Q: Do Coilcraft models include fixture mounting pad parasitics and lead parasitics?
A: When measuring the inductors on which the models are based, we use fixture compensation techniques to remove fixture parasitic effects. The specific details of measurement, calibration and fixture compensation are described in the model documentation. This allows you to customize your simulation to match your specific circuit board, ground plane, and mounting parameters. Since leads or terminations are an integral part of an inductor, they are included in the model.
Q: Should I use your lumped-element (SPICE) models or your s-parameters?
A: S-parameters are often easy to add to a schematic: A blank 2-port s-parameter element is placed in the schematic and the appropriate s-parameter file is linked to the element. However, s-parameters are not included in statistical (e.g., Monte Carlo) yield analyses, and the frequency sweep resolution (about 400 points over the frequency range of the model ) is set in the s-parameter file. Although the lumped-element (SPICE) models require slightly more effort to add to a schematic, they do allow for statistical analysis, and the frequency resolution can be increased in your simulation sweep setup. If a library or netlist is available, there is no need to recreate the schematic.
Q: Why are the Q values resulting from some of your models different from the published ratings?
A: An instrument used for measuring Q must measure a large reactance and a very small resistance at the same time. If the Q is 100, the resistance is 100 times smaller than the reactance, so a small change in the resistance measurement causes a large error in the Q measurement. Therefore, high Q factors are difficult to measure precisely. Our Q values can only be as repeatable as the measurements used to create them. This is why a typical Q is often specified on the data sheets.
Our published Q values and curves are based on the most accurate available method of measuring Q. This method uses an impedance analyzer and an accurate phase calibration standard. In order to provide deembedded models at very high frequencies, a network analyzer is used with a different fixture and a different set of calibration standards. Calculations of Q based on the s-parameters from network analyzer measurements produce greater error than the impedance analyzer method.
When creating our models, we use the impedance analyzer measurement as a guideline to verify the Q values. Because our models represent deembedded behavior, the Q will not always match the published values. The simulated Q resulting from the model changes, depending on the circuit board parameters entered into the simulation.
Q: The measured values of an inductor often depend on the measurement method and equipment. If the measurement results for the same component depend on how the part is measured, which method correlates the best with actual circuit performance?
A: The answer to this question depends on how well the simulation program mimics the entire circuit and its interactions with the surroundings, and the sensitivity to component tolerances of the final circuit.
If your circuit operates within a wide-band that is not very sensitive to component tolerances, the differences in measurement method may not affect the performance. If your circuit is a very sensitive narrow-band tuned circuit, one measurement method may correlate better with final circuit performance than another.
The ideal measurement method covers the frequency range of interest with sufficient resolution, and effectively removes all environmental and fixture effects through calibration and compensation. Coilcraft makes every attempt to come as close to this ideal as possible with the best available equipment. We also use correlation, when appropriate, as a standard of comparison to ensure that our manufactured component values are consistent with our specifications.
Q: The inductance in your lumped element model does not match your published rating, and the DCR rating does not match the series resistance in your model. Why don't they match?
A: Our inductor models are optimized to match the measurement of the component over the entire specified model frequency range. As a result, the individual model elements may not necessarily correlate with the separately-measured published parameters. The models should only be analyzed as a whole from input to output port.
Q: How can I predict the changes in inductance, Q and SRF when mounting your RF inductors over a specific PCB material or ground plane?
A: The effects of ground plane proximity depend on inductor geometry and test frequency, along with substrate thickness and dielectric constant. In many cases the effect is fairly small, but it can be significant in very tight-tolerance applications. The interaction of the magnetic field of an inductor with a metal ground plane is a complicated 3D E/M problem that requires careful measurement or simulation under specific application conditions. For this reason Coilcraft cannot quantify these effects in general.
Modelithics has measured many Coilcraft RF inductor series with a variety of different substrate types and thicknesses. They have created global models that scale the substrate-sensitive parasitics for accurate simulations. Their models may be viewed at: http://www.modelithics.com/mvp/coilcraft/.