7

I was just going through a text book on digital communication where something related to ac coupling was mentioned.It was mentioned that if a long stream of one(1) is encountered at the receiving end then an erroneous bit is supposed to occur (sorry for not being able to quote the text).So for that purpose they are somewhat DC balanced in order to avoid DC WANDER.

I tried to look for it but did not get a proper, thirst quenching response. Is it possible to get a conceptual explanation for what they mean by DC wander?

Kortuk
  • 13,412
  • 8
  • 62
  • 85
nvade
  • 123
  • 1
  • 6

3 Answers3

7

Visualisation aid:

  • "All models are wrong.
    Some models are useful" ... George Box.

Imagine feeding a signal through a capacitor.

If you send AC (sine or square waves etc) with a mean DC level of zero, then the signal passes through it unimpeded.

If you apply DC to a capacitor it will charge exponentially to the driving voltage and current flow will fall exponentially to zero - a capacitor does not pass DC..

A comms channel of the sort you are dealing with is effectively an AC coupled circuit. You must maintain the AC nature of the signal so that it does not "appear to be DC".

Russell McMahon
  • 150,303
  • 18
  • 213
  • 391
  • It indeed,seems to be a nice comparison for comprehending the concept.But it would be rather satisfying if you just elaborate a bit on dc wander.Thanks a lot – nvade Aug 16 '12 at 05:06
6

Another kind of visualization aid. A drawing... before and after waveforms showing AC coupling effects

The original signal is at the top. The second line shows the same signal after the "AC coupled" treatment. With no changes, the dc voltage decays back to zero. When that low-going pulse comes through, it appears as negative voltage. This is probably not what your detector was expecting to see.

gbarry
  • 8,745
  • 22
  • 32
1

DC wandering is a common phenomenon in NRZ-L based data streams. (In NRZ-L systems ZERO is 0V, and a ONE is 5V or viceversa). If we have too many ones back to back (or too many zeros if our data is defined as 0 being the 1V or 5V) in AC coupled systems, the back to to back ones look like a DC line as the drawing above suggests. Due to capacitive coupling between the circuits (AC coupling means they used capacitance to connect two circuits) following circuits (if it is a detector) stop operating or start evaluating all the data as zeros. This would not happen if the data is BiPhase-L (such as Manchester coded data). However, the Manchester code uses twice the frequency bandwidth as the data levels change right in the middle of the bit sequence (a ONE could be a change from 5V to 0V and a ZERO could be a change from 0 V to 5V). In NRZ systems a period of data is two bits (could be same binary bit or opposite).

AC coupling is desired if there is a bias voltage (DC) used to turn on a gate quicker, or any other reason, in the previous stage of the circuit. Usually low bit rate data circuits do not prefer to use AC coupling due to long periods of ones and zeros.

Hal
  • 11
  • 1