| |
G6CRV > WLAN 29.09.03 08:08l 63 Lines 3020 Bytes #999 (0) @ WW
BID : 63097-GB7CRV
Read: GUEST DB0FHN OE5RCO DK5RAS DO6NP
Subj: RE: 2.4Ghz rx sensitivity.
Path: DB0FHN<DB0RGB<DB0IGL<DB0FSG<I4UKI<IK5SKV<IW5CWB<IK5CKL<IK1ZNW<GB7CRV
Sent: 030929/0518Z @:GB7CRV.#16.GBR.EU #:63097 [Heysham] XFBB7.04j $:63097-GB7C
From: G6CRV@GB7CRV.#16.GBR.EU
To : WLAN@WW
The following is a quote from Glen Elmore, one of the microwave guru's
who has had amateur high speed links in operation for years. :-
-------------------------------------------------------------------------
Thermal noise level is expressed as K*Tsys*B where K is Boltzmann's constant,
Tsys is the effective system temperature (mainly a combination of antenna
temperature [electrical not physical], feedline and input circuit losses
and receiver noise), and B is the bandwidth in Hz. For Tsys=300 Kelvin,
more or less room temperature and a reasonable starting guess for typical
system temperatures, in a bandwidth of 1 Hz this is about -174 dBm.
Center frequency is not an issue, what is of concern is the information
or channel bandwidth. For 802.11b,g,a with a channel bandwidth on the order
of 20 MHz, this number is 20 million times, or 73 dB, bigger and the noise
is about (-174 + 73 =) -101 dBm.
Receiver sensitivity usually relates to the minimum signal level for which a
receiver can properly functionreceiver can properly function, in a digital system this is often described
in terms of some bit error or frame error rate.
However, if that were all that mattered, amplification could simply be added
to a "poorer" receiver to make the signal big enough to allow it to operate.
Instead, what is normally the case with decent receivers is that the
limitation is on input signal/noise ratio rather than signal level.
Adding further amplification to an already low noise receiver doesn't
improve this ratio, only the absolute levels, which doesn't improve the
performance.
For 802.11b systems running at 11 Mbps, the input signal/noise ratio
(usually referred to as C/N for carrier/noise ratio) this ratio must be
greater than about 7 dB. For 11a systems at 54 Mbps it needs to be almost
25 dB. This means that the lowest signal a perfect 11b receiver could receive
would have to be about 7 dB above the noise
(on the order of -100 dBm from above) or -93 dBm. This level is the limit.
In actuality there are cablIn actuality there are cable losses, other noise sources and receiver
imperfections that probably make the reality ~10 dB worse than this so that
one probably doesn't find 11b hardware capable of going much below
-83 dBm sensitivity at full rate. If a manufacturer claims much more it
is time to be skeptical.
For 802.11a systems, things are roughly the difference of the required
C/N worse, (25-7)=18 dB and levels in the vicinity of -65 to -70 dBm are
probably the best sensitivities that will be common.
With all three standards, 802.11b, 802.11a and 802.11g, the slower fallback
rates require less C/N and these can operate at somewhat lower C/N.
This is the reason for having them.
I guess this was the long answer to your question. It's channel bandwidth
that should be considered and usually C/N rather than sensitivity.
regards,
Glenn Elmore
n6gn
73 de Dave G6CRV SysOp GB7CRV
g6crv@lurpac.lancs.ac.uk
Read previous mail | Read next mail
| |
