Antenna Comparison Testing

This page details the testing method and results when testingsome downpipe waveguides, a collinear, some biquads, a cantenna,and a modified Conifer dish.
Contents:  Background |Test Equipment |Antennas |Test Setup |Testing Methodology |Test Results |Conclusions |References
Background
Over the past several months, I‘d constructed a number of antennas,including somebiquads, an 8-element collinear omni,and a cantenna, andMarcus had recently constructed some 8-slot 180-degree waveguide antennasusing cheap downpipe.
We wanted to test the waveguides, and compare the performance of thevarious antennas.
By testing all the antennas in a single session, using identical configurationsin the same location, allows us to easily compare the performance of thevarious antennas, as there are minimal factors which would be affectingthe results.
Test Equipment
We used two laptops, one at either end of our wireless link.
The specs for the laptop at the remote end: Pentium III 1GHz with 128Mb RAM
EnterasysRoamAbout wireless card
Windows 2000 SP2
Enterasys7.44 drivers and client utility
running on battery power for the duration of the tests
The specs for the laptop at the antenna end: Pentium III 700MHz with 256Mb RAM
EnterasysRoamAbout wireless card
Windows 2000 SP2
Enterasys7.44 drivers and client utility
running on battery power for the duration of the tests
Antennas
We had the following antennas to test:modified 24dBi Conifer (ex-Galaxy) dish
cantenna
two biquads
two 8-slot 180-degree downpipe waveguides
8-element collinear omni antenna
The 24dBi Conifer dish has beenmodified by myself,with a custom-made copper and brass dipole. This is my reference antenna,and has been used in previous antenna testing sessions(referred to as Feedhorn #3 in the May 2002Conifer Antenna Testing, and m3in the September 2002Conifer Antenna Testing II).
The cantenna was constructed by myself using a steel can with diameter of 100mm,and a length of 175mm, with a panel-mount N connector riveted into the appropriatelocation.

the panel-mount N connector

the feed inside the cantenna
The first biquad (referred to as biquad1) was constructed in a similar way tothose made byTrevor Marshall, and has two "lips" on the sidesof the reflector. This biquad had approximately 2m of CNT-400 coax attached to it.
The other biquad (referred to as biquad2), is the one featured on myBiquad Antenna Construction page. A short length ofCNT-400 coax is attached to this biquad.

biquad1

biquad2
Note that biquad1 has not been built optimally, as it is not correctly impedancematched (due to the use of the wire stakes to mount the biquad to the reflector).
Biquad2 has beenconstructed with the coax extending throughthe reflector, encased in the copper pipe for additional strength.
The collinear was constructed using 8 elements of CNT-400 coax, as perthe details atwireless.gumph.org,with the dimensions appropriately adjusted to suit the velocity factor ofCNT-400 coax.

the collinear nearly complete
It has been mounted inside a length of 25mm electrical conduit,to provide some physical strength, and to weather-proof it.

the completed collinear
Marcus constructed the two 8-slot 180-degreewaveguidesusing readily available downpipe.
The first is based onTrevor Marshall‘sdesign (referred to as the TM waveguide), while the second is basedonRob Clark‘sdesign (referred to as the RC waveguide).

the two completed waveguides
(RC waveguide on the left, TM waveguide on the right)
Test Setup
We placed one laptop on the parcel shelf of a car which we parked approximately200m down the road. No external antenna was connected to this laptop -we were relying on the internal antenna in the RoamAbout card.

the remote laptop is in the red car (circled), approx 200m distant,
with the top of a waveguide shown in the foreground
Each antenna was connected to the wireless card in the other laptop using a2 metre length of RG213 coax (terminated with a male N connector at one end,and a female N connector at the other end), and an appropriate pigtail,with the wireless cards at both ends configured to operate in ad-hoc mode.

testing the waveguide

testing the collinear

testing the cantenna
Testing Methodology
The "Link Test" mode in the Enterasys Client Utility was used to monitor the link strength,with each test configuration being monitored for a couple of minutes.

the Enterasys Client Utility displaying link statistics
Once the reported link details had stabilised, the SNR, signal strength and noiselevel were recorded for both the local and remote ends of the linkwere recorded.
All antennas were tested in both horizontal and vertically polarisedorientations.
For horizontal polarisation, the waveguides are vertical, the biquadis vertical, the collinear is horizontal, the cantenna feed elementis horizontal, and the Conifer dish is vertical.
For vertical polarisation, all antennas were rotated 90 degrees.
The waveguides were briefly tested on channels 1, 6 and 11, andas they performed best on channel 6, this channel was used for allsubsequent testing of the waveguides and all other antennas.
As we wanted to plot azimuth graphs for the two waveguides, we measuredthe signal, noise and SNR details after rotating the waveguides at10 degree intervals.

testing the waveguides at different angles
Signal readings at different rotation angles of the waveguides were onlymade with the waveguide horizontally polarised, as it is designed tooperate in this orientation.
Test Results & Calculations
The tables below show the recorded signal, noise and SNR valuesfor both the local (ie, the laptop connected to the antenna) andremote (ie, the laptop in the car) ends of the connection.
horizontal polarisation:
local remote
antenna       SNR
(dB)    signal
(dBm)    noise
(dBm)      SNR
(dB)    signal
(dBm)    noise
(dBm)
TM waveguide 18 -80 -99 22 -75 -97
RC waveguide 25 -76 -100 24 -73 -98
Conifer 30 -70 -100 30 -70 -100
biquad1 19 -81 -101 21 -80 -100
biquad2 19 -80 -100 21 -80 -101
collinear 15 -83 -99 18 -82 -100
cantenna 17 -84 -100 22 -79 -100
vertical polarisation:
local remote
antenna       SNR
(dB)    signal
(dBm)    noise
(dBm)      SNR
(dB)    signal
(dBm)    noise
(dBm)
TM waveguide 30 -70 -100 28 -74 -100
RC waveguide 30 -70 -100 29 -72 -100
Conifer 37 -64 -100 34 -66 -100
biquad1 26 -74 -100 23 -78 -100
biquad2 24 -78 -100 20 -80 -100
collinear 23 -78 -99 20 -80 -100
cantenna 26 -74 -100 25 -76 -101
To calculate the gain of each antenna, the remote signal readings werenormalised, using the recorded results for the 24dBi Conifer dish.
The remote signal readings provide an indication of the transmit performanceof each antenna.
Note that the gain of the Conifer dish has been conservatively estimatedat 22dBi.
The normalised results for each antenna are:
horizontal polarisation:   antenna gain
(dBi)
Conifer 22
TM waveguide 17
RC waveguide 19
biquad1 12
biquad2 12
collinear 10
cantenna 13
vertical polarisation: antenna gain
(dBi)
Conifer 22
TM waveguide 14
RC waveguide 16
biquad1 10
biquad2 8
collinear 8
cantenna 12
Azimuth plots of the transmit performance have been generated foreach of the waveguides, showing the performance over360degrees.
Note that testing was only performed over 180 degrees, butsome quick tests indicated the results for the other 180 degreeswere within 1 or 2 dB.

azimuth plot of the TM waveguide
(normalised dB)

azimuth plot of the RC waveguide
(normalised dB)
Refer to Marcus‘pagefor more details on the waveguide test results, and additional azimuth plots.
These azimuth plots almost exactly match the theoretical azimuth plotswhich Trevor Marshall shows on hiswebpage.
Comments & Conclusions
By testing all the antennas in the same location, on the same day,using the same hardware at both ends, we‘ve attempted to minimiseany factors which may affect the results.
When calculating the normalised gain of the antennas, we assumed thegain of the 24dBi Conifer was 22dBi.
Our results generally agree with the antenna gains other people have reported.
Claimed gains for the biquads are typically 10-12dBi, while we measured 8-12dBi.
Interestingly enough, biquad1 measured 10-12dBi, while biquad2 measured 2dBiless when vertically polarised, despite having a stub (the mount between thereflector and the quad itself) which is impedance matched.
I‘d suggest these differences are due to the lips on biquad1, which wouldreduce the beamwidth of the antenna somewhat.
Various people have reported gains for the 8-element collinear anywhere between6-10dBi (6dBi for RG-213 construction, while my collinear uses CNT-400),while we measured 8-10dBi.
Similarly, people have reported cantenna with gains varying from 10-14dBi,while my cantenna measured between 12 and 13dBi in these tests.
The theoretical gain of the TM waveguide is 15dBi (according toTrevor Marshall),while commercially available 8-slot 180-degree waveguides aremarketed with gains between 14 and 16dBi.
Our results show the gain of the TM waveguide being 14-17dBi,and the gain of the RC waveguide being 16-19dBi.
The RC waveguide provides approximately 2dBi more gain than the TM waveguide,which is consistent withRob Clark‘stesting.
Note that the dimensions of the downpipe are slightly different than thealuminium tubing used by Trevor Marshall, but the slot spacing was notadjusted to compensate for this.
However, when constructing the RC waveguide, the measurements werere-calculated to suit the downpipe dimensions, and was adjusted to havea air column length of 5 wavelengths (while the TM waveguide has an aircolumn length of 4.75 wavelengths).
We intend to perform some more testing with the waveguides,and I‘ll update this page with the results, when they are available.
During our testing, we noticed that while the cantenna and theConifer dish have a very narrow beamwidth (Conifer specs say 7 degrees),the biquad have a very wide beamwidth.
With such a wide beamwidth, the biquad is extremely easy to aim,while antennas with a narrow bandwidth can be more difficult to aim.
References
Conifer (ex-Galaxy) Antenna Modifications:Modifying Confier Antennas for Wireless Networking
Conifer (ex-Galaxy) Antenna Testing (part 2)
Wave Guides:Marcus‘ Downpipe Waveguide Comparison
Trevor Marshall‘s Slotted Waveguide 802.11b WLAN antennas
Rob Clark‘s Downpipe Waveguide
MikeN‘s Slotted Waveguide Antennas for 802.11b Applications
Collinear Omni Antennas:Easy Homemade 2.4Ghz Collinear Omni Antenna
Biquads:Biquad Antenna Construction
Trevor Marshall‘s BiQuad 802.11b Antenna
Cantenna:WLAN Cantenna
Credits
Photos byMartin,azimuth plots byMarcus,other graphics byMartin.