Print

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

[lf2k]

Evenin' all,

After my fiddling on Saturday, I've been sitting here waiting for the DLM to relent and take interleaving off... As I write this I've had just shy of 5.5 days uptime with no resyncs.

Having looked at my stats on MDWS (which is now becoming an obsession!) I notice that unlike everyone else, I've got a shed-load of CRC errors on my upstream (which is *NOT* interleaved - it's just my downstream that's interleaved).

Since I can't find anyone else with similar issues on MDWS, I'm wondering if anyone has any insightful information as to how to minimise these intermittent problems?

I've just been to Tescos and picked up an AM radio - a quick scan around the house shows that the rats nest of cables behind the TV is generating interference; my router (Netgear DGDN3700v1) being the noisiest thing I've got switched on (this is using the WAN port connected to the HG612).  The new DECT phone is switched off at the mains, my Synology Diskstation has been been removed etc.  I've even turned the cooker off, since that was very noisy...

Obviously I can't keep these things switched off for ever, so I was wondering if any bright spark can give me some feedback on the following...

• I've walked my line (it's all underground) and since I live in a cul-de-sac it's fairly easy to walk the entire thing with certainty of where my wire is going to the cabinet (I make this about 490 meters) - am I expecting too much?
• I've got some new surge protectors, is it worth swapping them out for the one behind the TV?
• Most of my CRCs are on the upstream - we're talking one error every ten minutes or so - does this indicate local interference?
• Why am I seeing CRC errors on the upstream rather than the downstream (or is this the interleaving helping the DS)?
• This connection started life with good upstream (18Mbps), but now I get I "only" get 10Mbps - there's a big chunk of missing "U2" missing, is this due to line length/crosstalk?
• If I switch stuff out (like the surge protector), is it better to do this in the morning (I've read this is the time to resync)?  Do I have to allow 30 minutes downtime?
• Could this potentially be a line fault, and if so would plus net take it seriously given I get reasonable (by most peoples standards) speeds?

I'm kinda scratching my head here - what do I need to do to get the interleaving to stop?  I was hoping for G.INP+Vectoring, but I guess I'll have to carry on hoping - anything I can do in the meantime?

Cheers,
Rob

[Chrysalis]

how do you know which side of the cul de sac the line goes? I dont understand when people say they walk the line, unless its above ground how do you know where it is?

Paste your D1 attenuation and I will guestimate your line length.

Whilst it is odd to see more CRC on the upstream, one every 10 minutes is nothing anywhere near a problem.

Your U2 bitloading looks too low to me tho suggesting there is an issue, normally U2 snr is stronger than the D2 and D3 because of less crosstalk.

[lf2k]

See attached... hopefully that's what you need (according to the checker here that's about 1.3KM!)

I appreciate that walking the "PO", "GPO" and "BT" manhole covers isn't exactly a scientific way of measuring your line length, but when you live in a village it's a bit more obvious (they're not going to run my line across the fields either side - it's going to follow the only road, to the closest cabinet).

Despite the potential line length disparity, that still doesn't really answer the nub of my query - why can I get semi-decent downloads (which are more inline with my guestimate of ~500 meters), but I get more CRCs on my upstream rather than downstream?  I can't find anyone else on MDWS with a similar predicament.

After all, if my line was really 1.3KM long I wouldn't be getting ~50Mbps down.

The reasoning behind my questions is that I see Bald Eagle has a long line but no (that I saw) CRCs...

[Chrysalis]

my guess is approx 550-650m.

[lf2k]

Whilst it is odd to see more CRC on the upstream, one every 10 minutes is nothing anywhere near a problem.

Your U2 bitloading looks too low to me tho suggesting there is an issue, normally U2 snr is stronger than the D2 and D3 because of less crosstalk.

Thank you - that's basically what I'm driving at - at least I don't feel like I'm going mad now!

My line in to the house passes through the concrete floor, and as such I have no slack on it.  The previous owner cut through the cable about 20cm short of the master socket.  I've spliced in some CW1308 to make up the lack of cable reach.  Is this likely to be causing "reflections" or something similar on the line?  And moreover, why is it only affecting my upstream?

(I appreciate you might not have the answers, but I'm wondering if it's worth digging up my drive or not!  )

[lf2k]

my guess is approx 550-650m.

So I wasn't a million miles out... that's good to know, thx.

[boost]

Is this likely to be causing "reflections" or something similar on the line?

In my black and white world, I'd say yes.

Whether it's affecting your upstream, I've no idea

Can you paste your HLOG?

[lf2k]

Hi Boost - thanks for the feedback!  I've attached my most recent HLOG from this morning...

Just to give a little background, I replaced the surge protectors etc to see if that would improve things and there was a marginal improvement when I resynced at 10am yesterday.

I also swapped out my old Netgear for a TPlink I picked up, and it was when I was messing around with this I noticed that the CRCs being uploaded to MDWS and showing in the HG612 Modem Log are out of kilter with what the HG612 is showing in it's web interface (the web ui is showing the value for "RSUncorr errors" (stats prog) in the CRC box, and as such there are no upstream CRC errors according to the web ui).

Back to the "reflection" theory - is just splicing in 20cm of cable is too short for some of the wavelengths involved?  I just played with this calculator here and have confused myself slightly...

Could I potentially improve things with a longer piece of cable spliced in?

[boost]

JGO is your man for that!

I just look for anomalies in the HLOG so I can blame the copper

To that end, need much greater resolution on all your bands. However, interesting about the disparity between the UI and MDWS? I wonder what it's referring to?

[Bald_Eagle1]

I also swapped out my old Netgear for a TPlink I picked up, and it was when I was messing around with this I noticed that the CRCs being uploaded to MDWS and showing in the HG612 Modem Log are out of kilter with what the HG612 is showing in it's web interface (the web ui is showing the value for "RSUncorr errors" (stats prog) in the CRC box, and as such there are no upstream CRC errors according to the web ui).

Ignore the error counts in the HG612's own web UI as they are reported incorrectly.

They are reported correctly in the raw data obtained via a telnet session though and this raw data is used by the HG612 Modem Stats programs.

[les-70]

 Upstream errors can suggest a noise source near the CAB/DSLAM but your upstream error rate is low and does look a concern.  The upstream rate is a long is a long way from values that might might cause DLM intervention i.e 120/min for Plusnet and the downstream errors are worth more attention.  Also since your downstream is interleaved -which will dramatically reduce error rates- the upstream rate may just be a counterpart to the downstream.  You may just have a noisy line and it may be best to just see if the current downstream error rate is influenced by anything.

[lf2k]

They are reported correctly in the raw data obtained via a telnet session though and this raw data is used by the HG612 Modem Stats programs.

They are indeed correct via Telnet, I should've checked that first!

Upstream errors can suggest a noise source near the CAB/DSLAM but your upstream error rate is low and does look a concern.  The upstream rate is a long is a long way from values that might might cause DLM intervention i.e 120/min for Plusnet and the downstream errors are worth more attention.  Also since your downstream is interleaved -which will dramatically reduce error rates- the upstream rate may just be a counterpart to the downstream.  You may just have a noisy line and it may be best to just see if the current downstream error rate is influenced by anything.

Interesting - thanks for the feedback Les!  My cab is sat next door to a garage forecourt, but the noise is there even when the garage is closed (perhaps it's got nothing to do with the garage at all).

Looking at my ES+CRC graphs there an obvious correlation between downstream/upstream errors, but the interleaving (as you point out) is masking the downstream to a degree.

I think I've come to the conclusion that the source of noise isn't in the house (I've literally turned *everything* off in the house for a couple of hours, bar the modem, and the error rate is still there).

I've got a guy coming over towards the end of the week to help dig up the drive to see if I can remove the short splice from my line... at the very least I want to know the route the cable goes after it disappears in to my concrete floor.