I have been gathering information on Homebrew Wireless Antennas and have made a couple as an experiment. Basically its for accessing my wireless network from my garage which is some way off. My latest one out of a Mini Cheddars Can seemed to work well, I think. I have been measuring the signal strength using Net Stumbler but I am a bit confused and need some quick pointers.
I understand that signal strength and Noise is measured in dBm and think from what I have read on the Antenna Shoot out that the lower the dBm the better.
Netstumbler shows a graph with a left hand Axis that shows the Signal Strength and Noise according to the help file the signal strength should be in green( The higher up the graph the better, higher up the scale the lower the dBm reading) and noise in red if available. My graph only shows a green bar so I guess thats signal strength. Last night while testing my PCMCIA card only I got a signal strength of about 60 dBm when I attached the Antenna I got about 40 dBm (Wrote it down somewhere) if that is the case I just increased my signal dramatically.
I know it got better because when accessing my Network the wireless networks Icon increased from 36Mbps to 54Mbps.
Does anyone know of a software tool for bechmarking or measuring wireless signal strengths in order to accurately record results.
Mind if I not answer what you're doing wrong and simply tell you the right way to measure gain with Netstumbler? The problem is that most wireless cards simply do not have enough dynamic range to do a decent antenna pattern or gain measurement. The granularity also sucks in that the signal and noise levels are derived from an 8 bit (256 levels) value.
There right way(tm) is to use a reference antenna and an adjustable attenuator. The reference antenna can be any antenna with a known gain. It should be something with similar gain to the test antenna, but it really can be almost anything. The adjustable attenuator goes between the antenna and the receiver.
The test setup is important. It's important to eliminate any obvious sources of reflections and multipath. The worst offender is a roadway. There are web sites showing antennas mounted on tripods about 3ft above the roadway. The test result are fairly erratic, probably due to ground reflections. In addition, the midpoint Fresnel Zone hits the ground, which causes additional path problems.
It is also very important to set the over the air data speed to a fixed value. It can be just about any speed that gives a useable measurement, but it should change. That's because the receiver sensitivity varies with the speed. If the speed varies, the pattern and reference levels will be ruined.
What I do is setup my test antenna on my flat rooftop and aim it at a mountaintop about 4 miles away. The ground in between is a rather large forested valley, which offers few sources of reflections.
Once you have a repeatable test setup, just setup the attenuator for some arbitrary attenuation. I use about 10dB. Make VERY sure that the test radio does not pickup or belch RF directly. All of it should go through the antenna connector, not leak through the plastic box. If you're stuck with a plastic box test radio, wrap it in aluminium foil.
Setup a reference level using Netstumbler or preferably a passive sniffer such as Kismet. Record the level. Then, remove the reference antenna and replace it with the test antenna. Adjust the attenuator until Kismet or Netstumbler reads exactly the same value as with the reference antenna. The difference in attenuator readings is the gain difference between the two antennas. For example, if my reference antenna has a gain of 8dBi and the attenuator showed a difference of
+2dB gain, therefore the test antenna has a gain of 10dBi.
Measuring the antenna pattern is done exactly the same way. Rotate the antenna perhaps 15 degrees. Make the same measurement and record the attenuator reading necessary to get the same measured level with Kismet. Keep doing it all the way around with the test antenna. You should get something like the theoretical antenna pattern. The only added complications is that since the antenna is now pointing in all kinds of different directions, the opertunities for reflections are greatly increased. One test I helped run acted weired until we move the cars parked behind the antenna.
If you have any furthur questions, it would nice to know:
dB or decibels is a ratio. The 'm' on the end references the ratio to 1 milliwatt of power at the characteristic impedance. This ratio is actually expressed as ten times the logarithm of the power ratio - and a positive number means the ratio is larger than 1. A negative number would mean that the ratio is less than 1. As far as received signal (and noise) levels are concerned, it's rare to find signals larger than 1 milliwatt coming out of a receiving antenna, so the levels would be negative dBm. Now it seems that you aren't seeing that negative sign (trust me, it's there), but normal (usable) signal levels would be between say -90 dBm and perhaps -20 dBm. To put those numbers into absolute power levels, -90 dBm is 0.000 000 001 milliwatts, while -20 dBm is 0.01 milliwatts. Thus, a smaller negative number is a stronger signal.
Take those numbers with 10 kilos of salt, but yeah. Again, those are supposed to be negative dBm.
OK, so you went from -60 dBm to -40 dBm, which is a 20 dB improvement. Where those numbers reported with a calibrated device, that would be quite good. In the direction the antenna is pointing, that's the equivalent of a 10:1 increase in range capability in free space (which really means a path free of reflections, trees, dogs or what-ever).
A software tool is only reporting what the perceived signal strength is, based on what the hardware is saying. You're assuming that the hardware reports bear some resemblance to truth. This is HIGHLY unlikely. The numbers you see are a best guess - and may be absolute fables. You could connect the receiver directly to a calibrated signal generator, and record the indications while varying the signal strength (in effect calibrating the receiver). Another way would be to connect a calibrated variable attenuator to the receiver set things up with strong signals, and then vary the attenuator setting while recording the indications. In either case, you are only going to be as accurate as the calibration of the signal generator or attenuator, and that is only true at the same temperature, supply voltage, and may vary as under the influence of the color of the moon, and the number of days since you last paid your bookmaker.
The test range is about 600ft long. At 2.4GHz, the Fresnel Zone is about a 7ft radius at midpoint about the center line. For this test range to work, all the test antennas would need to be elevated at least 7ft off the ground. That obviously wasn't done.
Nope. I can't recommend an antenna without knowing what you currently have in the way of hardware, and what the neighbors or hot spot has that you're trying to mooch off has in the way of hardware. Just calculate the minimum necessary gain and the type of antenna will be rather obvious. See: |
formatting link
the link calculations. If there are any obstructions in the line of sight, it probably will not work.
Whilst carrying out some more recordings the N Connector connected to the Pigtail connected to the PCMCIA card fell off of the table ripping the mini connector out of the card hey presto a broken pigtail and maybe a broken PCMCIA external Antenna connection !!!!!!!!!!!! I knew I should have cable tied it to the Laptop
Now plugging in the Homebrew Antenna the signal strength drops!!! Oooops
The pigtail was bloody expensive as well for what it was, at least the cards internal ariel still works!!
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.