Now considering your observations: 25.0 dB DS & 14.8 dB US; old Hlog plot and 23.5 dB DS & 20.2 dB US; new Hlog plot. Unfortunately I can not make any sensible deduction from those observations. 
In my previous response, I mentioned that the new Hlog plot looked the "better" of the two; I preferred the new Hlog plot. That preference was due to the lesser attenuation being indicated in the lower frequency region of the graph -- the lower frequency tones being, of course, those of the US band. Hence my quandary. My visual examination of the graphs makes me prefer the new Hlog plot because it appears to show less attenuation in the US band. Yet that is completely the opposite the opposite to what you have noted . . .
I have an idea why my upstream LATN was higher despire that my US Hlog was lower. I think US Hlog is measured by CPE and CO and upstream SATN is calculated by US Hlog measured by CO. I see on all Hlogs that on upstream there is noise floor at used upstream frequiences. I've had about 80 kbps lower upstream and 400 kbps highered downstream on that day. I think it is very probable on that day I've had signal reflections moved to slightly lower frequiences of ADSL2+ band.
It is curious. One day pavement resurfacing works caused reconnect slight change at Hlog. Next day there was also pavement resurfacing works also caused reconnect and Hlog became again ususal as before pavement resurfacing work. Is still abnornality cause unknown?
Best regards
konrado5
