After four weeks of begging, I managed to get Janet to volunteer to tidy up the layout of my three modems, their dc cables and most importantly the cables from modems to wallsockets and get everything well away from any mains coming into the modems’ dc power bricks. (I didn’t want to stand upright for that long because it can make me feel dizzy and horrible.)
Four weeks ago two more ZyXEL VMG 1312-B10A modems were dropped in in place of the former DLink DSL-320B-Z1 units. One ZyXEL modem had been installed somewhat earlier still. But four weeks ago all three modems were just parked any old where in evidently a horrible physical arrangement because there was a huge performance drop evident on modem 1 for example the moment it was fired up after they were all moved.
Now Janet has moved mains away she tells me and downstream sync rates have improved this afternoon, just from this simple change, as follows
modem cwcc@a.1 downstream sync 2861 -> 2975 kbps - +3.98%
modem cwcc@a.3 downstream sync 2832 -> 2935 kbps - +3.64%
modem cwcc@a.4 downstream sync 2821 -> 2894 kbps - +2.59%
However modem 1 is still not back to the sync rate it had 4 weeks ago, which was 3055 kbps, so still another 2.7% short. Back then I think Janet had the first modem well away from all other modems as well, and when she moved it, 6.35% of d/s sync rate was lost and we still have not got all of it back yet. (That last figure is a percentage of the higher, earlier number, in case you are noticing that the differences do not add up.)
So more needs to be done regarding placement, but I am not quite sure what. Could separate the modems physically, but I would not have thought they could be emitting any rf themselves. Won't they be shielded with some kind of conductive coating or some such inside the plastic casing? I could ask Janet to check where the dsl cables are right now just in case they are near to anything nasty.
The very first time ZyXEL #1 was plugged in, its downstream sync rate was a glorious 3139 kbps, quite a way above the 2848 kbps delivered by its predecessor a DLink DSL-320B-Z1, a 10.2% improvement in downstream sync rate. But 80k disappeared fairly soon, and then, over time, all the rest of that 10% eroded away too.
One thing that has not been done yet is putting ferrites on the ZyXELs’ DC cables. This was something that the earlier DLink modems had and so if ferrites give any benefit at all then there is another possibility for improvement. However, the first time one ZyXEL was plugged in it did not have the benefit (if any) of ferrites attached.
Another thing that I may have done in the past is folding up the DC cables to shorten them. The superstitious and optimistic idea here is to try and stop them picking up any interference, so what I have sometimes done in the past is to bend the cable into a u- or hairpin-shape, then I out a twist or plait into it and finally wrap the whole thing up into a compact volume. The wacky idea anyway is to prevent it from being a decent receiving antenna. Now I cannot even remember if that was done with the DLink modems. I suppose there might be a chance that the evidence is still there, depending on what Janet did with them when she put those modems back in their boxes in the spares dept.
Anyway, all a bit disappointing, frustrating and mysterious.
The ZyXELs are still a hit though, slightly more performance, no downside apart from the reduced upstream performance, which is really annoying and the much greater purchase cost, which is 4 or 5 times more. They are far more stable as their downstream SNR margins do not drop below the initial 3dB downstream target set and they never fail and resynch, they just carry on calmly forever. I don't know if this is to do with the ability to support monitored tones. It could be that the DLinks are set to be too aggressive and just actually cheat by going below the selected downstream target SNRM by design, because they show a lower SNRM than the target pretty much immediately and it makes you wonder if aggression and a bit of cheating is the secret of some of their speed. In that case, to compare like with like, I should level the real current downstream SNRM values on one or other or both modems by tweaking somehow, and the ZyXEL would then gain some extra speed relative to its rival as a result, because it currently always runs at or slightly above my chosen 3dB downstream target SNRM with no ‘cheating’.
I could do with some suggestions about physical shielding or location or anything like that that I can do.