Erratic 50Mbps service is it an infrastructure problem?

[hd1080ts]

Hi All,

Had 50Mbps installed last week in the Plymouth area and it's been a bit erratic to say the least.

First speed test was just over 2Mbps, it got better during the first 24 hours and hit 49.5 Mbps down and 1.5Mbps up a few times both directly connecting to modem and via GigE on a D-Link DIR-655 using OpenDNS.

Now speed goes up and down a lot and seems to stall at times, I use to have the old 20Mbps service and that was rock solid before I moved to ADSL when Virgin's Traffic Management got annoying.

Spoke to the Virgin 50Mb helpline and they accessed the modem and thought it all looked fine but have attached my modem info with MAC details removed for a second opinion.

I think it's Virgin Media's infrastructure that's the problem but want be sure as I have just under 28 days to cancel.

Thanks.

H

[Peter_]

The RXmer/SNR is rather low for 50Mb and the Upstream is borderline to high for 50Mb if you called the 50Mb Support Line you will not have got a Liverpool agent so call tomorrow after 0800 and you should get a member of the dedicated 50Mb team.

Check out my 50Mb power levels Sticky HERE (http://www.cableforum.co.uk/board/34896851-post3.html)

[hd1080ts]

Thanks, will give them a call.

[Sephiroth]

Although I'd not be ecstatic with 31 dB SNR/RxMER, it's just about within limits for 64QAM modulation including a headroom margin for the variations in power you get this time of year (temperature changes). The number of correctable codewords is the indicator of how bad things are in the downstream. The number isn't worrying at all at this stafe.

It's your upstream power that would worry me. Your modem is struggling. A visiting engineer would need to know what he/she is doing to resolve this. The CMTS to which you connect records the SNR/RxMER for the upstream channel and the correctable/uncorrectable codeword count. If VM looks at that, they'll see a problem that needs resolving.

If you look at your stats again, tell us if the upstream power varies. It might even rise to 58 dBmv, the maximum for your modem. Upstream is struggling and you've got the T3 errors to prove it.

[hd1080ts]

I have a principle technician visit booked for Wednesday

Thanks Sephiroth & Masque for your help.

Lastest Downstream/Upstream info attached.

H

[Peter_]

The Upstream is more of an issue and I expect the technician to sort it out for you.

[Sephiroth]

Ok - I take that back about the downstream SNR. Your correctable codewords count is rising.

But, if at the same time you have an upstream problem (high power, struggling to push through) - and yes, the upstream is moving around which is what happens in T3 error scenarios.

So, IMO you have a common cause fault between your modem (or the modem itself) and the street cabinet affecting both upstream and downstream. An example of common cause fault would be coax not screwed in properly somewhere. Another, would be water having got in somewhere, oxidation, dipole effect and reflections on the cable.

For this type of fault the so-called "principal engineer" needs to be longer that the usual 5 minutes. He needs to take into account what I've said and you need to be on top of this.

Before he goes, you need to see the following in several successive stats:

Upstream always < 50 dBmv (55 is a threshold, but healthy is below 50)
Downstream SNR > 34 dB. 31 dB is **** poor even if in spec.

Get his phone number for follow up as may be necessary.

HTH.

[Ignitionnet]

The correctable codewords count is rising so slowly it's a non-issue. 1400 in over 12 hours is nothing, the modem would have received hundreds of millions of unerrored codewords in this period. 31dB SNR is fine, it can potentially drop into the mid-high 20s. Nothing poor about it at all, it works! The target bit error rate on cable networks is usually <10^8, this modem has shown a rate of zero post-FEC and not a lot pre-FEC, 1400 corrected codewords out of at least 1.052 billion, about 10^6 pre-FEC error rate. Nothing too alarming and won't cause major performance issues, a corrected code word every 50 seconds or so.

Upstream wise a single T3 timeout in over 12 hours isn't cause for concern really either. 55dBmV isn't a threshold, the modems can go to 58dBmV on 16QAM. There is no healthy criteria that says it needs to be low it entirely depends on where the modem is in the network and how the network is set up. On a noisy network Virgin can actually increase the transmit power of all the modems by attenuation on the upstream port in order to improve SNR - this is commonly done on the 50Mbps CMTS.

There are no hard and fast figures on upstream, which is why the modem has such a wide variation in transmit power. A figure above 50dBmV doesn't indicate a struggling modem, on 50Mbps 58.21dBmV transmit power indicates a struggling modem. Some areas have an extremely long fibre run back to the CMTS so have higher transmit powers than others, in other cases a modem may be at the end of a long cable run and have a relatively low amplification return path for performance reasons.

I suspect there is an upstream impediment or load problem there somewhere causing issues however the present upstream transmit power is fine, as are all the downstream stats.

The downstream SNR is unlikely to be caused by an issue local to the cabinet. There's probably an issue closer to the headend that VM are aware of which is why it's set to 64QAM not the standard 256QAM.

Would be good to know what the problem is, could be something as silly as a loose connector somewhere :)

[hd1080ts]

Upstream power level has gone up and Downstream power level has gone down.

[Sephiroth]

It's a brave or foolhardy person that takes Igni on. He's the #1 guru and his generalisations are solid, of course. He's absolutely right about the long cable run and the effect on power. But I've seen Igni refer to 54 dBmv modems as cooking and capable of frying eggs.

In the OP's case, I was referring to the consecutive sets of 16 x T3 timeouts in the event log when concluding that there might be an upstream struggle. If you did your stats every few seconds when there are T3 timeouts, you might well see 58 dBmv. If upstream is struggling, it'll be for a noise related reason.

Turning to the downstream SNR, as I said, 31 dB is in spec for 64QAM. Now, Igni might well be right about it being at the CMTS end and he's certainly right about the reason for 64QAM (and thus tolerance of lower SNR). All I've done is to provide a balance of probability assessment seeing the poor downstream SNR which is lower in your case than most 64QAM SNRs reported here.

At least Igni and I agree on the loose connector possibility!

Next blast please!

[Ignitionnet]

It's a brave or foolhardy person that takes Igni on. He's the #1 guru and his generalisations are solid, of course. He's absolutely right about the long cable run and the effect on power. But I've seen Igni refer to 54 dBmv modems as cooking and capable of frying eggs.

Next blast please!

Hah #1 guru you're having a laugh, I'm a beginner and I'm sure real cable gurus regularly have a good laugh at my ramblings.

You can tell my knowledge improving from the nonsense I used to spout on this forum, back when I did actually think that 54dBmV was a cause for concern. It actually is on the older network but on the DOCSIS 3 network the specs dictate 58dBmV max power for 16QAM instead of 55dBmV on DOCSIS 1/1.1/2, hence my relative indifference to the number.

I got a lot more knowledgeable right about the time I learned I knew nothing. :)

---------- Post added at 12:32 ---------- Previous post was at 12:29 ----------

In the OP's case, I was referring to the consecutive sets of 16 x T3 timeouts in the event log when concluding that there might be an upstream struggle.

That was seen when the modem was trying to come online, you don't see that message on the Ambit once a modem is locked you see a T3 then a T4. Those messages actually may relate to a downstream issue as easily as an upstream one and in any event aren't an issue.

[hd1080ts]

At least Igni and I agree on the loose connector possibility!


In my line of work it always is!

[hd1080ts]

Engineer came.

Made noises about 50Mb helpline being over cautious and was trying to get out of doing anything.

Finally agrees to attenuate the line and does a test, kept saying that anything over 42Mbps is a pass.

Downstream Power Level was actually -0.2 earlier today and have attached latest status lots of correctable codewords.

Performance did improve but this evening bandwidth is up and down again.

Currently mulling over cancelling.

[Sephiroth]

They ought to do something about the SNR, IMO. It's in spec for 64QAM - but only just with headroom for marginal conditions. I'm reasonably sure that the FEC errors are down to the noise level, possibly even noise that gets in and is not reflected in the RxMER.

Having said that, correctable codewords do not slow you down as they're resolved in flight by the modem. Peak time speed degradation is normally part of your "up to" contract.

[hd1080ts]

Here's a speedtest using Virgin's own tester, it was around 45Mbps this morning, not so good this evening.

Download #1
------------------------------------------------------------------
File size transferred : 100.0 MB (104857600 bytes)
Total time taken : 35.11 seconds (35112 milliseconds)
Throughput : 2986.0 KB/sec [Kilobyte-per-second]
= 2.99 MB/sec [Megabyte-per-second]
= 23888.0 Kbps [Kilobit-per-second]
= 23.89 Mbps [Megabit-per-second]

Download #2
------------------------------------------------------------------
File size transferred : 100.0 MB (104857600 bytes)
Total time taken : 44.85 seconds (44854 milliseconds)
Throughput : 2337.0 KB/sec [Kilobyte-per-second]
= 2.34 MB/sec [Megabyte-per-second]
= 18696.0 Kbps [Kilobit-per-second]
= 18.7 Mbps [Megabit-per-second]

Download #3
------------------------------------------------------------------
File size transferred : 100.0 MB (104857600 bytes)
Total time taken : 24.06 seconds (24058 milliseconds)
Throughput : 4358.0 KB/sec [Kilobyte-per-second]
= 4.36 MB/sec [Megabyte-per-second]
= 34864.0 Kbps [Kilobit-per-second]
= 34.86 Mbps [Megabit-per-second]

Upload #1
------------------------------------------------------------------
File size transferred : 10.0 MB (10485760 bytes)
Total time taken : 54.58 seconds (54578 milliseconds)
Throughput : 192.0 KB/sec [Kilobyte-per-second]
= 0.19 MB/sec [Megabyte-per-second]
= 1536.0 Kbps [Kilobit-per-second]
= 1.54 Mbps [Megabit-per-second]

Upload #2
------------------------------------------------------------------
File size transferred : 10.0 MB (10485760 bytes)
Total time taken : 54.13 seconds (54134 milliseconds)
Throughput : 193.0 KB/sec [Kilobyte-per-second]
= 0.19 MB/sec [Megabyte-per-second]
= 1544.0 Kbps [Kilobit-per-second]
= 1.54 Mbps [Megabit-per-second]

Upload #3
------------------------------------------------------------------
File size transferred : 10.0 MB (10485760 bytes)
Total time taken : 62.39 seconds (62389 milliseconds)
Throughput : 168.0 KB/sec [Kilobyte-per-second]
= 0.17 MB/sec [Megabyte-per-second]
= 1344.0 Kbps [Kilobit-per-second]
= 1.34 Mbps [Megabit-per-second]

[Peter_]

Engineer came.

Made noises about 50Mb helpline being over cautious and was trying to get out of doing anything.

Finally agrees to attenuate the line and does a test, kept saying that anything over 42Mbps is a pass.


42Mb would be a pass as it is an up to 50Mb service.

The RXmer/SNR is still to low and that is not being over cautious.

[hd1080ts]

T3 Timeouts 43
T4 Timeouts 4

Upstream issues?

Event log (MAC removed)

Thu Apr 22 08:48:01 2010 Thu Apr 22 08:48:01 2010 Critical (3) Received Response to Broadcast Maintenance Request, But no Unicast Maintenance opportunities received - T4 time out;CM-MAC=;CMTS-MAC=;CM-QOS=1.1;CM-VER=3.0;
Thu Apr 22 09:16:47 2010 Thu Apr 22 09:17:03 2010 Critical (3) No Ranging Response received - T3 time-out;CM-MAC=;CMTS-MAC=;CM-QOS=1.1;CM-VER=3.0;
Thu Apr 22 09:17:33 2010 Thu Apr 22 09:17:33 2010 Critical (3) Received Response to Broadcast Maintenance Request, But no Unicast Maintenance opportunities received - T4 time out;CM-MAC=;CMTS-MAC=;CM-QOS=1.1;CM-VER=3.0;
Thu Apr 22 09:31:01 2010 Thu Apr 22 09:31:16 2010 Critical (3) No Ranging Response received - T3 time-out;CM-MAC=;CMTS-MAC=;CM-QOS=1.1;CM-VER=3.0;
Thu Apr 22 09:31:46 2010 Thu Apr 22 09:31:46 2010 Critical (3) Received Response to Broadcast Maintenance Request, But no Unicast Maintenance opportunities received - T4 time out;CM-MAC=5;CMTS-MAC= ;CM-QOS=1.1;CM-VER=3.0;

[Sephiroth]

Although I'd not be ecstatic with 31 dB SNR/RxMER, it's just about within limits for 64QAM modulation including a headroom margin for the variations in power you get this time of year (temperature changes). The number of correctable codewords is the indicator of how bad things are in the downstream. The number isn't worrying at all at this stage.

It's your upstream power that would worry me. Your modem is struggling. A visiting engineer would need to know what he/she is doing to resolve this. The CMTS to which you connect records the SNR/RxMER for the upstream channel and the correctable/uncorrectable codeword count. If VM looks at that, they'll see a problem that needs resolving.

If you look at your stats again, tell us if the upstream power varies. It might even rise to 58 dBmv, the maximum for your modem. Upstream is struggling and you've got the T3 errors to prove it.

BUMP. Igni might disagree - but the facts are still in evidence. You've got a noise problem on your circuit.

[Chrysalis]

why are codeword stats missing from 20meg modems? they seem useful for diagnostics.

[Sephiroth]

why are codeword stats missing from 20meg modems? they seem useful for diagnostics.

You bet. The swine stopped our own SNMP access so we couldn't get to those MIBs. Did I fread somewhere that access has been restored by some special procedure?

[hd1080ts]

Really bad today!

[hd1080ts]

It's getting worse!

[speedfreak]

blimey I feel abit happier with my line after seeing that! Did someone mention codewords :p:

[hd1080ts]

blimey I feel abit happier with my line after seeing that! Did someone mention codewords :p:

This should make you feel even happier :)

[speedfreak]

lol it does but its ruined by the fact I feel sorry for you ;) :p:

[hd1080ts]

lol it does but its ruined by the fact I feel sorry for you ;) :p:

How did the CEO Office get involved with your issues?

[speedfreak]

How did the CEO Office get involved with your issues?

I had to crawl on my hands and knees up to the CEO's front door wearing raggedly clothing begging "please sir,no more of this sir" :D

In reality this forum put me in touch with them but Id been in contact with VM for several weeks before that happened. The idea is to explore all avenues of support before reaching them, if everyone got in touch with them at the first signs of a complaint then the system wouldnt work, I think theres only 12 of them, but if you are that fed up then I suggest an email to the CEO to sort it out. if you havent already Id go on VM's support forum first and see if second line can help.

[hd1080ts]

In reality this forum put me in touch with them but Id been in contact with VM for several weeks before that happened. The idea is to explore all avenues of support before reaching them, if everyone got in touch with them at the first signs of a complaint then the system wouldnt work, I think theres only 12 of them, but if you are that fed up then I suggest an email to the CEO to sort it out. if you havent already Id go on VM's support forum first and see if second line can help.

My tale of woe is on the VM support forum.

http://community.virginmedia.com/t5/Up-to-50Mb-broadband/New-install-issues-43-T3-Errors-4-T4-Errors-and-poor-speed/td-p/48329

Customer services have extended my Money Back Guarantee period beyond the first 28 days as it does not look like it will be sorted by then.

My case is now "in the hands of Networks" and they are monitoring my modem.

Modem totally locked-up just now and had to be rebooted.
:rolleyes:

[speedfreak]

Wow. I thought I was patient. Id give networks till the end of the week and if its not sorted by then Id send an email to the CEO. Its odd that they all have the same tools available yet it took a CEO office guy to spot noise on my line. Didnt realise you had a thread over there already. The thing that my faults taught me was that patience comes in handy but Id keep that 28day thing going. Good luck. At least with your fault, things are that bad that if they are monitoring it it should be easier to fix rather than an intermittent fault and the more disconnects and resets the better whilst they are watching it, well lets hope so

[hd1080ts]

Thanks Speedfreak, Executive Office are now monitoring the situation.

[hd1080ts]

Had a Principle Technician visit who and did a very comprehensive health check resulting in replacing everything within the house (cables, isolator).

Good news is there was only one T3 error and no T4 errors over a 24 hour period.

However with the local connection fixed it is obvious to me that there are infrastructure limitations elsewhere in the Virgin Media network, which may be specific to the Newton Abbot UBR.

So I have cancelled and will stay with my 18Mbps ADSL connection for now.

Thanks to all the helped.