Thursday, February 21, 2013

Double BiQuad 3.5 GHz - WiMAX - G4 antenna Analysis and Testing


There are loads of information, how to build correct biquad or a double biquad antenna for Wi-Fi (2.4GHz) adapters. But there are only a few sources in the internet which shows how to make a cheap proper handmade G4 WiMAX antenna for 3.5 GHz frequency. In this analysis I will try to figure out which dimensions are prime choices for the best performance for 4G WiMAX 3.5 GHz Double BiQuad sector antenna. Will be testing two types of antennas with "lips" and without them.


Antenna simulation, optimization and analysis was done with computer software 4nec2 (antenna modeler and optimizer), version 5.8.11.

My target - frequency 3475 MHz.

Wave-length in vacuum at frequency 3477 MHz equals to wave-length in Earth atmospheres air at frequency 3475 MHz. So lambda=86.2446 mm=8.62446 cm.

Simulation NEC Input files:

Configuration symbols:

  • ED  - Element square diagonal length / 2
  • WR - Wire radius
  • S - Spacing between element and reflector
  • RH - Reflector height / 2
  • RL - Reflector  length / 2
  • LH - Height of reflector "lips"
All dimensions in .NEC data files and graphs are in wave-length ratio.

Main 4nec2 windows (antenna with lips on the left, without on the right)

 Geometry (antenna with lips on the left, without on the right)

Smith Chart

3D Radiation pattern and Smith Chart

 Tot-Hor Gain (antenna with lips on the left, without on the right)

This antenna should get the best performance at horizontal polarization.

 Wire radius (antenna with lips on the left, without on the right)

Thinner wire gives slightly more Gain, but the main impact is in SWR. As you see ‘lips’ = better results.
With lips 0.0059×86.2446×2≈1.017mm≈0.1cm
Without lips 0.007×86.2446×2≈1.207mm≈0.12cm

Element spacing (antenna with lips on the left, without on the right)

SWR and Refl.coef. graphs displays significant impact of distance. This means that the element should maintain correct distance without bends up or down on the sides. Lips boost the performance.
With lips 0.0915×86.24467.891mm0.8cm
Without lips 0.0805×86.2446≈6.943mm≈0.7cm

Element dimensions (antenna with lips on the left, without on the right)

SWR and Refl.coef. graphs displays a huge impact in element dimensions. This means that you really want to maintain a perfect size.
With lips  2/2×0.1789×86.2446≈21.820mm≈2.18cm
Without lips 2/√2×0.1777×86.244621.673mm≈2.17cm

Reflector length (antenna with lips on the left, without on the right)

The optimal reflector horizontal length is 20.2cm
With lips 1.169×86.2446×2
201.639mm20.2 cm
Without lips 1.17×86.2446×2
201.812mm20.2 cm

Reflector height (antenna with lips on the left, without on the right)

The above SWR and reflection coefficient graphs show that using lips you must maintain perfect size (9.1cm). Though reflector without lips indicates no significant impact in SWR.
With lips0.5282×86.2446×2
Without lips 0.555×86.2446×2

Height of reflectors “lips”

The optimal “lips” length is 23mm as it is shown in graph above.
The main impact is in SWR and reflection coefficient.
Lips boost the performance.

Frequency range (antenna with lips on the left, without on the right)

The above SWR and RF.COEF. graphs shows that best performance at frequency 3477 MHz. Lips boost the performance.


Spacing and Wire thickness test:

As the 4nec2 modeling shows significant impact on the wire’s thickness and element spacing, it was time to see what would happen in real life with various thickness of wire and various spacing for element. Sadly I don’t have a laboratory, neither can find one around, so the testing was made in home conditions as best as I could (It took all weekend to test 4 types of wire with different polarization and spacing, repeating tests hundreds of times).
I made 4 elements of various thickness wire as you can see below.
To make one element nicely takes about an hour or a bit more. Note: The thicker the wire the harder to bend it right.

Use this PDF file for bending wire contour. Print without changing scale and contour will be correct size on paper.
From left to right: copper 1mm; copper  1.2mm, aluminium 2mm and the last copper 2.2mm

Two days careful testing averages below:

VP – Vertically polarized; HP – Horizontal polarized.

CINR [dB] - Carrier to Interference + Noise Ratio (sometimes SINR: Signal to Interference + Noise ratio). Higher values are better.
RSSI [dBm] - Received Signal Strength Indicator. Higher values are better (Note: Its negative, higher is closer to zero).

The main impact is on CINR as its shows the channel quality, possibility to choose better modulation.

Hence, the tests shows that 4nec2 modeling don‘t lie to much. The prime choice is for the copper wire of 1mm in diameter with spacing 0.8cm over reflector.
Aluminium wire displays a good performance although its thickness over 4nec2 predictions. Still, I found it very hard to solder, almost impossible. 


  • Copper wire of 1mm or 1.2mm in diameter with spacing 0.8cm over the reflector demonstrates best performance.
  • Use “lips” on the reflector. Theoretically “lips” gives slightly better performance.
  • Maintain correct element dimensions.
  • Maintain correct reflector dimensions.
  • Use horizontal antenna polarization. (antenna is horizontally polarized then reflector is vertical)


All measurements in the simply schemes so even a horse could understand

Scheme for double biquad with "lips"

Scheme for double biquad without "lips"

How-To Build antenna: here

Last updated 07 Jul 2014

1 comment:

  1. For dimension 2.18 cm and 3.1 cm, dimension 12.3 cm is impossible.
    For dimension 2.18 real dimension is 3.083 and 11.977 cm
    For dimension 2.18 and radius of corner r=1 (minimum) real dimension is 3.0001 and 11.8248
    Which dimension is main dimensions 2.18, 1.3 or 12.3. I think that is 2.18

    Standard dimension of wire: for 0.75mm2 d=0.98mm, for 1mm2 d=1.13mm

    Which the thickness of blank is recommended, and material