spiral inductor question

[myem]

An spiral inductor has Self-Resonant Frequency.
I can see curve at higher frequency.
For example.
A spiral inductor SRF is 12GHz.
After 12GHz , what does the curve mean at higher frequecny ?

 

[jian]

Hi, myem: Basically, everything is distributed. Some components can be approximated as lumped elements when the frequency is low. A typical spiral is in fact far from an ideal spiral. However, designers still consider it as an inductor. Its frequency response is quite different from an idealized spiral. In short, you have to consider it as a distributed element when the frequency is high. Regards.

 

[goodboy_pl]

just like a shorted Transmition Line which has the first rsonance frequency at lamda/4, spiral inductors behave the same.

after resonance freq. the inductor converts to capacitor!
it is possible only in distributed circuits.


BEST!

 

[maxwellian]

myem:

I think you can think of it as somewhat similar to a growing length of transmissin line with a characteristic impedance that is higher than your system Zo. At short lengths, this will appear to be inductive. At longer lengths, it will appear capacitive, and then alternate as you make it longer and longer.

With a fixed length of transmission line, you see it alternate between inductive and capacitive as the frequency gets higher.

The spiral inductor behavior is more complicated than that, but I think this is a reasonable analogy to use to understand the behavior.

--Max

 

[popsy]

I agree with Max, the curve at higher frequency means that the inductor is no longer behving like the inductor its is like the capacitor. As a general rule of thumb, we use the inductor much below its self resonant frequency.

Popsy

 

[elek_fyc]

Any lumped practical (not theoretical) element is made of a resistor, a capacitive effect and an inductive one.
A spiral is an inductive tanker coupled to capcitive effecte (because of lines proximity) and a resistor (for electric losses). Syou spiral wil be represented by a self coupled to a capacitor (serial and/or parralel) and a resistor: this is a thanker model.
For all tanker there is a natural frequency that devide the behavior of the lumed element as "Inductor" or "Capacitive" effect. That is the meaning of your curve.

 

[rautio]

Hi Myem -- So, is it distributed or is it lumped?

Strictly speaking everything is distributed and nothing is lumped.

Now, if you are willing to tolerate a certain amount of error (you get to decide how much error), then suddenly some components over some frequency ranges can be modeled as lumped (even though they are really distributed).

I have been doing a fair amount of spiral inductor on Si modeling over the last couple years. Turns out that you can get pretty good (i.e., two curves very close together) lumped models of most spirals over the entire frequency range you have plotted, both above and below resonance.

Even a simple model, like a PI network, with a series inductor and two shunt capacitors works well. Of course, you have to add a resistor in series with the inductor to model metal loss. Then you also have to add a resistor in series with each capacitor to model substrate loss.

So, above resonance, is it a capacitor? Well, kind of. But why look at it as a single lumped component, when we can say that it is really a PI net of components at all frequencies you plotted, above and below resonance?

The big problem with lumped models is when you get to Q. Now skin effect becomes important and all the resistors change value with frequency. Until we get a sqrt(f) resistor in SPICE, we have to satisfy ourselves with constant value resistors, or perhaps Taylor series approximations using inductors and resistors. Could be a rewarding research area.