Intermediate Frequency vs Baseband Signal Processing

[chingkui]

In the digital communication class I took, people talk about how to do detection and processing at the baseband. But from what I read, it seems that many systems do the signal processing at some intermediate frequency without mixing it all the way back to baseband. My questions are why and how:

1) Why would you want to process signal at IF band? wouldn't it be "easier" if you process baseband signal? And wouldn't the sampling rate be lower at baseband?

2) How exactly do you process the signal (such as detection, etc)? I know how you do it at baseband, but how do you accomplish the same thing with a signal oscillating at IF?

I have not yet been able to find the answer online and in books, I would greatly appreciate if anyone can point me to the right direction. Thanks.

 

[JakeBrodskyPE]

The answer to your question is that books don't usually discuss practicalities. Base-band processing may require significant gain across many octaves of bandwidth. This is not easy to do while keeping the power supply decoupled, and maintaining a good noise figure.

IF processing can handle a similar bandwidth with lower noise (the infamous 1/f flicker noise from many mixers is less at IF); it is easier to decouple the power supplies; and linearity is simpler.

So, unless there is a compelling reason why you want to do things at base-band (such as an A/D converter that you use for signal processing), IF processing is the typical way that many high performance receivers do their processing.

There are variants to IF processing such as using an image of the A/D converter, but that takes us down a rat-hole of discussions that people write papers and even books about.

 

[sophiecentaur]

It all totally depends upon the particular system involved. The huge advantage of using an Intermediate Frequency in a receiver is that all channels can, on an individual basis, be mixed down and filtered through just one well defined filter (perhaps a crystal or SAW filter). I.F. 'processing' would normally just be limited to this filtering and, perhaps level / gain regulation, followed by demodulation / detection. But, as I have already said, the choices of what happens when and where are made on a system by system basis - there is no universal rule. Companding or surround sound processing would have to be done at baseband.

I think you may have an inappropriate idea of what constitutes 'baseband'. A baseband signal would normally be thought of as a signal that could come directly from a sound or video source - or that could be fed to a loudspeaker or monitor.
In all cases but high power AM transmissions, the modulation would be achieved at a convenient IF frequency and then up-converted to a required transmit frequency. AM (mf) transmitters very often actually modulate the kWs of carrier right at the transmitter output with enormous modulation transformers and modulation amplifiers. Great scary devices behind safety screens. This is much more efficient than using linear amplifiers to generate such high powers.

 

[rbj]

i might suggest going to the USENET comp.dsp newsgroup and asking them. they'll help you flesh out the question. if it's a software radio, there are a few reasons for processing IF with DSP rather than processing the IF in analog.