| Author |
 Topic  |
|
|
buenos
United Kingdom
2 Posts |
 Posted - Jan 18 2009 : 11:43:38
|
hi
what type of EM simulation the fasthenry and fastcap do?
is it full-wave, or quasi-static?
(if its quasi static, then its not good for GHz frequencies, or is it?)
if i do a simulation an a pcb interconnect with fasthenry then with fastcap, then can i combine the results (adding imag values) to get a realistic impedance versus frequency data? or does the summing have to be done on every unit-cell separately (probably impossible with these tools)?
there is a newer fieldsolver from MIT: fastimp. I was trying to compile it to windows, or even to linux but i couldnt suceed. compilers have thrown lots of errors. It would do realistic impedance versus frequency data with full-wave EM simulation. ht*p://w*w.mit.edu/people/zhzhu/fastImp.html |
|
|
Enrico
545 Posts |
Posted - Jan 18 2009 : 22:37:44
|
Quasi-static. If it is good or not a GHz frequencies depends on the dimensions of your structure. However if you are looking at PCB interconnect, then the dimensions does not allow full 3D simulation at GHz frequencies (though you can still calculate the parameters for transmission lines with constant cross-section).
On how to combine FastCap and FastHenry results, yes that's not so easy; you can look at Matt Kamon's PhD thesis, you can find it on the MIT web site (follow the link on the 'Links' page).
Yes I'm aware of FastImp, however at the moment we do not support a Windows version. I know compilation is a bit complex, also because FastImp is built upon various libraries you have to compile as well.
Best Regards,
Enrico Di Lorenzo |
 |
|
|
buenos
United Kingdom
2 Posts |
Posted - Jan 31 2009 : 00:04:23
|
hi
thanks for the answer.
"simulation of e.m. behaviour of a 3D structure with SPICE-like simulators"
how can a simulation in spice to be accurate if capacitances and losses are not taken into account in fasthenry, and inductances in fastcap. is it still describing the behaviour?
how do you combine the fasthenry with fastcap for example for a connector simulation. (with macromodelling in spice-like simulator)
"you can look at Matt Kamon's PhD thesis" -ok, its long, but please give me a clue: do some mixup with fasthenry/fastcap output files, or it needs some more extensive computing after using fasthenry+fastcap? or even no way for combinig them for a meaningful result?
"Yes I'm aware of FastImp, however at the moment we do not support a Windows version" -it seems we dont even support a linux version. i have tried to compile on 2 linuxes. the problems are not missing libraries, but syntax errors, for example: "<>>>" is not recognised as a function... |
|
|
|
jk_1420
1 Posts |
Posted - Jul 11 2009 : 04:08:42
|
I realize that this post is well after the fact but I think will be useful for others attempting to compile fastImp v1.1 by Zhenhai Zhu at MIT.
gcc 4.2 on Linux yields numerous compilation errors. After spending hours modifying the source I was able to get "it" - pfft++, fastImp, and fftw ... - to compile. However, fastImp hung on the first test run. The driver tests for pfft++ also hang on the vector matrix mult, so pfft++ is probably one of the problems. gcc/g++ compile options that address c++ dialect issues have no effect.
Another attempt on the original fastImp source using gcc-3.4 also failed.
I found identifiers in the source that were never declared, anywhere in the code! Perhaps Mr. Zhenhai Zhu was careless when he created the archive pulling from stale source. |
|
|
|
Enrico
545 Posts |
Posted - Nov 06 2017 : 17:42:12
|
Resuming an old topic, but for completeness, full cleaned-up sources as well as Linux and Windows are available from our download section.
-
Enrico
|
|
|
| |
Topic |
|
|
|
what type of EM simulation?
