Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[WWWombat]

One aspect that intrigued me about this is an anomoly. Or it appears to be an anomoly initially, but turns out to not be an anomoly ... and turns into a question about how this improves anything!

The obvious first "big change" is that the CAL is set to zero for the whole cabinet, so the PSD mask will not reduce power on any tone (ie no attempt to be backward compatibility with exchange-based ADSL). This will apply to all lines, not just the long-range ones.

The obvious second "big change" is that the maximum power is increased from14.5dBm to 20.5dBm. This applies only to the long-range lines.

The apparent anomoly is that long-range lines can run at higher power ... so would appear to be able to cause extra crosstalk on the non-long-range lines.

But those long-range lines are actually restricted to the same PSD mask as the short-range ones (and is broadly the same as any "plain" cab outside the exchange building) - which ends up restricting power on a tone-by-tone basis. So the long-range lines can't run higher power on individual tones ... right?

How does an increase in the total aggregate power help, if the power for individual tones is unchanged?

[WWWombat]

Seeing as ECI doesn't support upstream G.INP I don't suppose this will be any use in ECI.

The requirement for downstream vectoring might get in the way too.

They should get some sort of Behemoth cabinet that does ADSL/VDSL and G.Fast all in a single PCP or if you run G.Fast at 0-106 does it have the range of ADSL2+?

No. G.Fast is strictly short range.

[gt94sss2]

But those long-range lines are actually restricted to the same PSD mask as the short-range ones (and is broadly the same as any "plain" cab outside the exchange building) - which ends up restricting power on a tone-by-tone basis. So the long-range lines can't run higher power on individual tones ... right?

How does an increase in the total aggregate power help, if the power for individual tones is unchanged?

It's late so I might have missed something but the STIN talks about

Removing the CAL shaping for LR-VDSL2 lines so that no PSD shaping is applied to the downstream VDSL2 signals, other than the G.993.2 limits.

As well as setting the CAL to zero and vectoring to deal with crosstalk..

[Chrysalis]

Seeing as ECI doesn't support upstream G.INP I don't suppose this will be any use in ECI.

They should get some sort of Behemoth cabinet that does ADSL/VDSL and G.Fast all in a single PCP or if you run G.Fast at 0-106 does it have the range of ADSL2+?

adsl2+ has worse range than adsl1/2 because it allocates less power per tone (it doesnt need the full power per tone as generally it is only supposed to be used on good short lines). On ukonline my 50db line synced at about 6-6.4mbit on adsl2 during the daytime but using adsl2+ it was only circa 4.5mbit.
adsl2 is the best adsl tech for long poor lines.
vdsl should have comparable range to adsl2 as long as it can use the lowest tones at max power, if it has to do power cutback then it will be very poor on long lines.
g.fast at its original spec would perform worse because it had a max bitloading of 12 per tone, plus I think it has a similar issue than adsl2+ had where it allocates less power per tone, not really a tech designed for long lines.  However BT got the spec modified to allow 15 bits per tone although I dont know if the power per tone is back up to par with vdsl.

[Black Sheep]

Yes. The document specifies the insertion loss of the D-side.

Strictly, I think BT counts the D-side as the cabinet to DP, and measures the loss of this section in their cable records. I think the dropwire is considered separately, but perhaps BS can confirm what is likely to apply...

Correct on all counts, W3 ............. 

[WWWombat]

From what I've seen, the amendment to G.Fast allows for 14 bits rather than 15, and the amendment to power is by increasing the allowed aggregate total from 4dBm to 8dBm.

However, the PSD doesn't change. The total power increases, but the maximum power of any one tone doesn't. Presumably the gain comes from places when the modem chooses, for any one tone, to not make use of the maximum allowed to it, and trying to do so would take the aggregate out of bounds.

[WWWombat]

As well as setting the CAL to zero and vectoring to deal with crosstalk..

Yes, setting CAL to zero is what removes any artificial PSD shaping for ADSL compatibility.

Note, though, that any DSLAM located within 100m of the exchange will already have a CAL of zero.

[Weaver]

Have there been any more developments in this area?