What is baseline wandering in digital transmission?

[Geek007]

hi there,
can someone please explain what is baseline wandering and how the continues 1's and 0's cause baseline wandering in digital transmission?
Regards

 

[256bits]

Hi Geek007,

Best I can do is this, as I don't want to get too far ahead of myself.
http://pdfserv.maximintegrated.com/en/an/AN1738.pdf

Maybe somebody from Electrical Engineering Section can describe AC coupling and problems.

 

[analogdesign]

Because the transmitter and receiver often have different electrical requirements, sometimes you have to AC couple the signal, which means you put a series capacitor in the circuit that blocks the DC component of the transmitted signal.

Now, remember that DC is just the average value of the signal. Imagine you are using the simplest tranmission scheme, which is Pulse-Amplitude modulation, Non-return-to-zero (PAM-NRZ). This means the voltage is high for 1, low for 0. Also, say that in your transmission system the level for 1 is 2V higher than for 0, or as it is often described, the signal alternates between 1 and -1 V (relative to the DC level).

Some examples:
If you have an equal number of 1s and 0s, the average value is 0.5*1 + 0.5*(-1) = 0. Which is what you expect.

If you have 9 1s on average for each 0, your average value you be 0.9*1 + 0.1*(-1) = 0.8.

If you have all 1s, the average value is just 1.

As you can see, this is no good because the DC level at the receiver is data dependent and this causes all kinds of problems.

There are various ways to fix this. For example:

1. DC couple your TX and RX
2. Use a "DC balanced" line code. Basically encode the data such that 1s and 0s are approximately zero (even if they aren't in your real data). Examples of a DC balanced line code are Manchester encoding or 8b-10b encoding.
3. Use a DC restoration circuit. Basically a slow filter which holds the baseline as the data changes.

 

[anorlunda]

In more severe cases,you may even have to use frequency modulation for 1 0 encoding.

 

[sophiecentaur]

Manchester coding - AKA Biphase coding avoids variable DC offset by sending a 1 as 01 and 0 as 10 ( or the other way round) so each binary digit has a mean value of 1/2. It involves doubling the bit rate, of course, but the analogue bandwidth needed to transmit biphase need not be excessive.