Hi all,


just wondering, if I have an inductor's S-parameters for example, is there any software tool which will allow me to input my own RLC model, and then give me a best fit?


thanks,
Aaron

Hi,

another good software doing this (and much more) is Aplac  :)!! It is a good analog circuit simulator which contains also good optimizers.

Bye Stephan

Hi all,


I've had a look at the responses, but I'm not sure if they do what I was asking.

1. Origin. I've used this before and as I remember its a bit like microsoft excel. Am I looking at the wrong software? How would you use it to create an RLC model? thanks!

2. Aplac. Actually I haven't been able to get much information on this tool. Do you have any link which shows how it can accomplish what I need?

3. ADS. As I understand it, the only ways to do it in ADS are by manually curve fitting using the tuner, or using their own pre-defined networks. Am I missing something?

Manual curve fitting can be very tedious especially when the model becomes complicated. Also, the predefined models from ADS (ladder, pi, T...) sometimes result in models which lose their physical meaning or are overly complicated. This makes it difficult to find problem areas in the circuit and modify as necessary.

I figured that since all PDK inductors, capacitors and resistors (components really) need some RLC model, there would be some software which would automatically generate the parameters for a user-defined model based on S-parameters.


thanks,
Aaron

In the meantime, I downloaded the trial version of IdEM from http://www.emc.polito.it/software/IdEM/idem_home.asp and played around with it a bit. It does not allow you to fit to a pre-defined model, but it can generate a netlist model suitable for transient simulation from S-parameter data (similar to Sonnet's Broadband Spice Extraction, see http://www.sonnetsoftware.com/products/sonnet-suites/bb_spice.html).

The trial version of IdEM is limited to 2 ports and order 20, which should be perfectly adequate for a normal inductor. For building the model, I selected the FDVF-AS algorithm with minimum order 2 and maximum order 20. After that, I used the Hamiltonian passivity check and compensated the model if necessary. Finally, I checked the accuracy of the model and generated the Spice model.

I would expect that the full version of IdEM has no arbitrary limits. I have no experience with the Spice Model Generator from ADS.

aaron_do wrote on Jan 6^th, 2012, 10:33pm:

That's absolutely correct. If you want go get direct insight into your design from the parameter values of the fitted model, then IdEM is not the right tool for you.