Companding and Non-uniform Quantization.....

[Rohit Chatterjee]

What is the difference between companding and non-uniform quantization???I believe both are utilized to increase the number of quantization levels when the signal amplitude is small,but both are entirely different concepts....correct me if i am wrong.....

 

[Warpspeed]

A very subtle difference !

With companding, the system gain is controlled by an externally applied control voltage, which usually goes into some kind of analog multiplier.
Where that control voltage comes from, and how it is derived determines if the incoming amplitude dynamic range is expanded or compressed.
Companding is often used for audio applications.

Non linear systems are often typically logarithmic in response to amplitude.
The human ear is like that.
Some applications (like radar) you might want to compress a very wide dynamic range in incoming amplitude to something with much less dynamic range that can be more easily seen on a radar screen.
You don't want weak signals to be so faint that they are invisible, or super strong returns to be so bright they overload the display with maximum brightness.

Similar thing with a gas chromatograph, You want to reliably detect trace amounts of some compounds, but don't want to be driven way off scale by massive concentrations of other
things. So your vertical scale on your chart might be made logarithmic.

With companding, the gain slews up and down over time, perhaps controlled by some feedback signal, but the system remains linear with no amplitude distortion.

With a non linear system, the signal amplitude itself determines the gain. It is instantaneous and very convenient for many things, but it does introduce amplitude distortion into the signal.

 

[Rohit Chatterjee]

The part about non-linear system is quite interesting....but i didnt quite get the rest....
As far as i know,the step size is reduced when the signal amplitude decreases,keeping the number of quantization levels same....is this change in step size what u r referring to as the gain of the system????

 

[dougy83]

Companding is a contraction of Compression and Expansion. It may be applied to an analogue signal; it does not necessarily have to be quantised and may remain analogue and continuous-time.

Non-linear quantisation obviously involves quantisation. Quantisation is done using differing step sizes. e.g. for audio, smaller step sizes are used for smaller signals which allows the audible quality to be increased for signals having a large dynamic range.

By using a companding (well the compressing part of it) front-end to a linear quantiser, a non-linear quantiser is effected.

@Warpspeed, I think what you were describing is ALC/AGC (although, I have heard it called compressing also - the compression occurs when the amplifier is reducing its gain to limit the output signal - the compression being of the long-term gain of the system/amp, whereas companding is applied directly to the instantaneous value of the signal). I think companding consists of non-linear distortion of the signal during compression and the reverse distortion during expansion.

 

[Warpspeed]

Doughy, yes, AGC, ALC and compression are pretty much the same thing. Expansion is the inverse function.

Non linear systems that that I have seen, generally have all been analog in nature using true logarithmic amplifiers, but they can also be digital, where the bit step size varies over the whole input range.

Non linear quantization is much more difficult to do directly at the analog to digital conversion process itself, and is not commonly used for high performance systems.

Far more common is to have a logarithmic amplifier followed by a standard a/d converter. This effectively decreases the bit step size at low amplitudes, and increases the bit step size at high amplitudes.

 

[Rohit Chatterjee]

Companding is a contraction of Compression and Expansion. It may be applied to an analogue signal; it does not necessarily have to be quantised and may remain analogue and continuous-time.

Thats what i was referring to-compression,followed by quantization(uniform) and finally expansion to get the digital signal from the analog signal....So in this case the step size is unaltered while the signal undergoes non-linear distortion....And in case of non-uniform quantization the signal is unaltered while the step size varies as the signal amplitude varies....Right????Did i miss anything????

 

[Rohit Chatterjee]

No one.......
please i need to clarify ceetain doubts......

 

[Warpspeed]

That is rather difficult old chap, if we have no idea what it is you doubt.

 

[Rohit Chatterjee]

Thats what i was referring to-compression,followed by quantization(uniform) and finally expansion to get the digital signal from the analog signal....So in this case the step size is unaltered while the signal undergoes non-linear distortion....And in case of non-uniform quantization the signal is unaltered while the step size varies as the signal amplitude varies....Right????Did i miss anything????

The former case refers to companding and the latter to non-uniform quantization.....

 

[rogs]

An example of non linear quantisation can be found in the telephone world, where 'mu' law or 'a' law compansion is used to improve the signal to noise ratio of digitised telephone signals.

This is a pretty good place to start for more info:

?-law algorithm - Wikipedia, the free encyclopedia

I always tend to think of companders in the analogue mode. This is a good example:
http://www.datasheetcatalog.org/datasheet/philips/SA571.pdf --

I use that device quite a lot, as a simple way of significantly improving the quality of 8 bit digitised audio signals, in the analogue domain.

I usually think of non linear quantization - like 'mu' law and 'a' law -as a digital process, where the use of higher resolution digitisation of lower signal levels, and lower resolution at higher signal inputs can improve the overall quality of a telephone signal quite dramatically, for example.

In fact 'mu' law and 'a' law can be implemented as an analogue process as well, but the digital version better describes you non linear quantisation question.

 

[Warpspeed]

Non linear pre processing before linear digitisation is a reversible and definable process.

Companding with some type of variable gain LINEAR amplifier that adjusts with some time constant is not reversible, because you have absolutely no idea what the original amplitude was.

Non linear pre processing is ideal for instrumentation and measurement. You can say that so many output volts are equal to some input quantity at the source.

Companding is ideal for something like audio processing, where you are just listening to some sound, but could not care less how many millivolts or microvolts there were at the microphone during the original recording session.