More Error Seconds than connect time???

I've been going stir crazy over the last few days as my ADSL connection throughput has dwindled from 3.8Mbps to er...
Down 0.48 / Up 0.64 Mbps with ping latencies from 49ms up to 350ms on speedtest.btwholesale.com
The BT phone line 1470... quiet test, is... quiet.
The BT Home hub 3 is directly plugged in the master socket.
I just swapped the filter out and am in the test socket now. After doing this I got a 3.18 Mbps download speed for one test and thought I'd found the problem. Now the throughput is miserable ISDN-liek speeds. I'm on a 4Mbps connection (profile?).

I swapped from the HHub3 to an old Netgear DG834Gv2 to see if the HHub3 was the culprit. Doesn't look that way. Throughput almost identical through the NetGear.

After a router reboot the speed 'seems' to pick up for a few test runs and then slowly slithers down to US Robotics 58K kind of throughput.

The router stats are at the bottom of this post. I have some non-telecomms engineer questions:
1) If I'm basically on a two wire system which handles both up and down traffic, how come I can get such massive attentuation differences between down and up? ...or are they measured differently?
2) A spin off of (1) - my upload is rock steady at ~600K and hardly twitches while the downloads lurches drunkenly from one throughput to another. I'm running these tests now at midnight so there's not much contention. Why does the throughput swing arround quite so badly?
3) With 5400 secs of uptime, how come I have 9K of error seconds? [This really freaks me!]
4) No FEC events? Is that cos HHub3 isn't using FEC or what?!?
5)For a zero loss of framing, I have a colossal load of CRC events which almost all are HEC events too. (I guess an HEC triggers a CRC statistic or vice versa or no?). I'd expect to have some losses in the next network 'layer' with such a large number of errors. (I'm more used to looking at FCAL stats.)

I haven't a clue in the world what's dropped out here. I got to work from home as my wife has MS but this bandwidth isn't going to cut it for my work! (

I spoke to a bright young BT lady in India who could see all this and we put it down to excess water round underground cabling - especially as the water board have been flooding the road outside while installing meters everywhere. However the water table drops in a few days round here and this is stubbornly persisting after nearly a week.

I renewed my BT packages and they're going to move me to Fibre but that's 2 weeks away and may not even cure the problem.

Any input/help/wisdom eagerly received!

Thx

Connection Information
Line state: Connected
Connection time: 0 days, 01:27:26
Downstream: 3.43 Mbps
Upstream: 787.7 Kbps

ADSL Settings
VPI/VCI: 0/38
Type: PPPoA
Modulation: G.992.3 Annex A
Latency type: Fast
Noise margin (Down/Up): 6.4 dB / 6.2 dB
Line attenuation (Down/Up): 51.8 dB / 32.9 dB
Output power (Down/Up): 18.0 dBm / 12.6 dBm
FEC Events (Down/Up): 0 / 0
CRC Events (Down/Up): 275974 / 36
Loss of Framing (Local/Remote): 0 / 0
Loss of Signal (Local/Remote): 0 / 0
Loss of Power (Local/Remote): 0 / 0
HEC Events (Down/Up): 271809 / 14
Error Seconds (Local/Remote): 9020 / 4

Ring your service provider, jump through their hoops and have them get someone out. Those stats is nasty.

Just for your information, the error seconds do not get reset to zero on a re-synch; a re-boot would reset everything to zero.

Ian

Answering your question

1)......or are they measured differently?

Yes they are..... expect downstream to be twice upstream attenuation

2).....Why does the throughput swing around quite so badly?

Are you using wireless....make sure you use an ethernet cable DIRECTLY to the router (no homeplugs)

3) With 5400 secs of uptime

Answered in my other reply

4) No FEC events

Your line is on fast path......no interleaving so no FEC's

So to summarize try a speedtest on this site (and post a link to the results) when you are connected to the router via an ethernet cable.

Upstream attenuation is always something like half the downstream. Yours looks OK. It's when they aren't we wonder why. Explanation, no idea sorry.

Re water, it should've have any effect. If it does then the place the water is getting to the copper needs fixing. The tiniest bit of damp can cause trouble, so if it is water it can take a while to right itself even after the cabling is no longer in water.

The sync of 3.43Mbps looks low to me. I'd expect 4Mbps or so. Which us what you say you have - but haven't at the moment. Upstream tends not to fluctuate as much as downstream.

Are you connecting wired or wireless to the router? Maybe a new wireless router nearby is swamping your signal, but that would only affect throughput, not sync or error rates.

It's possible that you have or a neighbour has recently acquired some new electrical/electronic kit that is creating noise. Often from the combo power plug/power converter units. E.g. landline phones, Sky or other TV-related boxes, TVs themselves. Even moving stuff like that the around and putting the router on top of something electronic. Or tidying up cables and running power and signal right next to each other in a neat bundle. Electrical exercise kit like static bikes and running machines can cause havoc.

G.992.3 is ADSL2, not ADSL2+ (G.992.5). That in itself is odd unless you have set your router to it, which can increase sync at that sort of attenuation but routers aren't normally smart enough to do that until things are really bad.

Did I really take that long to type my reply!!!! Maybe you posted 3 seconds after I started . One finger tapping away at an iPad, in bed, is my excuse.

Did I really take that long to type my reply!!!!

Over 30 mins....that is s.......l......o.......w

Regarding the two different attenuations, I think although it is the same wiring, it is the difference in the Frequency bands used within the ADSL allocation.

The Upstream signals are at lower frequencies with both lower inductive and capacitative losses, hence lower attenuations.

If it was 0 Hz (DC), these losses would disappear, leaving only resistive.

As the frequency increases, although the resistive losses stay relatively steady (there is a very slight increase as the current concentrates towards the outer skin of the wire), the inductive and capacitative losses increase more significantly.

VDSL generally is at frequencies higher than ADSL; and thus has even higher losses per unit length, although in general it is the same line/wiring from the PCP thus very much shorter, with again differences between the VDSL Upstream Attenuation and the corresponding VDSL Downstream Attenuation.

On ADSL, my Downstream Attenuation was about 24 db over 1,286 Metres.

On VDSL, my Downstream Attenuation is about 13.4 db over 300 Metres.

Thank you all for your input.

Thanks ian007jen - didn't know the relationship between interleaving and FEC. That answers that one!
It generates the question - is this set by user, provider or either?
Thanks also for the 'error seconds' stat not reset except on reboot info.
Afraid I have to hold my hands up to use WiFi even though I'm almost sat on top of my router and I'm on a channel that nobody else in the building is using. I *did* use a cable a couple of times and unplugged every IO generating device and I still got poor throughput so I got lazy again. I'll check throughput against this site and post.

The frequency altering attentuation. Oh me oh my. My electronics lectures "frequencies of interest" soundbyte is coming back. Differing frequencies will see different impedences and capacities. (? that the right spelling?) How could I miss that? Good Answer! Thank You eckiedoo. That ticks that box.

Thanks RobertoS for the other background info. )

Right... back to testing...

By the way - the faceplate where my wiring in this flat (c1974) appears has 'GPO' written on it! <Sigh! - the good old days> The master socket is on an extension of the same cable, pulled for about 8-9yds through the loft. Apart from rolling your eyes or laughing out loud - is anyone going to know if there's anything definitively bad about wiring from that era?

Either way - it all worked reasonably, if a little jittery until this week. It's these drop outs that are the killer as any secure connection will only stand so much delay before it'll give up and force a user-refresh/retry. If the slow bandwidth was reliable I could wait for remote desktops to paint themselves onto the screeen etc.! Think I'm going to read the next post in this forum... Drop outs? What are they and where do they come from?

Be wary of WiFi Channels.

It may be that there is a lot of non-WiFi interference at that channel, from other unlicensed devices using the same 2.4GHz allocation, such as microwaves, baby alarms etc, so the other WiFi devices in your vicinity are actually avoiding it!

There are arguments For and Against setting to a fixed channel, as opposed to letting the WiFi system AUTOmatically choose what it reckons is the "best" channel/s.