Noise Reduction System: Preemphasis/Deemphasis Filter

[Swapnil]

I know that purpose of the preemphasis filter in a noise reduction system is to increase the amplitude of the input signal at frequencies where SNR is low prior to transmitting the signal. I also know that a deemphasis filter is located at the receiving end and its purpose is to undo the effects of the preemphasis filter by decreasing the signal amplitude back to its original size.

I get the general idea but I have don't know how all of this fits together. So I was wondering could anyone please tell me some specifics/details on how this type of noise reduction system works and maybe just expand upon what I said above?

 

[chroot]

Imagine that a channel has a hard time transmitting some frequencies -- meaning that the channel greatly attenuates those frequencies. When you pass a signal of that frequency through the channel, the receiver only receives a very small amount of energy. The noise in the channel might be relatively constant over all frequencies, though, so the signal's SNR is significantly reduced -- at least, compared to that of another signal of another frequency at which the channel is more gentle.

If you can pre-emphasize those frequency components that the channel does not transmit well, you can maintain the same SNR across all frequencies.

- Warren

 

[Darryl]

i believe its only used in FM, (my theory is OLD..).

ive always read it as:.

there is more information in the higher frequecies of the audio spectrum, but the modulation techniques used. put more power into the lower frequencies.

so preemphesis is used to boost the higher frequencies or the audio spectrum, and attenuate the lower freqs. making the power spectrum of the signal more linear with the frequency spectrum.
(as i said its been awile).

for amplitude modulated signals i believe the same effect is achived with compresson and expansion. (AM and SSB). reducing the dynamic range i guess in the goal.

and if Preemphises is used for FM modulation, u can assume probably VHF or higher, and its generally not SNR that is the problem. its fidelity.
FM is not a good mode for high noise low signal condition.

but my theory is a bit rusty.

 

[mdelisio]

My take is that you use pre/de-emphasis (PDE) to improve SNR. In general, if you know the spectral characteristics of your "message", the transfer function of your channel, and the noise spectral characteristics, you can come up with an optimum PDE filter pair that maximizes SNR at your receiver.

In general, PDE is used in broadcast FM. The big problem with FM is that your information is in the frequency, which is the time derivative of the instantaneous phase. This differentiation in the demodulation accentuates the higher frequency noise (it gives it a f^2 dependence). If you've ever had to take a derivative of a noisy signal, you know exactly what this means. At the high end of the spectrum, most "messages" (music, speaking, etc.) have very little energy, as explained below. You might be tempted to just filter off the upper end of the spectrum, but the ear will detect this as a loss in fidelity. Try turning down the treble control on your stereo --- many people describe the sound as "muffled" or "muddy". PDE gives the high-frequency message signals a boost so as to be able to be heard above the accentuated high-frequency noise.

I believe the actual PDE filters used in broadcast FM are not strictly optimum, but were chosen such that they can be realized inexpensively (like a simple single-pole RC filter on the receiver for de-emphasis). Nevertheless, the SNR improvements can be considerable --- I'm guessing around 10 dB.

I think PDE is also used in some of the noise reduction techniques used in tape recording (like Dolby, for example) to minimize high-frequency tape "hiss".

To dig deeper, I would suggest some EE communication texts, like Lathi or Haykin. Pierce and Posner is also a real gem, although a bit dated.