# Forum

## Understanding OFDM Part 4

4 posts by 2 authors in: Forums > CWNA - Enterprise Wi-Fi Admin
Last Post: November 18, 2009:
• Another excellent tutorial from Rick Murphy. For those new to RF, don???¡é?¡é?????¡é???¡ét worry if you don???¡é?¡é?????¡é???¡ét ???¡é?¡é?????¡­?¡°catch???¡é?¡é???????? all the concepts at once. Learning RF takes time. I???¡é?¡é?????¡é???¡éve been at it for 25 years and learn something new all the time.

The two most important concepts to learn from this tutorial are:

1. When you have a wide bandwidth input to your receiver, you receive more TOTAL noise than when you have a narrow bandwidth receiver. So how wide is wide and how narrow is narrow ? At the input to the receiver [ Wi-Fi AP or STA ] we have filters which filter out what we want [ our signal ] from what we don???¡é?¡é?????¡é???¡ét want [ other channels, noise etc ]. We can???¡é?¡é?????¡é???¡ét have the filter too narrow, because all signals ???¡é?¡é?????¡­?¡°drift???¡é?¡é???????? or vary in frequency a bit. We have to have a ???¡é?¡é?????¡­?¡°window???¡é?¡é???????? that allows for that drift. Conversely, we can???¡é?¡é?????¡é???¡ét have filters that are too wide, or we end up ???¡é?¡é?????¡­?¡°taking in???¡é?¡é???????? too much noise etc. There are other factors as well, but me fingers is sore.

2. Noise Figure is the ratio of Signal to Noise ratio at output of device / Signal to Noise ratio at input to device. When we pass a signal through an electronic device, it gets amplified, but SO TOO DOES ANY NOISE INPUT. If nothing else happened, the Signal to noise ratio at the output would be the same as the signal to noise ratio at the input. True, the output signal level would be higher than the input signal level and the output noise level would be higher than the input noise level, but the ratio of S/N at the output would the same as the ratio of S/N at the input. Unfortunately, real life is not like that. As the RF signal passes through the electronics of the receiver, It picks up internal noise from the motion of the charged carriers in the transistors etc. This adds to the total output noise. So how do we reduce this internal noise ? Well older electronic devices used things called bi-polar transistors. These have two charge carriers. Then FET???¡é?¡é?????¡é???¡és were developed [ Field Effect Transistors ]. These devices use only one type of charge carrier, dramatically reducing the internal noise of the device. The next method is to physically cool the device down. When NASA wants to receive signals from far away probes going to Mars etc, it uses giant dish antennas [ parabolic antennas, similar to your outdoor link types ].

http://deepspace.jpl.nasa.gov/dsn/antennas/34m.html#BWG
http://ed-thelen.org/ATA/HatCreekATA.html

http://www.astrosurf.com/luxorion/qsl-mars-communication2.htm

They physically cool down the amplifiers using liquid helium which drops the temperature of the electronics down to almost absolute zero. Electrons move much less ???¡é?¡é?????¡­?¡°frantically???¡é?¡é???????? and the noise drops way down. One of my first jobs was working on these refrigeration systems. These systems are very expensive and difficult to maintain. Fortunately, our Wi-Fi systems can tolerate higher noise levels due to the fairly short distances involved as the signal levels are still relatively high. Due to the fact that the electronics develops noise [ thermal noise, shot noise, intermodulation noise for mult-channel radio receivers etc ] the output S/N can never be higher than the input S/N.
The noise figure tells us by how much the output S/N differs from the input S/N.
When chosing between different AP???¡é?¡é?????¡é???¡és, with all other factors being equal, always go for the unit with the lowest noise figure.

Dave

• Nice follow-up explanation, Dave. I especially like your references to the "giant dishes" and helium cooling. But your last sentence says it all; "When chosing between different AP???¡é?¡é?????¡é???¡és, with all other factors being equal, always go for the unit with the lowest noise figure."

Many times I get asked to justify the reason why Enterprise-grade APs and client adapters are so much more expensive than SOHO-grade gear. One of the main reasons is the quality of the electronics. More silicon (or better silicon) results in quieter radios. You can tell which ones have the better electronics by reading the manufacturer's published noise figure statements. The lower the noise figure, the better the gear.

• Tks Murph

Yes, noise figure is a parameter that is very often ignored in the selection of an AP, yet it is crucially important. Receive sensitivity is tied to this parameter [ along with other things ].

When an RF wave is converted into an electrical current and passed through any electronic device, impurities in the semi-conductor matrix "cause" noise by a form of collision mechanism.HEMT's [ High Electron Mobility Transistors ] allow a "signal" electron to travel further without a "collision", leading to less internal noise.

The problem is that in order to purify the semiconductor matrix, very expensive procedures have to be used.

Essentially you have to take a rod of semiconductor material and pass it very slowly through a special oven. This process allows impurities to be drawn to one end, where they are physically removed from the rest of the rod. The more times you do this, the less the impurities, the less the number of ???¡é?¡é?????¡­?¡°collisions???¡é?¡é???????? [ actually a very complex process ???¡é?¡é?????¡é?€?? I use the word ???¡é?¡é?????¡­?¡°collision???¡é?¡é???????? lightly here ], the less the number of ???¡é?¡é?????¡­?¡°collision???¡é?¡é????????, the lower the noise figure.

Dave

• For those new to RF, you may be thinking "If that device actually degrades the S/N ratio [ actually a C+N/N ratio...but I'll leave that for another post ] what on earth use is it ?"

What happens is that a device called a demodulator lives inside the AP [ and your WiFi client card ]. This device "strips" the modulation from the carrier [ or carriers ]. In order for it to operate at it's most efficient point, we need to adjust the input signal level to the correct value. This correct value is called the "middle of the dynamic range". think of it as a "sweet spot". Not too high in level, not too low in level. If the signal becomes too high we get very complex compression effects ocurring which are not good [ on a transmitter, overdriving the amplifier will cause an actual physical change of the signal shape - another reason why it is better in many cases to use a higher gain antenna instead of just "cranking up the power" ].

At the "front end" of every AP, we have amplifiers, filters, variable attenuators and downconverters [ devices that change the SHF input frequency 2.4 Gigs, 5 Gigs to a more suitable value for the demodulator]. Signal levels are very important, and amplifiers are very necessary.

Even though an amplifier actually degrades the "S/N" ratio, we need them in order to boost the signal level when necessary to help get the received signal level in the middle of the dynamic range.

This all usually carried out in a closed loop dynamic AGC [ Automatic Gain Control ] system, and the actual operator does not have to do any physical corrections.

Similarly due to poor quality oscillators, frequency drift due to temperature changes etc, doppler shift [ yes, we'll have to take that into consideration for the new WAVE stuff ], an AFC [ Automatic Frequency Control ] system is used as well.

Dave

Page 1 of 1
• 1