Structures, comparison...
Nice to see a new forum so interesting for me!!
I can give up later some structures which have been simulated and measured:
-patch antennas and antennas for mobile commun.
-horn antennas, double ridged horn
-transitions from coax to waveguide (pi-type)
-microstrip capacitor (vertical stub) and maybe more planar circuits
But give me time, I'm very busy now!!!
Best regards,
Eirp
Hi,
here comes part 1 of the first patch.
It's koch3 snowflake built on GIL GML 1000 (h=1.5mm) with er=3.05 and tandelta=0.002. Feed is coaxial, 0.625 mm is the radius of inner coaxial conductor. Position of the feed can be seen from the figure. It has been simulated by IE3D and MWS and measured too.
Now I'm sending .dxf and .geo of the snowflake, tomorrow I'll send meas. and simul. data.
Ground plane is infinite for IE3D and finite for MWS (details in next part).
Eirp
Hi again!
I've prepared the data. You'll find it in the attachment as standart TOUCHSTONE format.
I don't added the picture because you (experienced EMans) will prepare the graphs.
Notes:
- ground plane in MWS was 200x200mm (zmin=open!), IE3D-infinite GP.
- reference plane is at the ground plane (so there is inductive component due to the 1.5mm feeder)
- simulation was performed with the feeder of course.
- agreement between measurement and IE3D is excellent I think (13cells/lbda used). Data from IE3D are divided
into 3 parts. MWS is a bit worse; Of course,planar simulator will make here a better work.
- it was measured on Agilent E8364A vector analysator (very good one, 45MHz - 50GHz!!!)
Please cooperate with me and try another EM simulator so we can compare more (Sonnet??)
Next week I'll continue with other patches and the transitions also.
Now it's your turn, GOOD LUCK!!!
Regards,
Eirp
PS: Far field maybe come in future
I've measured farfield of the Koch3 patch I've sent first. See at attachment.
Please sorry, it's in my language, so I'll explain:
full black line - measured (mode #13 --> 2.759 GHz)
.-.-.- black line - simulated IE3D (mode #13 --> 2.756 GHz)
....... blue line - antenna's measured pattern at fundamental mode (only for info)
There is shown relative power as a function of theta with const. phi.
Regards,
Eirp
No feed was defined in this .geo model.
Yes...
1) You can made it by your own, isn't it easy?
2) I've sent .geo mainly for peoples who can have problems with attached .dxf (and have IE3D). I hope that someone try to simulate this structure with another simulator as Sonnet or any appropriate so we can compare results with measurement and think about accuracy.
That's it!
Eirp
Here is attached one of coax to waveguide transitions from my colleague.
I have MWS files now. I hope I'll get results from measuring too.
Now it's for your comments...
Regards,
Eirp
FUCK!!!
Attachments are deleted too, not only e-books....!!!
I'll upload it again later...
Eirp
Here are the deleted data files plus four from my own
koch_ensemble.s1p (s11 simulated with Ensemble 7.0)
s11_ie3d.pdf, s11_mws.pdf, s11_ensemble.pdf (simulated vs. measured data graphs in pdf format)
I'm looking for a PC to install Sonnet 8.5.2 to simulate the structure.
keep on waiting...
ChinaSky: what is your conclusion?
Chinasky,
thank you very much for your contribution!!!!
Seems that if fits also well - small resonance shift at higher frequencies is produced a bit by IE3D too..
Please, add some next info - how many memory it takes and the number of cells/lambda. Did you use uniform or nonuniform meshing?
Later I'll send my condition also...
Best regards,
Eirp
Hello eirp
These are the details about meshing/machine/etc..
Adaptative meshing at 4 GHz
Max. normalized variation in S parameters between 0.05
Max. passes 3
Normalized variation in S parameters at third pass 0.00744
Number of triangles in first pass 259
Number of triangles in last (third) pass 1130
Memory used in each pass: 45 Mb
2 mins. / frec (1130 triangles, 45 Mb)
Machine: UltraSparc 10 (450 Mhz 128 MB)
1000 points 39 hours CPU time (it is slower than my laptop Pentium III 450 Mhz )
I have included two gzipped PostScript files with the initial and final mesh (their pdf counterparts are bigger and bandwidth is very expensive nowadays )
Hi all!
Here are my simulation conditions in IE3D:
- 1819 cells total (9 rectangular + 1810 triangular), 1811 2D cells, 8 3D cells (due to the coaxial feeder)
- 2639 unknowns
- Time per frequency: 118 sec on Athlon XP 2100+ (512MB RAM)
- RAM used for simulation: 56MB
Regards,
Eirp