Manjurul Ahsan Riheen Hard to say without seeing the model. What type of calibration are you doing in your physical measurements and is your model de embedded to the same point to mitigate any extra phase?

-Adam

You have higher inductance at higher frequencies. This is why the open point is at lower frequencies than in the simulation.

Try using a linearly variable inductance with frequency and then redo the simulation with this inductance

Dear Manjurul Ahsan Riheen ,

I would propose that you plot R(f) and X(f) for the input impedance of the coil.

Or equivalently you can plot mag. Z and phase phi as a function of frequency.

Because of the parasitic capacitance it is expected that you have a shunting capacitance to the inductor and therefore you will have effectively a parallel resonance circuit. So, in order to make the measured and simulated impedances agree with each other you have to take into consideration the stray capacitance.

Best wishes

Adam Weidling Thank you for your reply. I have used regular VNA calibration (open, short, load) using Calibration kit. I have shared the HFSS model of the coil placed on a polyimide substrate.

Valentin Bogatu Thank you for your reply. Your are right that my design has higher inductance at higher frequencies. I am using HFSS for simulation. I am not clear how I should try your instructions in HFSS. Can you please elaborate?

Abdelhalim abdelnaby Zekry Thank you for your reply. I have added the R(f) and X(f) plot of the Z with frequency. I understand stray capacitance controls the resonance frequency of the inductor. So far I thought that HFSS considers the stray capacitance of the design inside the simulation. Please correct me if I am wrong.

If there is any metal (or good conductor) nearby it will reduce the inductance, I think. The closer it is, the larger the effect. In simulation it is easy to have any metal at infinity, but there may be some close by for the measurement.

The magnetic field lines won't go through the metal at 50 MHz. If the magnetic field would have been significant where the metal is, then the change in inductance will be significant. Metal 100 microns thick will exclude 50 MHz magnetic fields.

While inductance measured as same, lower parallel resonance frequency hints for higher parasitic capacitance than in simulation. Can be due to connecting wires or proximity to dielectric / metal although proximity to metal would diminish inductance as Malcolm pointed.

Dear Manjurul Ahsan Riheen ,

It is very clear that you have a parallel resonance circuit as expected. The resonance frequency at zero reactance. The presence of the capacitance is very apparent. In order to obtain matching with the experiment you need to add a parasitic capacitance across the inductor. It may be just the header capacitance which is the capacitance between the connection pads pf the inductor.

It is a normal practice to find the measured resonance frequency is smaller than the simulated one.

Best wishes

I think it is from the S11 measurement method connecting the DUT to the VNA that increases apparently the parasitic capacitance (some pF in more).

I suggest you to use a wireless method, using a dual-loop probe in S21 measurement on VNA.

This dual-loop probe associate two inductive loops together with a position of zero mutual,

such a prode is described in "Quick measurement of NMR-coil sensitivity with dual-loop probe, L.Darasse and G.Kassab, Rev. Sci. Instrum. 64 (7), July 1993".

With such a method you should obtain a better value for the auto-resonante frequency.

See the flowing pictures to illustrate this ...

Abdelhalim abdelnaby Zekry Thank you very much for the clarification. Comments of Malcolm White and Nikolay Pavlov are also aligned with your thoughts. Thanks everyone.

Thierry Thomas Appreciate your recommendation on the alternate measurement method.

Abdelhalim abdelnaby Zekry Malcolm White Thierry Thomas Nikolay Pavlov

I added an additional shunt capacitance (parasitic) value that is giving me similar resonant frequency in simulation as real measurement. Should I consider this capacitance as measurement environment compensating capacitance?

Now, if I measure resonant frequency of other type of loops like rectangular or spiral of different sizes, will it be the same additional parasitic capacitance value that I should use in simulation?

I am thinking whether it is possible to come up with a common capacitance value which will compensate my measurement environment effect in simulation. Will it vary based on size and shape of the loop? I have shared some different loops pictures as well that I have to measure.

Dear Manjurul Ahsan Riheen ,

You write that you introduce a capacitance of constant value in order to give a good match between HFSS calculations and S11-VNA measurements, independently of design. It's curious.

Is it about 3 pF ? (In your first post you wrote a shift from 66.3MHz to 45MHz with a inductance of about 2µH.)

Have you checked that your used calibration kit (SOL) corresponds to declared/selected in the VNA ?

But if the kit-parameter used by the VNA is not corresponding to the used cal-kit, the errors should be not exceed one pF or an offset delay of several 10ps (équivalent of about one cm of coaxial cable).

Then what about you add to connect your coil in DUT to the VNA, do you add short threads no taken into account in your calibration ?

You wrote that your substrat is polyimide. Is a flexible circuit material without adhesive layer between copper layer and polyimide ?

For my designs in flexible circuit, I usually use a circuit material without adhesive layer like DuPont Pyralux AP (AP8525) taking epsilon-r to 3.32. In case of adhesive layers, the thicness of such layers is not negligible upon the polyimide thicness (more than 50% !), then the contribution of the adhesive is important (particulatly the loss are increased).

Bests regards

Dear Manjurul Ahsan Riheen ,

welcome again!

Yes formally all the parasitic capacitances can be formally modeled by a discrete capacitance connected at the terminals of the inductor. Physically where it comes from and its origins, one has to make a through comparison between the simulation structure and the measuring set up including the device under test and its measuring circuit. The analysis introduced by Thierry Thomas may be helpful.

Best wishes