>> as @j0hn has pointed out it’s over longer lines it fails
This may tie in with the post I started to make the other day that I started, but didn't know how to conclude. As I still have it in notepad I'll post the unfinished item warts and all.
Its difficult to say with absolute certainty, but the new vdsl line looks to have a different PSD mask.
The clue is how bit load slightly increases after tone 500 (end of adsl frequency range).
In theory. as they are on the same DSLAM and have the same attenuation both lines should have the same spectral mask. Each DSLAM is given a specific set of masks to use which are based upon the cab's distance to the exchange. The DSM (Dynamic Spectral Management system) then allocates one of these masks based upon the the line's distance between home and the cab & performance.
DSM analyses the line during the sync up phase, one of the key parameters is attenuation but it can take other factors into account such as crosstalk or a poorly performing line in an attempt to boost SNR at various tones. Whilst there are many different masks available, they narrow down considerably based upon the DSLAM location.
After a considerable period of looking at many stats, the naked eye can tend to spot patterns from the plotted bit load graph to be able to tell the difference between long & short D sides and also if the cab is close or far from the exchange.
As I say, naked eye is no exact science - you'd need something like a JDSU to analyise each individual tone - its just something that you can spot and get a feel for when you see bit load patterns at certain tones. It's not something I could put in writing, but the main tone shapes I look out for are: any U0 bitload, patterns in the D1 bandset based at tone 250, 500 and the general shape of the U1 bandset due to PCB.
I started to say something about the even stepping and 6 bits.... but I've forgotten now where I was going with that without reopening the graphs and looking again.
The reason it ties in with j0hn's comment is that it's almost like it's selected a psd mask that reduces your available bit load in the U1 region. Bearing in mind that because its a long line DSM will allocate you a different mask than shorter lines. The long and short of what I'm trying to say is that its possibly the PSD Mask and PCB that is causing the reduced upstream. The more I looked at your 2 lines, the more convinced I became that you had been allocated a different PSD Mask.