After only a couple of days I have mixed feelings about the new modem.
In the day it synchs faster - about 7.6 Mbps vs 7.0 Mbps.
Daytime speedtests @ speedtest.net give about 6.4 Mbps vs 5.7 Mbps
Even so, CRC rate is about 7 per minute vs between 0.1 and 0.7 per minute.
But in the evening it becomes awful, and web browsing is noticeably sluggish vs ST546.
The old ST546's SNR Margin drifts down in steps from 9 to about 7. i.e. -2dB
Currently this happens between 21h00 and 22h30.
The new 2-Wire drifts down continuously (no steps) from about 8.5 to about 4.5. i.e. -4dB
The CRC error rate rockets up. It touched 100 errors per minute just before I forced a resynch
as the BBC TV programme I was trying to watch was having constant dropouts.
Yet the FEC rate was quite low - about 500 per minute.
The ST546 FEC rate was often between 2000 and 7000 per minute.
This suggests to me that the ST546 is doing a better job at correcting errors - opinions?
As neither router gives the per minute figure directly, I just divide the actual figure, e.g. FECs, by the uptime in minutes.
This always starts at 0 and increases over time so I think that I'm correctly comparing eggs with eggs between the two routers.
Resynching the 2-Wire around midnight the SNR Margin was again 9. Speed dropped back to 6.7 Mbps.
By morning SNR Margin drifted up to 11, with very few CRC errors.
I couldn't get RouterStats to work. RouterStats Lite doesn't show bitloading (or I couldn't find the option).
I need to discuss this over on the RouterStats forums.
Pretty easy to adapt my own monitoring Perl scripts with a new 2-Wire frontend.
It pulls the necessary data off two webpages, converts to .txt, parses and outputs to a cacti-like format.
Then the data looks similar to that from the ST546.
I did have a problem to find an explicit uptime - used "Time Since Last Event - DSL Training Errors" from Statistics.
Seems that the 2-Wire doesn't like being polled every 20 seconds à la ST546, but every 30 secs is OK.
Still need to to get to the master socket. Need to run extended tests to find out if the evening
noise is generated chez nous, or external.
Cheers, Peter