Simulating Complex Loads

[Evan.Cornell]

If I have data describing a complex impedance (say, 100 data-points of magnitude & phase at various frequencies) collected from a RLC / impedance analyzer, is it possible to insert that into simulation program for analysis in a larger system?

I am currently familiar with TINA-TI, so being able to use that program would be ideal, but if anyone knows how to do such a thing in other (preferably free) tools, that would be extremely helpful.

[Nitrousoxide]

If you have the frequency response (or time domain response) from the network, you can then use multiple methods to generate a transfer function in the laplace domain (s-plane).

A common method is design by bode plot, where the magnitude and phase response is observed and a transfer function is generated usually in zero, pole and gain form.

Once a transfer function is determined, it can be realised by either passive or active networks. A good example of this is a pure-integrator implementation.

If you're looking for a specific process, here is a good example with matlab: http://au.mathworks.com/help/ident/examples/frequency-domain-identification-estimating-models-using-frequency-domain-data.html?prodcode=ID&language=en

How complicated is the bode plot? (how many poles and zeros?)

[Evan.Cornell]

To add some more information, I have found out that TINA-Industrial is required to directly take impedance measurements into simulations. So, for free simulators, it does look like fitting the data to a lumped-element model is the best method for simulation.

See screenshot attached for shape of impedance data that I need to model.

[Nitrousoxide]

See screenshot attached for shape of impedance data that I need to model.

Would you be willing to share the data in CSV format? I would gladly have a go at modelling the transfer function.

[Evan.Cornell]

Please find attached. I would be most grateful if you could share your process for fitting model to the data, as I'm going to have to repeat this process in the future..

Thanks,
Evan

[Nitrousoxide]

Awesome, I will give it a show tomorrow morning as it is 2AM and brain no function 

One thing that I glossed over is to mention 2 port network parameters and how they are essential if you want to correctly model/analyse a system.

I'm going to assume that the data given is the Z11 parameter (i.e. the impedance "looking into" the input). (and before people yell at me, im approximating Zin = Z11, bad i know)
This is the case for a passive network with a single input, I am assuming this as the vertical axis in your plot has the unit of ohms.

[Evan.Cornell]

This data was taken by hooking up the test leads of B&K 891 RLC meter to the two leads of the device under test, and getting impedance magnitude data (in ohms) and impedance phase data (in degrees).

[T3sl4co1l]

Keep in mind that the impedance will vary greatly with what's on the other side of that... piezo transducer, if I had to guess.

Indeed it doesn't seem to be all that well matched, for how strong the main resonance is, and it also seems to be in a bit of a cavity, given the higher modes (not quite harmonics).  Pretty typical for air, I suppose.

The equivalent circuit will be, more or less: a capacitor (the terminal capacitance of the device), with branches in parallel.  Each branch is a series RLC, resonant at whatever frequency.  This should give you enough info to start playing around.  Though it's not all that easy to adjust; that's why I wrote this:
https://www.seventransistorlabs.com/Calc/Filter1~.html
(Hover over the schematic elements to view tooltips, transform, change, add or remove stages.  Set unused components to near-zero or near-infinite.  Click and drag values to adjust in real time.)

Tim

[nctnico]

The last time I played with piezo transducers I used a lumped model and calculated the values by using the characteristics from the datasheet. There is a lot of good info out there on how to do this. See the attached PDF.

[Evan.Cornell]

So a single RLC || C model (Butterworth VanDyke) doesn't really capture the phase information properly in the frequency region I care about (25-35kHz).

So I am back to wondering if there are any SPICE tools that can take in a list of impedance data points to run AC sweeps. Any ideas?

[rstofer]

I guess I would use LTspice.  If I wanted to find out if it can be done, I would join the Yahoo LTspice group because there are some VERY knowledgeable people over there.  Be aware that there are two LTspice groups on Yahoo.  One is a user forum and the other is a repository for files.

[Bud]

Many simulation programs have an ability to load and use S-parameters files. They would have a S-parameter component, you load your S-parameters file in it and use the block as any other component to set up an analysis. Look for such component in your favorite simulation program.

[Evan.Cornell]

I have a list of impedance values (polar or rectangular format possible) taken at various frequencies. How would I go about converting this into s-parameter file?

[Bud]

Look for Touchstone file format, there are different data representations that can be used to create a file out of your data.

[Evan.Cornell]

Any free SPICE simulators that can read in touchstone files and do AC sweeps?

[Marco]

Can't B-sources do this with LTSPICE?