SNR Margin and Speed

I have searched everywhere for an answer to this question and have had no joy - so I am hoping someone on these forums can help:

Why do higher speeds require higher snr margins? In other words, why are higher speed connections less tolerant of noise than slower ones?

Thanks

They don't!

You may think they are doing more work, but as long as the margin for any noise to rise is sufficient, usually 3 - 6 dB, it will be suitable for any speed.

The SNR margin is the level of noise which is "acceptable" on the line.

On any given line the SNR comes down as the speed goes up with everything the same.

It is not that higher speeds are less tolerant of noise - it is that if the SNR is lower the line is more susceptible to noise.

If the target SNR is 6dB then this will determine the speed that can be achieved. If the actual noise level (not the ratio) the attainable rate will be higher. If the noise level (not the SNR) is low enough the line can run at the maximum speed (8Mb/s for ADSL, 24Mb/s for ADSL2+ if the provider will allow it) and the SNR will be more than 6dB (which is good) but a higher noise level will cause the SNR to cause the line to sync at less than the maximum.

It is possible to run with an SNR of less than 6dB on a stable line, particularly a short line which gives a higher line speed.

All down to absolute noise level which then determines the SNR for a particular speed (faster speed gives lower SNR as the bin loading on the line needs to be higher and more noise tends to be present on the higher frequencies used for the higher speeds.

For the same bandwidth, sending at a higher speed requires that the transmitter sends a more complicated signal, with finer detail than a slower speed signal.

To be able to distinguish the finer details the receiver needs to have a better signal to noise ratio. If it doesn't get it, the fine details will be lost in the noise and the receiver won't be able to reconstruct the faster data stream.

The maximum data throughput for a link is the product of the bandwidth and a function of the signal/noise ratio as described here:
Shannon�Hartley theorem. (We assume that the transmitter is already using up all the available bandwidth on the link by sending at the Nyquist rate described on that page so we can only send faster by using more complex signals.)

For a particular type of broadband link (ADSL, cable) the bandwidth will be fixed leaving the SNR to determine the transmission speed. If the SNR is insufficient for a high speed link the transmitter will fall back to slower, less complex, signals until the receiver can accurately recover the data stream.

The sort of more complex signals that are needed are described here:
Quantized QAM.

In reply to a post by GrahamN2012:

For the same bandwidth, sending at a higher speed requires that the transmitter sends a more complicated signal, with finer detail than a slower speed signal.

To be able to distinguish the finer details the receiver needs to have a better signal to noise ratio. If it doesn't get it, the fine details will be lost in the noise and the receiver won't be able to reconstruct the faster data stream.

That is exactly the kind of explanation that I was looking for - thanks!

In other words the more refined (complex) the data, the 'cleaner' the line needs to be. :

In reply to a post by has:

In other words the more refined (complex) the data, the 'cleaner' the line needs to be.

You've got it.

Great thx - always wanted to understand that and now I think I do

the definition of the SNR margin is the excess SNR over that required for a pre-defined Bit Error Rate (BER). Some modems / routers have a BER test facility, it works at the ATM cell level. From memory the BER used is 1 in 10 million (10^7) to define 0dB SNRM.

I don't believe a higher speed requires a higher SNR or SNRM because higher speeds result from using more tones or bins - the process is the same, it's just happening in more frequency bands at the same time on higher speed connections.