I have to agree 0.5db is not a major difference, my line and most lines vary +-1db throughout the day. Irrespective, if the device is faulty - get it replaced which you have done.

Regarding hard wiring, it would be recommended to ensure a disconnect switch in sight of the device.

Removing the socket and soldering directly onto the traces on the board? Just suppose you've got to be careful not to(one day) yank the wire and rip the traces off.

In reply to a post by ian72:

0.5dB is not significant in my opinion. Whilst it may be a good percentage of the SNRM at your "full speed" it is little compared to the full noise margin available - if you forced the line to 512Kb then the margin would be a lot higher and 0.5dB would be a small percentage of the total.

A dodgy power connection means it is right to replace the router but the extra 0.5dB in your original post is not going to make a major difference to speed, even 1.5dB as going to be a relatively small improvement.

I strongly disagree.

I'm currently synced at 69129 with my attainable actually reporting 66996 due to currently having 5.7dB SNR.

That suggests that a loss of 0.3dB would reduce my connection by 2Mbit, which implies 1.5dB could be up to 10Mbit, a 1/8 reduction in sync speed if you are right on the edge of the limit. Its pretty much what HAS happened to my connection over the last few years, unfortunately due to crosstalk so not fixable.

Granted its probably not so linear in the real world, but I'd still argue its likely to be far from a "small improvement".

As it stands I'm keeping synced, but a sudden loss of 0.5dB and certainly 1.5dB could cause a resync. That alone is hardly an insignificant impact on your connection.

Edited by alexatkin (Mon 26-Nov-18 15:41:57)

Hi

No problem soldering direct if you know what you are doing.

It could be a slightly high resistance connection due to dirt or oxidation on the socket which may clean off, however the power isn't used directly by the modem as it goes through several DC-DC converters meaning a slightly poor connection is masked over by all the other noise and switching that happens before it reaches anything. Also because the power draw is quite high, a high resistance connection would either sort itself out or cause heating of the socket and perhaps complete failure.

It may just be noise being picked up by the wire from the power supply, and as you twist it, you might be increasing or decreasing the ability of common mode noise cancelling to work or not, i.e. twisted pairs picks up noise more evenly, therefore it's easier to filter that noise away. This could mean soldering the wires doesn't see an improvement.

Try twisted the power wire so it spirals, see if that helps first.

Regards

Phil