuracy instruments proving the validity of the model.  For a theoretical model of arbitrary accuracy, one need analyze only finitely many possible "sneak paths", or orders of approximation, or other subtle details about a system.

That we should be so lucky, as to work within a field that is complete!

It is the sole responsibility of the design engineer to understand how many degrees one must go to, to reach a suitable level of validity in that model -- in other words, to correct ones' mental model to the appropriate theoretical ideal. )

So with that said; think about why that is so; reactance is important.

Tim

[MrAl]

Hi,

There's also the actual BH curve to consider.  Note that most BH curves are not straight lines, but have a stretched out "S" shape to them, and are curved mostly at the start and at the end.  The curve near the end represents the saturation characteristic, while the curve near the start represents the initial bias characteristic.
This starting curve starts out with very low slope for many types of cores, and so the permeability may actually equal air until a certain amount of current flows in the coil wire.  With an increase in coil current, we'd see an increase in permeability.  Since H is based on the ampere turns, adding more turns should show a different result also but it may take more turns than you would like to have to add to see this effect.