Why does the Inductance change on an E-Core Leg when going from Small Copper Wire to Litz Wire?

In the first picture you can see 45 turns of standard 0.4 mm copper wire. The impedance analyzer (e4990) shows an inductance at 20 kHz of 11.5 mH. Which is inside the tolerance of around 12.8 mH +- 20%.

The impedance analyzer measures Ls and Rs, which is the equivalent series resistance and inductance of the impedance at a given frequency.

However, using the following litz wire coil:

Measured the same way on the same core:

Gives only around 8.2 mH. How is this even possible? I have gone through simulations, lots of models that could possibly cause this, asked some Phds but I can't answer such a huge deviation.

Here is what I already thought of:

Any capacitance can't be the cause of this, since for lower frequencies we see the same inductance value (also evident in circuit simulation).

The air between the core legs can cause huge deviations, but when it is pressed down with a clamp we get the same results.

Even if I missed a turn on the litz wire coil, it wouldn't cause such huge change (44^2/45^2).

Very high quality litz wire (3.05x3.05 mm 1280 x 0.071mm rupalit safety) -> impossible for shorts, which could have explained lower inductance.

Has anyone seen this before or are there papers on this?

Edit:

Here is the same measurement at home giving 11.17 mH.:

Now I added 45 turns of thinner copper wire but with huge gaps to increase leakage as requested by @tobalt:

As you can see leakage is increased by about 0.4 mH. Which in context to the original problem makes the litz wire measurement further implausible.

• That impedance analyzer should be doing well at that frequency and that value of inductance. The only thing I can think of- and I don't see why it changes the measurement- do you have good solid connections to all the Litz strands? What are the measured Rs values?
– Fred
Commented Mar 16 at 0:11
• Would the down-voter care to share the reasons for the downvote? Seems like a valid question and plenty of investigation before posting IMO
– Frog
Commented Mar 16 at 0:29
• I can't see anything in the first picture; could you submit larger pictures, or more closely cropped to show the subject? Commented Mar 16 at 1:17
• @Fred Good point, for some reason I turned a blind eye on the resistance measurement. I looked up the E4990 Help Page: "Equivalent series inductance Equivalent series resistance Ls - Rs". Then intuitively the Rs measurement of 5.2 Ohm seems wrong for the 0.4mm copper wire version. I will think about it for a minute. Commented Mar 16 at 3:02
• @SpehroPefhany yes you are correct. I'm mainly just talking about the series circuit. My argument is, that the equivalent inductance can never be lower if you account for parasitics at low frequencies (below resonance). Commented Mar 17 at 0:13