High downstream SNR
As can be seen from the above data, I have a very high D/S SNR. While this doesn't really worry me (I'm no gamer or suchlike), I was told by the last BT engineer who came out to resolve another problem, that he had switched me to a 'good' pair in the DP and I should get 11-13Mbps with no problem.
I did get around 12 for while, then dropped to 9-9.5 (still 'good enough' for my needs). Ever since the drop, occasionally DLM will reset my speed to around 13.1, but with an SNR of 3, which, considering my distance from the exchange is too low. It will hold this for 24-36 hours, then kick me back to around 9 and stay there for a while, until the DLM resets it - again.
The other thing, which IS causing me some concern, are the very high error figures.
Anyone care to investigate or offer suggestions?
Edited to add: I go away for 3 weeks 3 or 4 times a year, and disconnect the router/turn off all but essential electrics, and when I get back and restart, it is usually at the higher speed/lower SNR.
My line is doing the same 
Don't worry about the high FEC count - that is corrected errors. ADSL is designed such that it has a certain error rate, but that the encoding is such that they can be corrected. This is what FEC are - me I get a FEC second maybe every couple of minutes, but one second typically has a count of 20. Bad seconds can have a count of 2000 for me!
The CRC errors are more important. It means that the error correction did not work, so there is a known error. When this happens in a second, this creates an Error Second, and that is what is said to be used to set your line speed. Now you CRC errors are very low, so that shouldn't be what is causing the problem, but you have a huge number of error seconds. So why your Error Seconds and CRC are so different is odd, you would expect CRC >> ES. Anyway yes that high Error Second could explain the speed drop, question though what is causing it ...
For me what seems to give the dropping SNR (mine is at 15dB right now!) is that I occasionally get a few seconds of extreme noise, that takes the line down. Doesn't look like you have this from your statistics though ...
Thanks for the quick reply, @summers - appreciated 
I'm not concerned about the FEC count, but it could be lower. I am aware of the technology - I spent about 20 years in IT support as one of the facets of my job (no offence), but I don't understand the DLM behaviour.
The ES/CRC disparity is one of the things I would like an explanation for, too.
My SNR is fairly steady, and as I'm running Router-Stats, I can monitor that constantly. And no, I don't get extreme noise - again RS would alert me to that.
@jab1 IIRC (
) that's a TG582n router, which again, IIRC remember correctly, doesn't reset the error second count until rebooted, so assuming it's more than 4 days since the last reboot the error seconds might not be as bad as they look at first glance.
Having said, that, your line is clearly far from perfect and the sync speed consdierably lower than you might hope for - you might not get far trying to raise a fault if it's not actually dropping the connection on a frequent basis, although I'd be surprised if there isn't some audible noise on the line, even if intermittent, which could provide a route to raising a phone fault.
@JayG Correct - it is a TG582n -I'd forgotten about the ES not resetting until a reboot.
. Yup, I'm aware the line is pants, but unfortunately the noisy line had crossed my mind, so it is checked on a regular basis, so that doesn't seem like an option.
Last router reboot was early August on my return from my holiday.
@Gandalf Any chance you could cast your eye over this. Sorry to pick on you, but you have been very helpful in the past.
If you want to do any intrusive testing, I won't be inconvenienced this afternoon - going out for 2 or 3 hours.
TIA
Unfortunately @Gandalf isn't in the office now but I've tested your line.
Looking at the graph attached below it backs up what you're seeing
What's interesting is that your target SNR has been set and is still at 3db which implies something on the line is making it sync at the higher SNR
With regards to stopping your line being moved onto the 3db target there are settings we can change which essentially makes DLM think the line can't do any better.
If you want us to go ahead with that let us know
@ejs I understand what you say, but believe me please when I say 3dB appears (to humour you) too low, but 6dB leaves me with a higher stable speed - been there, got the tee-shirt (& shorts)
Hi @jab1
Our tests are showing a high amount of errors on the line now.
| xDSL Status Check |  |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
I've nudged the target SNR back up to 6dB and asked our suppliers again to pin it at that level.
Discussing this with Matt, what he said about making DLM think the line can't do any better is where we'd place a modify order to set a custom threshold that DLM works on try to stop it making positive changes like moving the target SNR down.
I did that around this time last year. 
Testing the line doesn't show the custom threshold set anymore and going through the modify order process again, the custom threshold I set is still showing, but I've submitted the order once more, which should complete tomorrow.
Let us know how things go over the next 24 hours.
[edit]
Coming back in to edit my response to paste the test results in to my post, I've checked the results in the supplier database of the test I ran, and the custom threshold is actually still set. Hopefully our suppliers have more sticky glue to pin the SNR!
Failing that if it drops down again, we could attempt to switch DLM off, though that may cause further issues if it's at all possible.
@Gandalf Looks like your nudge has taken effect - now showing a steady 6.3dB in RouterStats. Lets hope they get the pinning done quick.
Thanks - as always.