General Discussion
  >> Fibre Broadband

Register (or login) on our website and you will not see this ad.

  Print Thread
Standard User WWWombat
(fountain of knowledge) Fri 28-Mar-14 13:44:07
Print Post

BT SIN 498 - Annex for Testing

[link to this post]
I've just seen that BT's SIN 498 (for GEA-FTTC, including modem requirements) has a relatively new section detailing the tests that a modem has to undergo.

It is interesting to techies, anyway...

There are quite a few snippets of info that can be gleaned from it - and there would be even more if we could get hold of some of the "attached" spreadsheets (eq QLN and HLog plots).

One example is that they run tests for the modem at (simulated) distances of 200m, 500m, 1000m, 1500m, 2000m and 2500m. They repeat these tests for a variety of cabinet locations (ie they vary the exchange-cabinet distance, and use appropriate DSLAM settings), and they perform tests against both fast and banded interleaving settings.

Interestingly, from the banded profiles chosen in those tests, they imply that long D-side lines (1500m+) perform better when on cabinets that are far from the exchange, perform a little less well on cabinets that are near the exchange, and perform less well again on cabinets in the middle. Obviously the PSD maps have an impact!

A second example comes from the testing of vectoring. Their test setup involves placing 16 modems at 200m, 16 modems at 500m and 16 modems at 900m

A third example comes from testing of G.INP as a replacement for Interleaving. That tests up to 2000m, with crosstalk included, and additional REIN noise injected into the line.

Finally, they have tests to verify that the modem correctly reads QLN and Hlog values from line measurements, and returns the data to the DSLAM. These are run for distances of 200m, 500m, 1000m and 2000m.

In this test, it specifically mentions that the reported Hlog and QLN data is "used
for test and diagnostic purposes to reduce time to repair (T2R)".

The SIN mentions some companion spreadsheets. I wonder, if we could get hold of these, whether they would help show what the HLog and QLN graphs ought to look like for varying line lengths.

Edit: Added link. Whoops!

Edited by WWWombat (Fri 28-Mar-14 16:13:45)

  Print Thread

Jump to