clock phase and noise

[sonaiko]

if i have a digital signal that i want to recognize, i sample in the middle of the bit to know its a zero or one.

sampling in the middle can be done using a local clock that is 5 times faster than input rate, which results in each bit contains 5 local clocks and u can sample at the third local clock.

if there is sm amount of noise in my signal, it may result in 4.9 local clocks per bit which means an error in sampling after a certain amount of bits..

I want to draw a graph of my bit error rate vs noise/power for such a system. how?

if i have 4.9 clocks instead of 5, thats 2% noise. how much shall i consider No in this case?

thank you.

 

[blueroomelectronics]

Huh? What are you getting at? Why are you oversampling a digital bit? Is it Manchester encoded?

 

[sonaiko]

Huh? What are you getting at? Why are you oversampling a digital bit? Is it Manchester encoded?

no its NRZ.
Im oversampling becoz there is some kind of jitter on my received signal..

Since I am only simulating and I am generating my own digital signal with noise, I want to draw a graph of BER vs No/Eb. Where Eb is energy per bit. But I am not sure how to calculate No in this case..

 

[Mikebits]

Wouldn't that be Eb/No ? So your case is you either oversample by 1 bit or undersample by 1, I don't remember the math but I think it is about 7.5 dB of noise which might be significant.

Not sure if I worded the over/under sample thing right. What I mean is Noise might cause 1 of 5 to miss or hit a sample.

 

[sonaiko]

Wouldn't that be Eb/No ? So your case is you either oversample by 1 bit or undersample by 1, I don't remember the math but I think it is about 7.5 dB of noise which might be significant.

Not sure if I worded the over/under sample thing right. What I mean is Noise might cause 1 of 5 to miss or hit a sample.

Yeah exactly.
If every 25 bits I am oversampling by 1, then how would I graph this information into BER vs No/Eb?

 

[sonaiko]

Actually the question is:

Since I care about the baseband section of the receiver, i.e after demodulation and down conversion of the received signal,
Do I really need to care about the BER vs Eb/No?
I think the BER graph depends on the modulation and doesnt change when changing the decision techniques used in baseband, does it?

 

[Mikebits]

Well, if there is probability for error, then error would cause bit error even in baseband. To be honest, this is not my area, but I can recommend a good resource.
Digital Telephony by John Bellamy. Section 4 contains all the juicy math and such for BER. My edition is 2, but there is a ed 3.
The book shows BER Eb/No graphs.