Hello guys. Lots of nice antenna models. Good job!
Wonder if NEC/4NEC2 is a good tool to evaluate, develop or optimize antenna, if it has symmerical lines, build as Wire-structures, not as TR-Lines.
Many antennas of this kind in holl_ands album (vertical stack Dipoles, BowTies etc)
I did test model of 2xfolded dipols in vertical stack (the simpliest model to test symmetrical wires), and 2-Bay BowTie.
If I feed model with 2 voltage-sources, or with Transmission-Lines - everything works perfect.
But when I build TR-Line as Wire-structure, behaviuor is enexpected.
Is it fault in my model, or is this a limitation of NEC-engine?
858 MHz 2-Bay Bowtie with Hard-Wired Feedline Pair and NO Reflector:
I took a different tack....I added my usual FR/RP and CMD-EVAL Statements and ran it under 4nec2. Geometry and Segment Checks were OKAY, but AGT was WAY OFF....EXCEEDINGLY HIGH by about 15 dB. I adjusted Radius of the Simulated Balun SOURCE (GW1) until I succeeded in getting AGT=1.0 on desired 858 MHz.
Running the File with my FR/RP Statements [for 758-958 MHz Freq Sweep] shows that the "Impedance Resonance" [i.e. PHASE = 0] was at about 784 MHz, corresponding to MIN SWR with approximate Zc=100-ohms [look at 4nec2 Impedance Chart]...which would be require an unusual 2:1 Balun to match to 50-ohm Load. Raw Gain increased linearly from lowest to highest Frequencies, as is typical for a Simple Dipole and many other Antennas [LPDA is a rare exception to this "Rule"]. As shown in the fol. EVAL RESULTS, SWR=1.67 at 858 MHz was a very acceptable operating point, presuming Zc=100-ohms, which would be compatible with Transmitters limited to SWR<2.0, allowing for somewhat higher final SWR due to effects of the Transmission Line, etc.....and 0.5 dB higher Raw Gain at 858 MHz than if it were Re-Scaled to move SWR Min to 858 MHz.
If you intended to design a 2-Bay Bowtie with a DIFFERENT Characteristic Impedance (Zc)....and Balun, please let us know....that would require DIFFERENT Dimensions to the Feedline Spacing and perhaps also the Feedline Radius....and perhaps even changes to other Dimensions.....which is what nikiml's Python Optimization Scripts are good at determining......
I can not comment re the Separate SOURCE, nor the Transmission Line alternatives, since OP did not provide those 4nec2 Files.
EVAL RESULTS for 858 MHz 2-Bay Bowtie with Hard-Wired Feedline Pair:
600 MHz 2-Bay Folded Dipole with Hard-Wired Feedline Pair and NO Reflector:
Using same procedure as above [AGT wasn't off by much but did add Variable SYmbol for adjusting Simulated Balun SOURCE Radius so AGT=1.0], I ran revised 4nec2 File and found that SWR Min was at 644 MHz with Characteristic Impedance of about Zc=200-ohms and Raw Gain close to MAX Gain, which would be compatible with the usual 4:1 Balun. A Re-Scale from 644 to 600 MHz should be all it needs [Increase ALL Dimensions by F = 644/600 = 1.073.
EVAL RESULTS for 600 MHz 2-Bay Folded Dipole with Hard-Wired Feedline Pair:
I have no intention to build actual 858-Mhz (CDMA 800) antenna.
I was looking at TV antennas (stack dipoles, BowTies etc)
I had in-depth look how each part of antenna influence final impedance, including R of symmetrical lines, and their L (do they work as matched line, or as transformators. What is trans-ratio etc)
To test NEC2 behaviour I created 2 simple test models: 2xfolded and 2xbowtie (600 Mhz, whatever impedance (preferably at X=0, Z=R), not with practical concern)
Then tried to rescale it to 858 and decrase bowtie R from ~600 to ~400 with thicker wire and more screen offset.
My only question is: can we trust to 4NEC2 analysis for UHF antennas with thick wires, wire lines etc
HFSS vs 4NEC2 analysis of the same structure are going to be the same?
I did know about AGT concern, great thanks for clarification. Will read more about that and do more tests
I don't normally deal with Impedance at one particular spot frequency...so I don't really have a "feel" for how much a small Dimensional difference changes those values.
What are Raw Gain and SWR vs Freq Charts look like using HFSS???? [And where can I download at a User Guide for HFSS????]
First model with TR line as wire structure.
Very similar simulation data!
Gain sweep is +-0.02 dB identical to 4NEC2.
Impedances are a bit different (HFSS Real part is lower a bit), but in general behaviour is good.
For very wide band and modest-SWR solutions, NEC2 precision is quite enough.
For low-SWR and narrow band antenna (e.g. transmitting antennas), NEC2 can be used to develop draft, but need more accurate tool to verify design
HFSS is one of many competitive software, which employs "finite element method" (CST Microwave, FEKO - other popular)
NEC engine uses Method of Moments.
FEM method is extremely expensive in all ways:
- software is very complex and costly
- geometry/material model is very complex, it allow to draw any shape and use any combinations of materials. One can trace waveguides, cabling, PCB, radome etc
- simulation uses enormous amount of RAM and CPU power (BiQuad and 2-bay bowtie with 11 rods took ~5.5 Gb and 1 hour CPU time to proceed simulation + "Fast" freq sweep). Usually people use Xeon clusters for work or at least decent Workstation with 16...64 Gb RAM)
I asked experts theoretical questions about expected impedances of TR lines. They answered both theoretically and confirmed calculations using HFSS. I decided to learn HFSS as a hobby, and model some industrial antennas (already did a bunch) and verify some TV antennas developed at this forum.
From what I see by now, Far-Field is always almost identical (within error margin).
Impedances sometimes are very close, sometimes are quite different. Problematic cases are folded dipole (if it is low height and thick wire!) and TR lines as wire structures.
Good practice of antenna development is manufacturing prototype and measuring it's actual perfomance to confirm modeling.
With modern software and hardware it is possible to verify model without instrumental data. Experts said prototypes has never contradicted to CST or HFSS solutions.
If there is interest, I can convert more models to HFSS (and share *.hfss file) and provide solutions data.
I suspect that difference in (R)eal impedance increase at low frequencies because of error in TR Line transformation ratio.
At higher Freq, TR length is closer to 0.5 lambda, at lower Freq closer to 0.25 lambda.
Non-matched line do not influence impedance at 0.5 lambda length, and maximum transformation occurs at 0.25 lambda.
NEC2 inaccuracy increase if lines are 1/4L and decrease at 1/2L
I found that fol. Log-Yagi Antennas were the LEAST Believable when modeled using 4NEC2 (unfortunately, High Resolution NEC4 Engine is expensive and subject to U.S. Export Controls). Note that I attempted alternative Modeling Approaches for each Antenna. Hence we would be interested in what HFSS calculates (esp. Ch13 Performance for YA-1713). Models are based on MY Measurements and I still have both stored in my garage if any more photos or measurements are needed:
I suspect that HFSS calculations for SWR MIGHT be a bit HIGH. I would recommend running HFSS on a commercial Antenna where the manufacturer has published their own Model Results, such as the C2 [Dual Tapered Loop] from Antennas Direct. ADTech is active on this and other Forums and usually responds to specific questions...such as I'm not sure if "Re" (1/2 the thickness of the thin sheet metal Element) came from him or if I made a guesstimate....I'll have to check my emails and notes: Hi-VHF+UHF Tapered Loop - A-D C2 & C2-V [I believe the Dimensions are fairly accurate.] https://www.google.com/url?sa=t&rct...DS.pdf&usg=AFQjCNFHb3SNgxFgxIv-ambm-YnJYT-Jzg [Remcom, Inc. X-FDTD 7.0 Model]
holl_ands I'd be glad to make YA-1713 model in HFSS.
What is boom WxH (mm)? https://goo.gl/TwTayn yellow numbers are dimensions of real antenna? Purple is offset of wire's centre from top plane of boom?
Does it make sense to model plastic, or I can hang DE1/DE2 in vacuum and don't bother with dielectric mounting?
Boom is 1x1-inch (25.4x25.4-mm). Although I did a separate detailed Model for the Crossover between Active Elements, I modeled JUST the Boom with a SIMPLIFIED wire structure for the Crossover, but I didn't get very far along the path to combine the two models. After all, a detailed Boom Model wouldn't be NEEDED in 4nec2 anyway.....that's why Boom Corrections are added to Passive Elements when modeling WITHOUT the Boom: YA-1713 Detailed Boom & Crossover
WOW....it's been nearly 10-years since I started modeling YA-1713. In comparing one of the Boom Corrected 4nec2 Models against the Spread Sheet, the YELLOW numbers are Actual Measurements and there are alternative columns for how much Boom Correction needs to ADDED to Actual number to enter into 4nec2 Model (w/o Boom).
And just because I'm an American with one foot on each shore straddling the Atlantic Ocean, Spread Sheet is Metric [per formula source documents] and my Models are usually in Inches....so I keep the fol. CONVERSIONCALC Plus next to my Desktops [12 Hyperthreads each] and another next to my Laptop [8 Hyperthreads]: https://www.amazon.com/Calculated-Industries-8030-ConversionCalc-Professional/dp/B009FFS6IO
Since Plastic pieces are so SMALL compared to Element Length, I doubt that it makes much difference at Hi-VHF Freqs. But it's your choice if you want to simply omit them and "hang" it in a vacuum.
You should be interested in Dragoslav Dobricic [YU1AW] use of Finite Difference Time Domain methods (FDTD) or Finite Integration Technique method (FIT) to overcome known limitations of MoM Type Models (like NEC2/4). Esp. see various Yagi Analyses, beginning with "Boom Influence on Yagi Antennas", 2009: www.qsl.net/y/yu1aw/Misc/boom_yagi.pdf http://www.qsl.net/y/yu1aw/Misc/vhf_ant.htm [VHF/UHF Antennas] Home [Homepage]
Preliminary model, need more data on Logo-waveguides.
TR-Lines effective length is critical for phased array.
DE1 to DE2 offset is 270.0 mm
TR1 in your *.NEC is 353.8 mm (combined), TR2 is 371 mm (combined).
In my HFSS model using 8x0.5 mm ribbon, shortest path by edges is ~370 mm.
Top and bottom line are identical and mirrored vs center of vibrator wire.
Top ribbon is +39.09 mm above boom (Z=0 mm)
Bottom ribbon is -4.89 mm below boom (Z=-30.29 - (-25.4) = -4.89 mm)
"Ribbons vs boom" have characteristic impedance of a "wire over a ground plane", it is dependent both on radius/width and distance to ground/boom.
Impedance of this lines influence impedance of phased array, and what is more important - if 2 waveguides have different impedance, they radiate and SWR increase.
To improve model, I need 3 answers:
1) What if effective electrical length (shortest path) of TR Lines in YA-1713 sample. Are they identical in length?
2) What is exact ribbon width and distance to boom? Are they identical top and bottom?
3) How manufacturer trick to make equal distance from ribbon to boom?
H2+11RR Gain is very close, I got 13.8 dBi Max with 4nec2 vs 13.9 dB in the above Plot (presumably dbi ????). SWR per HFSS is still (AGAIN) a bit Higher than 4nec2...and again is Freq shifted DOWN, but mostly on the lower Freqs.
Looking forward to comparison vs A-D's C2V Modeling Analysis....
===============================================
Until I can dig into the Garage to find the YA-1713 and take some Photos (with ruler), the fol. should give you what you're asking for....I modeled all of the angles and jogs in the Crossover, presuming it was a THICK Wire right down the middle of the Ribbon. Download and View Item #20 4ne2 File: http://imageevent.com/holl_ands/logyagi/wgya1713/ya1713detail
PHOTOS [w/o Ruler...hey it was nearly 10-years ago and I've learned a lot since then] are on fol. webpage....which presumed a [Quasi-Optimized] Simulated Transmission Line Crossover between DE1 and DE2 using a 4nec2 "TL Card" instead of any REAL Physical Structure....so it emulates what Crossover SHOULD be doing if it was correctly implemented. Note that some runs were for MINUMUM Length DIRECT Connection, while others assume Lt=13-in, the total length along the ribbons (if I remember how I did it way back then), with Zt being the (Quasi-Optimized) Impedance of the TL.....Results were NOT very pretty: http://imageevent.com/holl_ands/logyagi/wgya1713/ya1713modelingthecrossover
Hmmm....it doesn't look like I investigated LONGER TL Lengths, such as you mentioned above....I'll have to remember to do that....
Angles and shape of lines aren't critical.
Critical is electrical length (path by closest edges) from tip of DE2 to the opposite tip of DE1. This influence angle of phase shift.
And critical is impedance of this line, which depends mostly from ribbon width/height and distance to boom from longest ribbon and shorter straight ribbon. Shape of connecting arc is not critical.
I found idea how to make distances from TR-Line to boom equal - just bend short ribbons down.
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Related Threads
?
?
?
?
?
Canadian TV, Computing and Home Theatre Forums
1.7M posts
114.9K members
Since 2001
A forum community dedicated to Canadian TV, computing and home theatre owners and enthusiasts. Come join the discussion about home audio/video, displays, troubleshooting, styles, projects, DIY’s, product reviews, accessories, classifieds, and more!