Thank you for reply. I'm not sure whether I understand rightly. Assume on DSLAM target SNR margin is changed from 6 dB to 3 dB. Are tones with only 3 dB SNR margin then enabled or rather only the same tones are used as at 6 dB target SNR margin and also more bits on the same tones?
Tones that were previously empty due to inadequate SNR will have 1 or 2 bits in them, depending on their SNR, existing in use tones will have an additional 2 bits per symbol.
I've heard SNR margin is depended not only on SNR on particular tones and synchronization rate. I've heard it is also depended on BER. I've notices strange phenomenon on my line. At the lower sychronization rate, higher power output, the same used tones I have sometimes continously lower SNR margin than at the higher synchronization rate, lower power output and the same used tones. For example, I have now 1207 kbps upstream and power output 12.3 dBm but my SNR margis is never higher than 6.8 dB however when I had 1225 kbps and power output 12.1 dBm I had SNR margin for upload about 7.1 dB.
Depends how it's calculated. Normally BER is not used to calculate SNR as, ideally, BER is zero. The SNR that is provided is after any error correction. When BER is zero error vector magnitude is the usual suspect to provide an SNR value. Where BER isn't zero a calculation can be done to estimate SNR based on BER and modulation in use.
The SNR margin depends on the sync rate, the higher the rate the less margin, the actual SNR itself does not and is the same whether connected at 24Mb or 64kb.
Recently I have strange interferences during first 20 minutes of new synchronization. On upstream SNR margin is only 4.6 dB but after 20 miutes SNR margin is already about 6.6 dB-7.0 dB. Is it not strange much interferences are only at first moments of new synchronization? Only at first moments of synchronization there are bitswaps for upload.
Not really. The same noise that caused the resync in the first place is probably still there for a while afterwards.
I have a similar effect but worse on my own connection, where it repeatedly disconnects over and over again until it reaches an upstream rate it can sustain even with the noise present.