In reply to a post by smurf46:

Apologies for jumping on this thread but I'm intrigued. Half the time I've checked the BT speedtester it reports a throughput in excess of the line profile by up to 1.5Mbps.

Please post a copy/paste of the tester result.

I also use a Fritz!Box 7390 on an FTTC line (for stability - the OR modem setup had download speeds constantly up and down, so it was suggested I tried a "better" router) and it reports an actual (not the even higher attainable) throughput at 6Mbps above the reported IP profile.

What connection speed does it report in these circumstances? Does the router really report a throughput speed?

I'd say this was an error but both the router and the tbbMonitor running on the PC regularly also report actual downloads running up to a couple of Megs in excess of the IP profile. I suppose it could be that the profile jumps about a bit and I've just used the test at the "wrong" times?

Are you wired or wireless to the router, and what IS product do you use?

Some AV suites do odd buffering while checking data, and this can result in spiky results for tbbmeter, and other monitoring applications.

In tbbmeter this can be spotted by switching out of All Internet Traffic (UDP/TCP) and using the data from the machines network interface i.e. Disable TCP/UDP Monitoring.

Thanks. The "old" results were with the OR modem and IDNet supplied Netgear WNR1000:
1. IDNET fibre: 20oct2010
Download speedachieved during the test was - 29727 Kbps
For your connection, the acceptable range of speedsis 12000-28778 Kbps .
Additional Information:
IP Profile for your line is -28778 Kbps
Upload speed achieved during the test was - 7389 Kbps
Additional Information:
Upstream Rate IP profile on your line is - 10000 Kbps
2. a result a couple of days later showed a lower throughpuof just below 28Mbps, but a lower profile of around 26.5 Mbps (which is what was expected on my line). Unfortunately I didn't cut and paste that result.

However it seems to have corrected itself as tonight I've just checked with the Fritz!Box modem/router and BTW shows:
15nov
Download speedachieved during the test was - 29517 Kbps
For your connection, the acceptable range of speedsis 12000-33867 Kbps .
Additional Information:
IP Profile for your line is -33867 Kbps
Upload speed achieved during the test was - 5805 Kbps
Additional Information:
Upstream Rate IP profile on your line is - 10000 Kbps

Fritz!Box stats (which have been consistent) are:

Receive Direction / Send Direction
Max. DSLAM throughput kbit/s 35000 / 7200
Min. DSLAM throughput kbit/s 17504 / 3600
Attainable throughput kbit/s 42100 / 6183
Current throughput kbit/s 34992 / 6184

Latency 8 ms / 0 ms
Bitswap on / on
Impulse Noise Protection 3.0 / 0.0

Signal-to-noise ratio dB 10 / 6
Line attenuation dB 20 -
Carrier record A43 A43
Profile 17a (reported wrongly, a known fault)

Error seconds/ With many errors/ Seconds With Remediable Errors (FEC) / last 15m / sNot Remediable Errors (CRC) per second / last 15m:

FRITZ!Box 304 / 0 / 6122 / 11436989 / 0.15 / 0
DSL central exchange 508 / 0 / 0.24 / 1 / 0.08 / 0

Laptop connected by ethernet cable, and running Norton 360 (v4). Interleaving has been on throughout.

Profile 17a (reported wrongly, a known fault)

Maybe you have switched over to 17a for real now.

It would explain your downstream going back up, coupled with a slight loss in upstream. Which is what most people have been seeing when switching over.

I thought of a switch to a 17a profile but according to the Fritz!Box graphing the frequency is still limited to 8842KHz which I understand is still 8b, and wouldn't a 42Mbps attainable limit be the sign of a very poor line and much longer distance from the cabinet than my 700m or so?

Edited by deleted (Tue 15-Nov-11 20:13:01)

In reply to a post by mr_bean:

Your line is probably connected at 39999 but the IP Profile is max'ed to 38717

Interestingly along with profile 17a I found my IP profile speed now reported as 38715Kbps. Not sure whether this slight change is universal.

Sync @ 39997Kbs^-1

My original line went throught 2 stages. The first 2 days was without DLM, when the IP profile was 38717. After DLM intervention (at 48 hours) to fix 4% packet loss, this dropped to around 34799. I went through this twice, with the second time when regrading to 10Mbps upstream.

When 17a was introduced, my downstream IP profile went up to 38716.

I haven't modded the box yet, so I don't know any more details yet. All existing figures come from BT's speedtester.

So slight differences do occur, without an obvious reason why.

In reply to a post by smurf46:

I thought of a switch to a 17a profile but according to the Fritz!Box graphing the frequency is still limited to 8842KHz which I understand is still 8b, and wouldn't a 42Mbps attainable limit be the sign of a very poor line and much longer distance from the cabinet than my 700m or so?

BT used profile 8c, but with a truncated spectrum - only going up to 7.05MHz. This is a limit point for the 997 bandplan; between 7.05MHz and 12 MHz, the spectrum is used for upstream (on profiles 12a, 17a and 30a).

If you are seeing your box making active use of anything above 7.05MHz, then you are definitely seeing the use of profile 17a. BT are actually employing both the 17a profile *and* switching to the 998 band plan, so you would then see downstream frequencies up to 8MHz, and then a switch to upstream for 8MHz up to 12 MHz.

As for 700m: You may well be reaching the limit for this line. With "plain" 17a, you would probably get 40-50Mbps. Once they introduce vectoring, you might get up to near 80Mbps. Take a look at figures 5 and 6 in this document about vectoring: Ericsson on VDSL2 Vectoring

As I understand it vectoring only helps recover bandwidth lost to crosstalk, I think it's highly unlikely crosstalk is accounting for 40-50 Mbps of lost bandwidth.

Very helful posts from everyone. Thanks.

I think it all confirms my first thoughts, which I'm happy about, that the line is maxed out on the 17a profile. It's a bit difficult to tell with the scaling on the router graph but I'm certain my tones don't get as far as 8MHz (though get close) so there's no upper upstream range: just a small one at the botton and a large one in the middle: i.e. there is two ranges for each of the upstream (first) followed by downstream, in each case.

Might the inability to get an upper tone range on the upstream - due to poor line quality and distance - also be why the upstream has the limit of 6Mbps out of the available 10 which it got very close to on the 8c profile?

Edited by deleted (Wed 16-Nov-11 09:46:25)

In reply to a post by cmn:

As I understand it vectoring only helps recover bandwidth lost to crosstalk, I think it's highly unlikely crosstalk is accounting for 40-50 Mbps of lost bandwidth.

I'm only going on what the telecomms experts are showing in their experimental data, but they suggest that "only" recovering the bandwidth lost to crosstalk is a significant proportion of bandwidth.

Take another look at that Ericsson document on vectoring, and especially at figures 5 and 6.

It looks to me, from figure 5, that someone at 200 metres can expect a best-case, vectored result of 175Mbps, and a worst-case, non-vectored result of 70Mbps - a difference of over 100Mbps. That's a *measured* result, not a theoretical one, and it shows crosstalk has a serious effect

Figure 6 show results over 0.5mm cable, using hardware limited to 100Mbps max. The results there show that someone on 2000 feet of cable (600 metres) gets 68Mbps with crosstalk, and gets the capped 100Mbps either without crosstalk or with vectoring. The figures are broadly in line with the "600 metre" results in figure 5, and I think are a better reflection on both the equipment in the UK and the limit that BT will place on the product offerings.

On my 600-650 metre line, I'm estimating that the 17a profile and the addition of vectoring will give me at least double the speed.