The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl Modeling >> Behavioral Models >> ac small-signal simulation of a magnetics behavioral model https://designers-guide.org/forum/YaBB.pl?num=1283897776 Message started by engelbrl on Sep 7th, 2010, 3:16pm |
Title: ac small-signal simulation of a magnetics behavioral model Post by engelbrl on Sep 7th, 2010, 3:16pm I have a question regarding ac small-signal simulation with magnetic fields. It may be more of a questio\ n concerning how Spectre actually works. I have written a compact model in Verilog-A which successfully \ describes the dynamics of a magnetic nanoparticle (abbreviated code shown below - Theta(Mpol) and Theta(M\ az) are the polar and azimuthal angles of magnetization respectively). I want to do an ac simulation to \ show ferromagnetic resonance at a particular frequency and have thus far been unsuccessful. The stimulus\ is the ac field and a DC bias field, both in the z-dir (perpendicular to the material plane), because I \ would like to see the magnetization rotate about the z-axis. The simulation runs but the output is extre\ mely small (zero really...) and without any resonant peak at the expected frequency. I'm wondering if there is a reason why the small-signal analysis would not work with the system of equati\ ons I have developed for this magnetic system (I realize that I might not have yet all of the conditions \ just right to get a reasonable output). I had assumed that since I had a DC operating point from other t\ ransient simulations, Spectre would linearize the system of equations around that operating point in the \ same way that SPICE does. I don't think it should matter whether the system uses V and I or M and H - Doe\ s anyone know if there is a limitation to the kind of system in which a Spectre ac analysis can operate? \ Does anyone know if Spectre really does do a differential linearization of the system equations in the s\ ame way SPICE does? Thank you, Linda (partial code listing) ----------------------------- nature Magnetic_Induction abstol = 1e-6 ; access = H ; units = "A/m" ; endnature discipline sig_flow_H potential Magnetic_Induction ; enddiscipline module SW3D(hx, hy, hz,Mpol,Maz,Mea ); input hx, hy, hz ; inout Maz, Mpol, Mea ; sig_flow_H hx, hy, hz, hMpol, hMaz ; // H-field vector components rotational Mea, Maz, Mpol, Anis; // port discipline for easy axis angle // and magnetization angles(spherical) : //parameters set : analog begin : H(hMpol) <+ ....eqn for H(hMpol).... H(hMaz) <+ ....eqn for H(hMaz)... Theta(Mpol) <+ idt((C1*G0)*(H(hMaz) + alpha*H(hMpol)),`M_PI/2 ) ; Theta(Maz) <+ idt(((C1*G0)/sin(Theta(Mpol)))*(alpha*H(hMaz)- H(hMpol)),`M_PI); end endmodule |
Title: Re: ac small-signal simulation of a magnetics behavioral model Post by Ken Kundert on Sep 7th, 2010, 8:36pm Did you apply the small signal stimulus? -Ken |
Title: Re: ac small-signal simulation of a magnetics behavioral model Post by engelbrl on Sep 8th, 2010, 8:58am Ken, Yes, I did apply the small signal stimulus, at least I believe I did it correctly. I use currents as inputs to generate magnetic fields via the behavioral expression: H = I/2w (w is the trace width). I put a DC current source (idc) in parallel with a sinewave current source (isin) both from the analogLib. The idc magnitude is 6mA (this is a small-moderate field strength effective for my system). However, I am confused about the ac source. The edit-object-properties box for isin asks for: 1) an AC magnitude -> which I set to 200uA (pk-pk swing) 2) a DC current -> which I set to 6mA, just like the parallel idc source <= is this a problem? 3) an "Amplitude" -> which I set to 100uA <= I assume this is the Peak amplitude? I have searched through the Spectre Users guide and the Spectre reference for what these object properties mean, and can not find them. Can I use just the isin source, without the idc source since I can specify a DC source value in isin? Am I using the AC magnitude and Amplitude options correctly - what Cadence manual talks about the object property settings? I saw that you have written a book on Spectre and Spice. In this book do you explain how these tools linearize and solve circuit equations? If it is not something that you address, can you cite a reference that does? Appreciatively, Linda |
Title: Re: ac small-signal simulation of a magnetics behavioral model Post by Ken Kundert on Sep 8th, 2010, 6:13pm Perhaps you should just give the netlist so we can look at it and see if you applied the ac magnitude correctly. -Ken |
Title: Re: ac small-signal simulation of a magnetics behavioral model Post by engelbrl on Sep 9th, 2010, 3:29pm The netlist is attached. Thank you for looking at this. Linda |
Title: Re: ac small-signal simulation of a magnetics behavioral model Post by Ken Kundert on Sep 9th, 2010, 9:29pm I can't see any thing obvious. Of course, you didn't give me much to go on. The testbench is missing, the models are missing, the analyses are missing, and a detailed description of what is wrong is missing. -Ken |
The Designer's Guide Community Forum » Powered by YaBB 2.2.2! YaBB © 2000-2008. All Rights Reserved. |