You could be right, but that would make DLM rather dumb. Unfortunately, as time goes by, DLM seems to elevate its dumbness.
One problem I have with "power" is that it is a tricky concept to deal with, and one that many people (including me) cannot handle properly.
For example, we are very used to seeing the idea that profile 17a is listed with a maximum power of 14.5dBm, and seeing modems report power - anywhere from -10dBm to 13+dBm.
But those numbers are aggregate power, accumulated from the individual power on each tone. And there is another limitation that affects the individual power on tones: the PSD (power spectral density), measured in dBm per Hertz.
So, when we see a modem report a lower-than-expected (aggregate) power value, it might be because
- there are fewer tones than expected
- the PSD mask has limited power of many tones
- "something" has reduced the power of individual tones (eg UPBO
There's also the gotcha that we're counting in log scales. Doubling power is +3dBm. Or, when calculating aggregate power, doubling the number of tones at the same power is also +3dBm.
LR-VDSL has a change that allows aggregate power to be +20.5dBm instead of 14.5dBm. A 6dBm difference means quadruple power ... but aggregated over (roughly) only half as many tones ... so could mean octuple power per tone. Wow.
Except it has to fit inside the broad parameters of the existing PSD mask (power per hertz), but benefitting from removal of the DPBO shaping that would otherwise allow coexistence with ADSL.
Who can tell, off the top of their heads, whether +6dBm in aggregate amounts to more or less than the change in shape of the PSD mask?
(*) - The UPBO article is interesting. It shows that the settings can be tuned to a "preferred distance", above which the upstream speeds fall away considerably.