Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[aparat]

Hi all.
I was just wondering how do these stats look like to you. anything to worry about ? maybe to many errors or are they all within limits? Thanks

[burakkucat]

Attached is a montage of the four snapshot plots, created from the data file that is attached to your initial post.

It seems there is a higher than normal level of background noise, it averages out to be -112 dBm/Hz per sub-carrier, and there also appears to be some evidence of cross-talk. There is nothing that can be done as they are artefacts of the physical infrastructure.

[re0]

I'm guessing you're around 200m (line length) from the cabinet with those stats.

Compared to my connection of a similar length line, the error rates for most counters are at least double for yours. Though I think that there is not too much concern about them since most of them are corrected (FECs). G.INP counters look OK from my perspective.

Perhaps you could describe how you filter your broadband connection? What master socket do you have, etc.?

[aparat]

I have openreach mk2 master socket and thats the only socket in my flat, and only one device connected, which is my modem. Then there is Asus RT-AC68U connected to the modem

[re0]

In all fairness, I don't think the error rate is going to be a detriment to your line from a performance perspective (at least noticeably, since it is very insignificant) and DLM perspective. You're at max sync with a bit of overheard so you should be fine.

If you want to do some testing and attempt to see if it possible to reduce the error rate, you could try and use a dangly microfilter (something good like the XF-1e) in the master socket and even try a MK3 faceplate (since that is compatible with NTE5A) and compare the results over a period of time. You could even try a different or shorter RJ11 cable; it's ideal to reduce the distance from the socket to the modem as it will reduce the loop loss, though if it is already 2m or less then I would just leave it.

You could go a lot further, but we leave that for the obsessed.

[aparat]

I will do some testing in few weeks I hope but I just wonder, because I remember few years ago, G.INP was also applied to the upstream. Is it not available anymore ?

[banger]

Upstream G.INP is only applied temporarily when DLM determines it is needed. I have only once seen upstream G.INP on my line and that was only for a few days.

[Weaver]

The xf-1e is very good, but I wouldn’t say it is better than the faceplate types. I doubt you would see any difference, and this is me who has been obsessively playing around with such comparisons. And dangly filters are vulnerable to damage of course. Use the best rj-11-to-rj4 cable you can get and as short as possible. I have used the Tandy cables which are RJ-11 to RJ-11. That url is to the shortest one. The wall socket end should be RJ45 so an RJ-11 plug is not a perfect fit but it it is certainly good enough to work. I have however found a good shop that will make up very short high quality RJ-11 to RJ-45 cables, so a perfect fit at the wall socket end and a satisfying snap.

I just have to remember where I bought those from now. I could probably spare one for you if an extremely short cable is ok (very short is what I explicitly wanted).

I certainly have used this run-it-direct one, or possibly shorter ones, before.

Only use the twisted pair ones such as the above types.

[dee.jay]

You can get shorter than 0.5m, I bought two 0.2m ones from eBay.

[re0]

The shorter the better, if it's a good twisted pair cable. Though I truly believe that you have very little to gain in reference to reducing the error rate, and zilch in reference to speed.

[dee.jay]

I added about 1000kbps from connecting shorter cables, I think!

[re0]

I added about 1000kbps from connecting shorter cables, I think!

In all fairness, I imagine that 1000 Kbps was probably no more than a ~2% increase on your line. I hate to be a downer, but taking into account other variables it may have even been within the margin. I'll just shut it and take your word for it.

In OP's case, there is no scope for increasing the sync anyway.

[dee.jay]

To be honest, you are right. When your sync is already 65,000 plus, an extra 1,000kbps, woopee.

[aparat]

The shorter the better, if it's a good twisted pair cable. Though I truly believe that you have very little to gain in reference to reducing the error rate, and zilch in reference to speed.

I think you might be right.
I just realised I replaced my cable when I ordered VDSL, and upon checking its 1m cat5e 4 pair 24AWG cable, so I think there is not much I can do to reduce errors to be honest, just like most of you said. I wish DLM could apply G.INP on upstream permanently 

[Weaver]

Indeed you can get very very short cables, which is a great idea provided everything else is up to the same standard. I have some 0.3m cables now. My beloved is in charge of modem fiddling, since I am usually too dizzy to make it from the bedroom to the office across the landing, where all the kit is. She complains if modem cables are too short, so iirc one is now 1m long. I now have RJ11-to-BT cables instead of using RJ45 plugs. I have standard telephony fronts on all the wall sockets (NTE5s). These have a standard BT telephone socket (British Standard BS 6312) in them, as in the old days before DSL, and there is no filter in them, no SSFP. This is because I have no POTS, so I don’t need any filters. I have not measured the benefit of this filter removal properly.