Kitz Forum
Broadband Related => Telephony Wiring + Equipment => Topic started by: GigabitEthernet on December 08, 2020, 06:53:39 PM
-
https://www.run-it-direct.co.uk/adsl-vdsl-faceplates/g-fast-openreach-faceplate/
Never seen one of these in the wild, anyone tried one on VDSL?
-
I see it has an additional RF filter.
-
While I have not tried one of these G.fast filters myself, I do have a few observations from the side-by-side photo comparison on that website.
Most notably the common mode choke, referred to as the REIN filter by Openreach, has fewer turns (so a lower per-leg inductance) on the G.fast faceplate than the one in the Mk4 VDSL filter. The number of turns on the VDSL Mk4 filter appears to be the same as on the VDSL Mk3 filter which I measured as having a per-leg inductance of 1.0 mH, while the G.fast filter appears to have a similar number of turns to the VDSL Mk2 filter. Back when the Mk3 came out, which differed from the Mk2 only by the increased number of turns on the REIN filter, there were a few remarks from people syncing near the full 80/20 on VDSL loosing a little sync speed when changing to the Mk3, possibly due to this new REIN filter having a little bit too much impedance at the highest VDSL frequencies (in theory this shouldn't happen for a differential mode signal passing through a common mode choke, but in reality there is some impedance). It would therefore make sense for Openreach to return to the lower inductance design for G.fast which uses higher frequencies in order to reduce this effect, particularly given that G.fast lines will be short so therefore less likely to need the improved REIN rejection of the higher inductance choke.
The other observation is that the PCB layout is different, with the G.fast variant using a (presumably) double sided board rather than wire links, and the use of SMD components for the small passives rather than through-holes. This design could potentially be better optimised for RF transmission through it, again more relevant for the higher G.fast frequencies.
It is hard to make out under the green soldermask, but the PCB material does not appear to be the same texture as the (cheap) paper-laminate type used on the VDSL, it could possibly be FR4. This may be for some manufacturing reason for the double sided board, but could also be for RF reasons as some PCB materials have more tightly defined dielectric constants than others and smaller loss tangent. For example, I often use Rogers RO4003C material for high frequency interconnect board designs at work due to these improved properties (though the G.fast filter almost certainly doesn't use this as it's very expensive).