First question, Are the modems with the center freq of 38 mHz, 3.2 wide or 6.4? If 6.4, you must be able to see on spectrum that they are set a few dB lower, due to the width and slightly more overall power into the lasers, which would give them a lower upstream snr.Next.Are you taking a return sweep reference at the node? I have noticed in many systems,(including the one I'm currently working) when looking at a RAW return frequency response at the node I will get a 1.5 to 3 dB tilt, where 5 mHz is 1.5 to 3 dB higher than 40 mHz.Assuming the sweep points are set equally, and if I take a reference at the node, I will carry that tilt to every amp afterward in cascade.Solution? Go to first active, flatten response, then take reference and carry that to all actives. Or, sweep all actives in raw return, with no reference to achieve unity gain back to CMTS across freq's 5 to 40 mHz. This may seem counter intuitive, but think about it, we need to eq the return in the opposite direction for the return, and the goal is to hit the CMTS with the same level across the return spectrum, 5 to 40 mHz.On a side note, we recently added a router at the headend, and after we migrated a few nodes to it, (to decongest some of our more populated nodes) I noticed a difference in the upstream snr, between the 2 routers. Our 6.4's had a better up snr than our 3.2's on the new router vrs the original router. All things being equal the 3.2 upstream carriers should have a better snr. I made a comment in passing to the headend techs and miraculously the numbers shifted to expectations. Hmmmm.
This is CABL.com posting #333275. Tiny Link: cabl.co/mbyRz
