| T O P I C R E V I E W |
| NikName |
Posted - Apr 08 2015 : 11:54:13
Dear Enrico,
I would be glad if you could provide some help concerning the further use of the computated FastHenry impedance matrix. Apologies for possible misunderstandings in physics, but this appears to be a real problem in my current situation..
FastHenry shall be used to characterize the behavior and losses of a simple one-port device (inductor) carrying a non-sinusodial periodic HF current with high AC ripple and DC bias. In my understanding, a larger AC ripple generates a greater value for the time-derivative current, resulting in higher parasitic resistance which eventually affects the DC bias currrent as well. My expectation of FastHenry is, that in contrast to datasheet values measured under small signal operating conditions , it should be able to capture such large signal effects and provide a much more sophisticated SPICE model for further loss calculations.
Although creating an input geometry, computing the Z matrix and even including the generated ROMs in LTSpice is working, I would like to understand the use of the Z matrix for loss calculation:
- For a generic input current, a fourier analysis will provide a DC offset and several sinusodial AC parts A_k*cos(w_k*t) of orders 1...k which should match the frequencies of the Z matrix. Why is the computed R in the Z matrix at a specific frequency (order k) not sensitive to the applied amplitude A_k at that order - in my understanding it should be to capture large-signal generated loss ? How do I add up the losses at each harmonic - via superposition of RMS values multiplied with the matching R? What about negative values for some harmonic amplitudes - use the absolute value?
- The parasitic resistances are generated solely by the AC part of the input current. What term/value describes the resistance/loss for the DC offset part of the input current? What resistance value is "seen" by the DC offset?
Thank you very much in advance for your support!
Best regards,
Nik
|
| 1 L A T E S T R E P L I E S (Newest First) |
| Enrico |
Posted - Apr 12 2015 : 23:44:50
quote:
- For a generic input current, a fourier analysis will provide a DC offset and several sinusodial AC parts A_k*cos(w_k*t) of orders 1...k which should match the frequencies of the Z matrix. Why is the computed R in the Z matrix at a specific frequency (order k) not sensitive to the applied amplitude A_k at that order - in my understanding it should be to capture large-signal generated loss ? How do I add up the losses at each harmonic - via superposition of RMS values multiplied with the matching R? What about negative values for some harmonic amplitudes - use the absolute value?
I'm not sure I entirely follow your reasoning. FastHenry calculates the impedance of your structure at a given frequency. The real part of the impedance is your resistance at that frequency, and the imaginary part is the inductance. You can repeat the calculation for a set of frequencies, and therefore have a characterization of the behavior of R and L over a range of frequencies.
Now if you want to perform a simulation, either you have a simulator that supports a model able to be specified using a set of discrete frequencies, of you can use a ROM model. This ROM model will approximate the impedance using a space state realization, so you will have a continuous function matching (up to a certain order) the behavior in frequency of the structure.
Now, using this model, you can have any input signal stimulating your structure, not only small signals. However, there is nothing changing the properties of the structure and therefore causing different effects if you have a large ripple w.r.t. a smaller one. If you were using a transformer, or a coil with a metallic core, then you would have some energy stored in the core, and this would affect in return your field behavior, giving rise to hysteresis phenomenons etc. However, FastHenry2 is not supporting any ferromagnetic material, so this is not the case. Nor we have any radiative effect, as the basic assumption is of quasistatic regime.
quote:
- The parasitic resistances are generated solely by the AC part of the input current. What term/value describes the resistance/loss for the DC offset part of the input current? What resistance value is "seen" by the DC offset?
This is maybe one of the misunderstandings. The resistance at DC is the resistance of the metal, given the cross-section through which the current flows, and its length. The conductivity you specify is used for this calculation. When the frequency increases, the effective cross-section is reduced because of the skin effect. This gives you a higher R. However at frequencies greater than zero you see an impedance of the structure, not a pure resistive term.
Best Regards,
Enrico
|
|
|
