4

I want to get an accurate model of a real inductor in LTSpice. Therefor I used an Bode 100 Network Analyzer und measured its impedance in a range of 100 Hz to 20 MHz. My idea was to calculate the capacitance of the inductor from the resonant frequency of the inductor like this (rearranged formula from wikipedia for a real parallel resonant circuit):

$$ C=\dfrac{1}{L\omega_r^2+\dfrac{R_L^2}{L}}$$

However, I have not been that satisfied with the results, when using this model in LTSpice. The inductor will be used in a resonant circuit, whose resonant frequency is approximately at 7 MHz. Can this inductor be modelled more accurately?

Impedance of the inductor (measured in inductance and serial resistance, the narrow one is the serial resistance):

enter image description here

Comparison of the impedances in LTSpice and the Bode 100:

enter image description here

Null
  • 7,448
  • 17
  • 36
  • 48
TonyDublov
  • 169
  • 6
  • The standard model of an inductor that comes with LTSpice is very likely an ideal inductor in series with a resistor. That will not behave in the same way as your real world inductor. But you can build your own model from ideal components. Just search for "Inductor model" to find some examples. – Bimpelrekkie Feb 23 '21 at 08:46
  • LTSpice has parameters like parallel capacitance, series resistance etc. in its inductor model. I just don't know how to get the most accurate values out of the measured impedance of the real inductor. – TonyDublov Feb 23 '21 at 08:53
  • How close do you get if you calculate your series R and parallel C and plot your impedance in LTspice? – winny Feb 23 '21 at 09:18
  • @winny I have added a comparison. – TonyDublov Feb 23 '21 at 09:45
  • 3
    @TonyDublov The analysis clearly shows multiple resonances which are displayed as slight notches because they involve damping. This means that a simple RLC won't do, you'll have to add more RLC cells (there may be reflections, too). This is commonly done. At any rate, judging by the pictures, it looks like either you didn't specify `Rpar`, or you did but it needs a lower value (more damping). Here is a [quick hack](https://i.stack.imgur.com/rgCq9.png) (single cell). – a concerned citizen Feb 23 '21 at 10:15
  • Plot both in either lin-log or log-log? – winny Feb 23 '21 at 10:18
  • @aconcernedcitizen I indeed did not specify Rpar. This seems to do the trick and should be accurate enough. Thank you very much! – TonyDublov Feb 23 '21 at 10:25
  • @winny The comparison is log-log, isn't it? – TonyDublov Feb 23 '21 at 10:26
  • Oh! There was a third. Thanks! Is it the dip at 700 kHz you are concerned about? – winny Feb 23 '21 at 11:30
  • usually the analyzer software has tools for creating models… for these resonances you usually model them as RLC groups in parallel with the main component – Lorenzo Marcantonio Feb 23 '21 at 13:09
  • 3
    _"The inductor will be used in a resonant circuit, whose resonant frequency is approximately at 7 MHz. **Can this inductor be modelled more accurately?**"_ why would you need to? – Bruce Abbott Feb 23 '21 at 17:24

1 Answers1

3

The analysis clearly shows multiple resonances which are displayed as slight notches because they involve damping. This means that a simple RLC won't do, you'll have to add more RLC cells (there may be reflections, too). This is commonly done. At any rate, judging by the pictures, it looks like either you didn't specify Rpar, or you did but it needs a lower value (more damping). Here is a quick hack (single cell):

test

a concerned citizen
  • 21,167
  • 1
  • 20
  • 40