FEC, CRC, HEC, And error seconds... what are they?

Hi Guys, still battling with BT over our internet issues...

Was wondering if anybody could help me make sense of my stats?

Connection Information

Line state: Connected
Connection time: 0 days, 03:46:26
Downstream: 1.031 Mbps
Upstream: 448 Kbps

ADSL Settings

VPI/VCI: 0/38
Type: PPPoA
Modulation: G.992.1 Annex A
Latency type: Interleaved
Noise margin (Down/Up): 4.3 dB / 11.0 dB
Line attenuation (Down/Up): 73.5 dB / 31.5 dB
Output power (Down/Up): 16.3 dBm / 12.4 dBm
FEC Events (Down/Up): 37616106 / 12139
CRC Events (Down/Up): 312583 / 6071
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): 878806 / 5326
Error Seconds (Local/Remote): 16136 / 5189

Now I know we have a long line (7.5-8km), but by my maths when the internet is in full use we are accumulating 2,723 FEC events, 25 CRC events, 50 HEC events, and 1 error second, every second! just seems rather a lot to me.

BTW speedtest comes back at 0.5Mbps with a 0.75 IP profile. Which would be ok-ish, but, I suspect due to our error rates, real world web browsing and download performance is inconsistent and dial-up slow! That was the second test, first one came back as 0.02Mbps!

Basically our largest complaint is that we used to have a rock solid 1Mbps connection for 7 years, but the past 2/3 years have been patchy and below par. We have finally decided to raise it up the BT's executive level complaints department, but we are still getting the same treatment ("your line tests ok", "you have a long line, that's as good as you're going to get", etc...)

Sorry about the jumble, but help would be much appreciated.

Thanks, JF.

Errored seconds definitely not looking good. I will leave others to comment fully on the stats but wanted to mention one word of warning.

As you have a very long line there is the possibility that if you push the issue that BT Openreach will declare your line as unserviceable and will refuse to provide you any broadband at all. This may not happen but be prepared for the possibility that you could go from having something to having nothing (Openreach would not take account of past performance if the costs to rectify are too high).

You may actually get a better connection if they limited it to a fixed 512Kbps rather than attempting to run it at these speeds. A higher SNR might also help things (4.8dB is not much of a margin for a long line - I've seen shorter lines running at 12dB margin just to keep stable).

Router Line Stats - Errors

Your ES are far too high. Your NM has drifted low, indicating noise on line. It is usually 6dB or probably more on such a long line. What was it immediately after re-syncing? That's what matters.

Have you tried running from the hidden test socket?

You are running on an extremely long line and could be said to be lucky to get any BB at all.

Okey doke, fresh line stats for you here

ADSL Line Status

Connection Information

Line state: Connected
Connection time: 0 days, 00:01:21
Downstream: 1.031 Mbps
Upstream: 448 Kbps

ADSL Settings

VPI/VCI: 0/38
Type: PPPoA
Modulation: G.992.1 Annex A
Latency type: Interleaved
Noise margin (Down/Up): 7.1 dB / 12.0 dB
Line attenuation (Down/Up): 73.5 dB / 31.5 dB
Output power (Down/Up): 16.2 dBm / 12.4 dBm
FEC Events (Down/Up): 120 / 24
CRC Events (Down/Up): 2 / 0
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): 5 / 0
Error Seconds (Local/Remote): 2 / 0

We have had several brand new master sockets (the engineers keep breaking them when they take them off!) and a proper BT Openreach filter faceplate. There is no internal house wiring. Only one BT homehub, and one cordless phone plugged into the master socket.

So I have just measured our line (thanks google maps!) and it is about 5.8km of main underground cable, then 1.9km of overhead cable from the exchange. We are second to last on this line (the last is an extra 2.2km of cable up the road!) and there are about 25-28 total customers on the entire run served by the underground cable.

Now, just as a side point, when you dial 17070-2 for the quiet line test, and get an assistant to dial any of the 25-28 properties on the run, you can hear the ringing pulse of their phone through the quiet line. It used to be louder than it is now (after several small repairs), but it is still there. Out of the roughly 20-30 engineers (I wish I was kidding) that have been out 1 suggested that it was an exchange equipment fault and referred it back to BT Operate (no further action), and another couple believe it is an underground fault, but cant do anything about it (overhead guys) and BT probably won't sort it (traffic lights = £££'s)

Sorry its all jumbled! I'd appreciate it if anyone can make sense of it, or knows the super secret word that magically makes them fix it... or the masons handshake.

Thanks, JF.

EDIT - Also meant to say it disconnects and eventually reconnects at a significantly lower speed when there's heavy rain. Yes, I've told everyone.

Edited by deleted (Mon 15-Sep-14 16:41:08)

Yes, you have a line fault. Your phone provider needs to get Openreach to fix it.

Alternatively, it's very likely the same fault is causing your broadband problems, so your ISP should get Openreach to fix it.

Edited by deleted (Mon 15-Sep-14 17:19:55)

Ahah! But BT really is dragging their feet! over two years, and we swapped from TalkTalk to BT over a year ago with the promise of getting better service.

Every time an engineer comes out they plug in their tester, the line tests ok except for a failure on discapacitance, which they blame on the length of the line then they go on to the next job.

Here's a way to get it fixed

In reference to aa.nu/broadband-trial.html:

Give us a month and we'll sort your problem broadband line or your money back.

If you are migrating your service to us, even though you know you have a problem with your broadband line, we'll take on the fault. We'll tackle the problem and get it fixed within one month. If we don't then you can migrate away and owe us nothing for your migration to us and your service charges for that month.

In reply to a post by BatBoy:

Here's a way to get it fixed

In reference to aa.nu/broadband-trial.html:

Give us a month and we'll sort your problem broadband line or your money back.

If you are migrating your service to us, even though you know you have a problem with your broadband line, we'll take on the fault. We'll tackle the problem and get it fixed within one month. If we don't then you can migrate away and owe us nothing for your migration to us and your service charges for that month.

If they do fix it then OP would have a minimum term of 6 months at £25 per month plus a £1 migration fee and a free router (Home 1) - definitely sounds like a good plan

they plug in their tester, the line tests ok except for a failure on discapacitance, which they blame on the length of the line

That'd be right. On very long lines the end to end PQ (pair quality) test will show a fail on the dis capacitance. Despite the overall 'red cross' the figure should still be between 98 to 100%

The figure that should be noted next time would be the AC balance. 55db and above is a pass, but for DSL 60db and above, better still above 65db would be preferable. AC balance is a measurement of the pair to repel external signals.

http://aa.nu/broadband-trial.html

I use that before on Plusnet as it much better. This isp know best to fix your line no problem as they work closer with BT. I was surprise all other isp's won't do it even Plusnet.

Edited by adslmax (Tue 16-Sep-14 14:23:31)

ok the way to deal with this is like this assuming your test results are reliably repeatable..

Report voice not bb fault.

Part of their diagnosis is doing a quiet line test.

When they start a test, ring one of those 25 properties from your mobile.

If the ringing is audible and the engineer then walks claiming no fault to fix, they breached contract.

Okey doke, had 2 vans and a cherry picker here all day yesterday.

So they started off swapping pairs on the main road UG run (5.8km ish) and managed to turn 2700kbps into 3500kbps at that point. However after that there is a further 1.5km of bundled pairs up a side road, then 400m from the junction into the property.

Anyway, after switching pairs on the main road to get 3500 they came to test at the house, but the speed here had not improved (JDSU syncing just under 1000). they then decided to swap the 400m of two pair (I think?) into the property. this improved things a lot and the homehub then synced at 1400kbps.

now 24 hours on the SNR has started to drift down again and the usability of the internet is getting worse.

Stats after 24 hours:
Connection Information

Line state: Connected
Connection time: 1 days, 00:00:29
Downstream: 1.406 Mbps
Upstream: 448 Kbps

ADSL Settings

VPI/VCI: 0/38
Type: PPPoA
Modulation: G.992.1 Annex A
Latency type: Interleaved
Noise margin (Down/Up): 5.3 dB / 13.0 dB
Line attenuation (Down/Up): 70.9 dB / 31.5 dB
Output power (Down/Up): 16.7 dBm / 12.4 dBm
FEC Events (Down/Up): 86840064 / 56084
CRC Events (Down/Up): 4197 / 622
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): 26942 / 411
Error Seconds (Local/Remote): 1821 / 323

IP profile still at 0.75, and I don't really want to reconnect just now, but I will if you think I should to show fresh line stats.

Tried the ringing pulse noise thing, but it is pretty faint now, and they guy said it was likely just induction.

Any thoughts welcome. Thanks, JF.

I seem to remember posts about marginal performance being improved by getting interleaving switched off (by the ISP). Something about fewer bad packets at the expense of higher latency. I'm no expert so what do others think?

Turning off interleaving is likely to make the line very unstable (and would reduce latency rather than increase). It is interleaving that normally helps a long line.

yep that CRC/FEC count he has, bad idea to turn of interleaving without a massive snrm bump The low sync doesnt give much room for a snrm bump either.

Thinking about it, the guy who was advised to switch it off had a rock-solid connection and wanted lower latency for gaming so, yes, not recommended in this instance. Thanks.

Some observations:

A) The downstream power is a bit low (17 dBm rather than 20 dBm). The extra 3 dB are desperately needed. But this may be beyond your control.
B) Replacing the wire improved the attenuation by 3 dB, which improved your line from 1 to 1.4 Mbps, with reduced error seconds. A decent step. Did they replace 0.4 mm wire by 0.5 mm wire?
C) Noise can vary throughout time. Try to resink when the noise is bad. The bandwidth will be lower, but you will get less errors. Alternatively try to change the snr margin settings from 6 db to 8 db or so, if you can.
D) At this snr margin level you should not get so many errors. Could be a modem issue.

Edited by deleted (Sat 27-Sep-14 07:28:24)

OK, so I went out and got a hold of a 2700HGV modem to try out, and the speed is fantastic! synced at 1888 with a noise margin of 6db on Friday, and errors much lower than the homehub. Also where the homehub needed reset every 12hours or so, the business hub just keeps chugging away!

I also downloaded linestats to run on the pc for a few days to keep track of things...

STarted off Friday afternoon 1888 and 6db, then on Sunday at 0715 it jumped to 1400 and 10db. Now, during the 1888 running it had a total of 9 tiny dropouts, but was generally pretty solid although I am now considering that there may have been other dropouts but linestats didnt pick them up. anyway, as of 0720 Tuesday morning its dropped again to 1184 and 12db! Not sure where to go from here, errors are no better than when it was at 1888.

A) How do I get the extra 3db?
B) Not sure, but sound like that could be right.

The Chairmans office are really keen to close this fault and give us £50 to go away, but I put her off till Thursday to see how it goes.

Thanks, JF.

EDIT: Router Stats

Since Reset Current 24-Hour Interval Current 15-Minute Interval Time Since Last Event
ATM Cell Header Errors: 12414 12414 9 0:00:08
ATM Loss of Cell Delineation: 1841 1841 3 0:00:08
DSL Link Retrains: 1 1 0 2:52:42
DSL Training Errors: 0 0 0 0:00:00
DSL Training Timeouts: 0 0 0 0:00:00
DSL Loss of Framing Failures: 2 2 0 2:52:42
DSL Loss of Signal Failures: 2 2 0 2:52:42
DSL Loss of Power Failures: 0 0 0 0:00:00
DSL Loss of Margin Failures: 2 2 0 2:52:42
DSL Cumulative Errored Seconds: 4604 4604 5 0:00:08
DSL Severely Errored Seconds: 159 159 0 0:47:07
DSL Corrected Blocks: 159605 159605 406 0:00:01
DSL Uncorrected Blocks: 15049 15049 9 0:00:08
ISP Connection Establishment: 2 2 2 12:00:25

Edited by deleted (Tue 07-Oct-14 10:18:28)

You probably need to set your linestats sampling interval to 5 or 7 seconds, the minimum such tools normally have, rather than the default.

That usually lets you see the margin degrade rather than just drop out, and I always found that more useful.

Also getting it to save the graphs (give those a nice short timespan as well so the graph gives more detail) and a CSV file are good ideas.

(Edit - the light suddenly dawned as to where your nick comes from! It has been worrying me for a while, as I recognised it but couldn't place it).

Edited by RobertoS (Tue 07-Oct-14 14:29:26)