Dear All, I am trying to Cosimulate Matlab and Cadence. That is my idea is to simulate the circuit in Cadence using Spectre or AMS and to feed the data to Matlab for deep analysis. For that I setup the SimulinkCoupler and the Fixed-Cell coupler. But whenever I am trying to simulate from AMS, (either in Parallel Matlab and AMS or Cosimulating Matlab and AMS) I am getting two errors: 1. LD_ASSUME_KERNEL variable set which may cause unstable simulation. 2. couple module 'tb_ieee_802_11a_demo.I10': *E,INITCF: ERROR (SIMLNK-66): [Internal] The connection to host 'localhost' and port 5023 failed Please any one can suggest some remedy for this problem? Also it says that some of the modules in the Simulink Schematic of the template "tb_ieee_802_11a_demo" (it comes as a part of Cadence Distribution) is outdated and compalins about the mex files. I am usimg the following version of Matlab, Spectre. Matlab : R2009a Spectre : sub-version 7.2.0.174.isr3 NCSIM: ncsim: 08.20-s021 ICFB: sub-version 5.10.41_USR6.127.29 Please suggest. Thanks in advance for your support in advance.. ------------------------------ Thanking you, Yours sincerely, Debjit ====================== Debjit Pal
Hi Debjit, Can you look at Cadence's AVUM workshop ? The AMS Virtuoso use model (AVUM) workshop is in $CDSHOME/tools/dfII/samples/tutorials/AMS There is one dedicated chapter for Matlab/Simulink with AMS Designer. I remember having seen some info about which Matlab version to set, Can't really recall off the top of my head. The workshop comes with both the doc and the simulation database you can exercise for a better understanding. let me (us) know whether you solve your problems with this. regards, Riad.
Am 10.03.2010 20:19, schrieb Debjit: If you need Matlab for postprocessing there is no need to do a cosimulation. There is a Cadence Toolbox available which can read simulation data into Matlab. It's available in the MMSIM instalations, check the docu for details. Regards, Marc
Hi Riad, I have worked out the workshop before I wrote into the group. But it is really to concise and I think very hard to have the deeper understanding of the actual tool flow for the co simulation purpose. Can you please provide any more good references? Thanks in advance. Cheers!!! Regards... Debjit.
Debjit wrote, on 03/11/10 09:29: Debjit, First of all, do you really need to do co-simulation, or just post-processing in Matlab (as Marc asked)? Also, did you get the workshop working? If you had, that probably would have made it clearer. In other words, did you get the problem you were originally reporting when using the workshop, or was that just when trying to use your own data and design? It should work with 2009.a (I have 2009.b on my computer). Regards, Andrew.
Debjit, I don't have any other references I am afraid. I guess the Other existings would tell you which versions you need. My best advice now is to understand what both Marc and Andrew have said, i.e. do you need the cosimulation ? If you problem still shows, then you may contact Cadence Customer Support. Cheers ... I meant to be on holiday right now ... Riad.
Hi Riad and Andrew and Marc, It seems to me that I really require the co-simulation. Because I will do some online monitoring and real-time signal processing on the simulated outputs. It would be better if you please explain the fine line of demarcation of Co-Simulation and Post Processing. It may be that what may seem to you experienced people obvious, may be non- trivial for me. So it will be of great help if you can explain. I have found the library which Marc referred. But I think for Real Time Signal Processing, Co-Simulation is necessary with Matlab/Simulink. Please suggest. Thanks and Regards... Cheerssss!!!! Debjit.
Debjit wrote, on 03/11/10 13:50: Hi Debjit, The Spectre Toolbox for Matlab allows you to open simulation results from Cadence simulators and read the data directly into Matlab. There you can use all of the Matlab processing tools to take that data and do whatever calculations you want. If instead you were simulating in simulink, but want to include the effect of the actual circuit in your simulink simulation, then the co-simulation approach is good. For example, say you want to use simulink to simulate all the DSP you're doing to generate the input signal to a circuit-level system, and then want to read the output of that circuit-level system and do post-processing on it (e.g. things like ACPR, EVM), then the co-sim with simulink is the right way. If you just want to read the output of a simulation and do (say) an fft and further signal processing on it, the toolbox would be sufficient (you're using Matlab rather than simulink in that case). You also didn't answer my question - did it work when you tried the example in the workshop? Regards, Andrew.
Hi Andrew, I am extremely Sorry. I missed to read that you asked whether the workshop worked or not. It worked in my setup with Matlab R2009a. Do you mean the collection of codes of Matlab and Simulink in the path $MMSIMHOME/tools/spectre/matlab and $MMSIMHOME/tools/spectre/simulink respectively as the Spectre Toolbox? If you mean them as the "Spectre" toolbox, then they are really good for offline post processing. But I want to do online processing, for which Simulink Cosimulation is the best candidate as you explained in your post. Thanks for the neat and concise explanations of the ideas. Thanks and Regards !!!! Cheers !!!! Debjit.
Debjit wrote, on 03/12/10 05:56: Hi Debjit, So I'm assuming that the issues you described in your original post (the first in this thread) are resolved? You say you got it working, whereas in the original post it was not working - you had various errors. Regards, Andrew.
