HH5 rebooting after f/w upgrade, noise margin/power too high

I've had problems with the HH5 after the most recent firmware upgrade - connection dropping randomly a few times a day.

BT sent out a new HH5, and I've had the same issues.

I was consistently syncing about 95000 down, 35000 up, with a noise margins of about 11 down and 15 up (on the 80/20 service).

I spoke to them again this morning, and was told the issue may be the power levels are too high and so he made some changes "at the exchange", which I assume are really at the fibre-cab. This dropped my noise margins down to around the 7db mark, and my sync speeds have stayed about the same for downstream but dropped to 25000 for upstream.

So far, the connection has been more stable - but early days. Anyone else heard of this being the cause of the HH5 connection drops?

It was an issue in the early days of ADSL - power level too high causing the modem to saturate/overload which caused distortion and reflections and thus produce large numbers of errors. Very likely to be the same sort of problem.

I would guess you are less than 100m from the exchange, maybe as little as 80m

In reply to a post by MHC:

I would guess you are less than 100m from the exchange, maybe as little as 80m

This is fibre, but yes - about 100m from the cab.

Yes, I meant cabinet - just came out after writing the ADSL comment!

If you're on FTTP, the cabinet is not involved.

How can you be synced at 95000/35000 ?

I read that to be the Max Achievable.

Yes, max. achievable on FTTC.

So not "consistently syncing about 95000 down, 35000 up" after all.

I think it's unlikely that changes were made by an ISP to Openreach equipment.

In reply to a post by BatBoy:

So not "consistently syncing about 95000 down, 35000 up" after all.

Why not? Here's the stats now, after the adjustment:

7. Maximum data rate: 26255 / 97695

and before:

7. Maximum data rate: 33580 / 108040

Because it's a theoretical estimate which changes all the time, not a sync speed which is fixed till the next resync.