|
bluedog
WiFi Guru
User
| Posts: 35 |   |
 |
Re: beamwidth and gain testing of some antenna feedpoints |
- 2006/06/11 01:07 |
|
#2 |
While I had this antenna testing range set up I decided to test the available wireless cards and measure their weak signal performance.
Note that I did not perform throughput tests or transfer any other data except for link esablishment purposes during this test, as it was primarily to establish if a card could sniff out a very weak access point using netstumbler 040.
The cards available were :
Cisco AIR-LM4800
DLink DWL-650+
Skynet/cabletron roamabout
I decided on 2 tests , the first being a flat-out weak signal detection test and the second a adjacent channel interference test.
Results:
Short cable test (4 metres of Lmr400)
All 3 cards recieved the weak signal source using the doublequad antenna and the displayed signal to noise ratios in netstumbler showed the weakest to be the roamabout/skynet adapters. The card showing the best signal to noise ratio was the cisco , however the link was unstable and there were occasional gaps. The Dlink card showed a stable unbroken link at slightly lower s/n ratio.(knowing how difficult these cards can be peer to peer it surprised me)
Long cable test: (12 metres of tandy RG8 20 years old)
Only the Dlink card received a signal when using the cantenna/coppertenna.
On the doublequad feed the Dlink was achieving a link while the other 2 received broken signals via the lossy cable.
The weak signal from the Dlink took a few seconds to aquire, and the link was barely usable. there was no link at all from the other 2 cards.
Adjacent channel test:
I decided to test the front end performance of the cards when exposed to strong local adjacent channel interference while linked to the very weak access point.
This will prove or bust the notion that strong adjacent stations kill links.
i decided on channel 6 as the weak test AP, and channel 11 as the strong AP. With stumbler showing a weak link using the dlink and the cantenna feed i applied a channel 11 AP and antenna 8 inches from the mouth of the feedpoint! pretty extreme , but the weak link held up with minor gaps, suggesting the signal level was the same but the data would be corrupted resulting in many retries
The same experiment was performed with each card with the same result, they all held a link except it was broken up a bit. So the rf part holds up , but the data transfer part probably retries a lot resulting in poor throughput.
its a work in progress , but someones got to prove/bust these things we read on the net
cheers
its a dogs life but someones got to do it |
.<br />31mm side dimentions as per standard design, brass plate as a reflector,N connector on rear<br /><br />Cantenna,coppertenna tubular waveguide with feed pin.<br />heavy walled copper tube, dimensions 8 inches long, 2.1 inches to feed point and 3.6 inches inside diameter.N connector on side<br /><br />Results : <br /><br />Signal level from the access point was dropped to extremely weak by adjusting the microwave oven door to virtually closed. The beamwidth of each feed antenna was then measured, the widest beamwidth was the cantenna, with the doublequad showing noticable better signal aimed straight onto the AP tapering off either side to about the same as or worse than the cantenna.(note there was nothing received by the dipole)<br /><br />This indicates that the cantenna is likelly to be the best feedpoint for a large dish or grid with short F/D ratio, also good for local area widebeam links by itself.<br />The doublequad has a primary lobe in the middle of its pattern indicating it would be better as a point to point short link antenna and as a feedpoint for a offset fed antenna with the outside of its main lobe still close to the cantennas performance. That will be tested on A band when time permits.<br />[img])