Forum

Discriminating two different wavelengths

5 posts by 4 authors in: Forums > CWTS - Enterprise Wi-Fi Fundamentals
Last Post: September 20, 2010:
• Apologies if this appears somewhat basic... but this has been bugging me. How on earth can a receiver discriminate between two different wavelengths, when as they are transmitted the resulting amplitudes seem combined? I mean in order to do this wouldn't you need a understanding of the surrounding interference to determine the actual signal (or in other words the original wave) you need to 'tune into'? This is basically related to the EMANIM program where it gives the combined wave. Thanks in advance, Nik. Probably missed something fundamental... :?S

• Nik

You have raised an excellent question. It is actually quite a complex process, and I won?t go into all the details here, as it gets very mathematical.

When you transmit a signal from a Wi-Fi radio card, we say that the carrier signal has been modulated. In most documents and books you will see the term ?carrier? referring to the modulated signal. That?s not strictly speaking correct. The signal is a MODULATED carrier signal, not really just a carrier.

However, in a similar manner to the term ?bandwidth in bits/s? not being correct ( bandwidth is measured in Hz ), ?popular use? has created these expressions and we are stuck with them.

Let?s say we are transmitting a QPSK signal from A to B at say, 5170 MHz. The transmitter at A and the receiver at B are both ?tuned? to 5170 MHz. What does this mean ?

At the receiver, an electronic circuit performs what is called ?carrier recovery?. This is actually a very clever process, as there can be frequency variations in the transmitted carrier due to poor quality transmitter electronics, variations in temperature, doppler effect ( very important in high speed vehicular systems etc ).

At any instant of time, the received signal?s center frequency is extremely unlikely to be exactly 5170 MHz. There will almost always be an offset. It is the function of the carrier recovery tracking loop to ?track? or follow the variations in the incoming carrier.

Once the carrier has been extracted, a bit timing recovery loop is implemented in order to extract timing information for the subsequent demodulation process.

Provided the receiver has been programmed with the correct value of frequency to receive ( you can?t tune to two frequencies simultaneously in one radio ), it should be able to ?discriminate? the correct value. Sometimes a device called a ?phase-locked loop? is used for this function ( you can find these devices in car FM radios for example ).

So, now we have the carrier signal properly recovered, what about all the other frequencies that we see ?surrounding? the carrier ? That is where the rest of the demodulation process takes place, and unfortunately I don?t have time to go into it in detail.
Another pointer to the issue of the ?combined amplitudes?: Using a process called Fourier Transformation, a complex waveform made up of a number of base frequencies ( ?looking like a jumble on an oscilloscope? ) can be processed by modern digital signal processors and the individual components extracted.

This is just a very basic overview to what is in fact an extremely complex process.

Dave

• there are two major ways to convert a signal in an RF circuit to a visual representation

the method Nik is thinking of is an oscilloscope. an oscilloscope has a mechanism that draws a horizontal line across a screen, and a separate mechanism that moves that line vertically. the end result shows voltage over time. if you looked at a cable tv coax or an 802.11 antenna the result would be a chaotic jumble.

in old shool analog scopes, the mechanism that created the horizontal line was a sawtooth waveform:

http://www.societyofrobots.com/member_tutorials/files/sawtooth.png

one sawtooth = one horizontal sweep. the ramp is linear, so there is a fixed relationship between time and voltage. the result is X milliseconds per CM of screen motion.

this is called "time domain"

the second method of viewing waveforms is "frequency domain". we view signals in the frequency domain with a spectrum analyzer. you feed the sawtooth to both the horizontal trace of the scope and a radio tuning circuit. you tune the radio across a band of frequencies with the sawtooth. you measure the voltage on the circuit, and rectify it so all the voltage goes positive on the screen. the resulting display is in mHz per CM.

the result looks like this in raw form:

http://www.metageek.net/files/images/usinginssiderwithchanalyzer.img_assist_custom.png

and like this cleaned up:

http://img18.imageshack.us/img18/7035/inssider.png

here is a spectrum analsis of an HF radio waveform:

http://www.bitscope.com/software/dso/32.png

on an oscilloscope this would be a jumbled mess. on a spectrum analyzer we see distinct and orderly grouping of RF energy. this is what the radio sees.

the vertical row of green on the bottom of the screen - "grass" - is noise. the height of the spike above the noise, measured in decibels, is the signal to noise ratio.

Trivia: the sawtooth waveform is the waveform of a violin string in free air ( without the body beneath it ). The bow drags the string to one side, until the tension on the string exceeds the traction of the bow. The string slides across the bow until the bow gets enough traction to drag the string to one side, until the tension on the string .....

• Nik

There are actually two issues here:

Firstly, the physical representation of a signal ( single frequency or composite, mixed waveform ) by means of either a time-domain or frequency domain test instrument ( usually an oscilloscope or spectrum analyzer respectively ). Both these devices will allow us to ?see? the signal displayed on a screen. The previous post shows some good examples of this.

In other words we can ?see? one of more signals plotted on a screen. By means of filters and sometimes software , these devices can ?show? the signal or signals.

The oscilloscope simply plots signal voltage versus time ( voltage can be plotted on a decibel, dBmv or linear voltage scale ). In order to see more than one signal we need to have a multi-channel input system.

The spectrum analyzer plots amplitude ( usually on a log scale ) versus frequency.

The frequency range of both instruments ( whether ?hardware? or ?software? based ) is very important.

Secondly, you asked the question about how an actual receiver as opposed to a test instrument can discriminate between two or more frequencies. As mentioned before, that is an excellent question, but unfortunately requires quite a bit of demodulation theory to understand properly.

You used a very good term in ?discriminate?. In FM theory the term ?discriminator? was actually used for the demodulator.

As you can see from the following, it gets quite complex

http://en.wikipedia.org/wiki/Frequency_modulation

http://en.wikipedia.org/wiki/Phase-shift_keying

In a nutshell:

Theory of displaying a single or composite frequency waveform on an oscilloscope or spetcrum analyzer ?..not so bad.

Theory of frequency discrimination in a receiver ??a bit more trouble.

Unfortunately in RF, physical explanations can sometimes only go so far ( isotropic radiator , big expanding ballon etc ), then the "'orrible math" kicks in.

An anology would be with speech. Imagine two tuning forks sending out waveforms and we display the sound waveforms on an oscilloscope via a transducer ( earpiece for example ). Now imagine two people talking together and viewing the result.

I think we can imagine that we could ?see? these waveforms.

Now imagine the situation where we have to explain how the brain ( the ?receiver? ) is able to discriminate between the two sounds and also in the presence of interference ( a radio station blaring above the two ). I think you can see that this becomes more complex.

Even in the case of the oscilloscope and spectrum analyzer, how the devices actually perform the discrimination requires quite a bit of theory ( as opposed to how the physical display operates ).

I have tought spectrum analyzer courses for over twenty years, and while the basics of how to use them ( ranging from little Anritsu handhelds to \$40,000 dollar devices ) can be taught in a few hours or a day, the theory behind them takes longer. Not everyone needs to know the theory, but it can produce a better troubleshooter. I can think off-hand of at least half a dozen cases where an understanding of the theory helped me resolve some problems that had effectively "shut down" satellite links carrying ( sometimes ) an entire country or island's communications.

Dave

• Give this a look: http://electronics.howstuffworks.com/radio1.htm

Page 1 of 1
• 1