FTTC speed prediction lower for flats?

I am intending to switch to BT Infinity shortly, but was slightly disappointed with the speed predictions: around 48 Mbps download and 8 Mbps upload. I am about 250 metres from the cabinet (the BT Wholesale checker helpfully confirms which cabinet I'm served from). I decided to check what sort of speeds some of my neighbours would get, and discovered that houses in the same road, which are maybe 50 or 100 metres further away from the cabinet, are predicted to receive 56 down and 16 up.

Why would the prediction for my flat be so much lower? Do BT predict lower speeds for flats because they assume poor internal wiring or something? The ADSL 2+ speed it predicts is pretty close to what I get, but I note that it predicts lower speeds for same neighbouring houses. Why is their ADSL predicted to be lower speed, but FTTC higher?

The likely speed I can expect is important as it could determine whether I go for Option 1 or 2 (given that the former is now available as an Unlimited option). Any explanations of how the predicted speed is determined would be appreciated.

That's pretty poor speed estimate @ 250m from the cabinet. I'm guessing either the estimate is way off, the flats are off a different cabinet, or the D sides take a long route to the flats

In reply to a post by Ribble:

... or the D sides take a long route to the flats

Yes, to the OP - are you measuring your distance from cabinet as the crow flies, or the likely path it takes via overhead cabling and/or footway boxes? Using Google Maps and its measuring tool, I'm 143m from my cabinet in a straight line, but when plotting the course of my line's route to the cabinet, it's 329m. So if you were measuring in a straight line, and the percentage increase between that distance and the cabling distance to cabinet is similar to mine, your copper VDSL link to the cabinet could be something like 600m.

For me the checker estimates 66.9 down, 20 up. So if your actual line is twice as long as mine, that makes some sense, although I still think it's a bit low, like Ribble. There's also a huge disparity with the upstream estimates.

Edited by deleted (Wed 06-Mar-13 06:02:09)

I guess it would lso depend how the phone line is routed within the building. OP doesn't say how big the block is that he lives in.

I lived on the 4th floor of a block of flats years ago and all the phone lines came into the building to a central junction box in the basement. From there the lines were routed through ducting to the various floors. I would probabaly estimate my line was about 400m long just within the building.

my house is 3 flats.

They all share the same dropwire so it may be related to the extra crosstalk expected from that?

The predictions also use information on how your ADSL line was performing relative to others locally in terms of noise.

Being in flats means that there is a possibility of a greater density of local noise/RFI from the various electrical/electronic appliances. Cross talk might affect the Upstream to a greater extent but on the Downstream the pairs will be running together from cabinet to building so the additional few metres will not do much.

Ideally in a block of flats, if there was access to the point where all lines entered, it would be ideal if you could install the modems there, and then run Cat5e for data and use the existing phone pair for voice.

In reply to a post by MHC:

The predictions also use information on how your ADSL line was performing relative to others locally in terms of noise.

Being in flats means that there is a possibility of a greater density of local noise/RFI from the various electrical/electronic appliances. Cross talk might affect the Upstream to a greater extent but on the Downstream the pairs will be running together from cabinet to building so the additional few metres will not do much.

Would this have such a great effect on FTTC but not on ADSL 2+? My predicted speed for ADSL 2+ is correct, and is higher than for the very same houses that seemingly have better FTTC speeds.

In reply to a post by Ribble:

That's pretty poor speed estimate @ 250m from the cabinet. I'm guessing either the estimate is way off, the flats are off a different cabinet, or the D sides take a long route to the flats

As I said, the BT Wholesale checker tells me which cabinet I'm on, it matches the stencilled number on the cabinet and is the same for the houses.

Obviously I don't know the route the cables take, although if they took a more roundabout route, they'd pass the houses first, in which case they should have a higher ADSL prediction too. It's the discrepancy between the predictions for FTTC and ADSL2+ I really don't understand.

Thanks for all the replies so far.

If I do go for Infinity Option 2 and discover the speed makes it not worth it, will BT let me downgrade to Option 1? I have searched old posts and it seems they will allow downgrades, although they reset the contract length to 18 months. Could anyone confirm that?

I will of course post results of what speed I actually get!

They could be on a completely different E side cable to the exchange

Well i live in a flat thats over 400M to the cab and my FTTC gives me 75 / 16 + . The speed predictions tend to be a bit on the low side for FTTC.

Why not go for option 1 initially, get a HG612 modem off eBay (if you get the ECI modem when the install is done), load the custom firmware and check the attainable rate. If you think the attainable is worth it, upgrade to option 2.

In reply to a post by Spasch:

Why not go for option 1 initially, get a HG612 modem off eBay (if you get the ECI modem when the install is done), load the custom firmware and check the attainable rate. If you think the attainable is worth it, upgrade to option 2.

Apart from the extra hassle and work required, BT will charge an extra £25 to activate Option 1, which I doubt I'll get back if I upgrade!

Ah, I see.

MHC I am not convinced on that the pairs will be together all the way from the cabinet.

eg. on a D side swap the dropwire isnt changed, its changed elsewhere, so its possible for adjacent pairs on a dropwire to not be adjacent in the ducts.

They may not be, but the point is that by the time they get to the building any cross talk will have occurred between various pairs in cable bundles and the signal levels will be so low that the chance of any occurring between pairs whether or not they have been adjacent before will be minimal.

stories I have read say otherwise.

I had Infinity installed on Monday. The engineer told me the speed on my line was 59 Mb/s down and 17.5 Mb/s up. That would be what my neighbours were predicted to receive and what I would expect given the cabinet location. However, I noticed he and his mate seemed to discuss seeing lots of CRC errors, although they didn't say anything about this or its impact on speeds to me.

When I run speed tests, I'm getting pretty consistent results of 55 down and 9 up. This is far closer to the original predicted result for my flat. Could the discrepancy in upload speed be due to a high error rate? And if so, is there really any point in contacting BT to complain about not getting the promised 17.5 once the two week settling in period is over?

Also, am I right in thinking that Infinity 1 limits the sync speed to 9 Mb/s upload, so the large number of CRC errors would mean I would only get around 5 Mb/s on the cheaper service?

In reply to a post by jrawle:

When I run speed tests, I'm getting pretty consistent results of 55 down and 9 up.

If you have an antivirus, advert blocker or anything of that ilk try the test without it/thenm. This is speaking from personal experience. There is a long running history of users who have 'accused' TBB's speed tester of having an upstream fault. I was one of them until I discovered that closing down (not just disabling) my ad-blocker fixed the problem.

Also try other speed testers. Oddly, speedtest.net was mostly immune to the problem and it's always good to have a selection of testers rather than relying on just one.

In reply to a post by Andrue:

If you have an antivirus, advert blocker or anything of that ilk try the test without it/thenm. This is speaking from personal experience. There is a long running history of users who have 'accused' TBB's speed tester of having an upstream fault. I was one of them until I discovered that closing down (not just disabling) my ad-blocker fixed the problem.

Also try other speed testers. Oddly, speedtest.net was mostly immune to the problem and it's always good to have a selection of testers rather than relying on just one.

Most of my tests were with speedtest.net. I always achieved the max upload speed when on ADSL. I do have an ad blocker installed, so will try it without that to eliminate that possibility.

In reply to a post by jrawle:

In reply to a post by Andrue:

If you have an antivirus, advert blocker or anything of that ilk try the test without it/thenm. This is speaking from personal experience. There is a long running history of users who have 'accused' TBB's speed tester of having an upstream fault. I was one of them until I discovered that closing down (not just disabling) my ad-blocker fixed the problem.

Also try other speed testers. Oddly, speedtest.net was mostly immune to the problem and it's always good to have a selection of testers rather than relying on just one.

Most of my tests were with speedtest.net. I always achieved the max upload speed when on ADSL. I do have an ad blocker installed, so will try it without that to eliminate that possibility.

I always got full upload speed with ADSL as well. I'm still puzzled about it but it's a known issue. Presumably it relates to the amount of processing so it's proportional to the speed, ADSL upload is too slow to notice a problem.

The upload issue on the old tbb speed test started in late October 2011 or early November, but seb has no idea what caused it as I can't give an even approximate date.

My 40/10 always showed over 8Mbps then suddenly dropped to the 6Mbps+ range.

07/12/11 00:24 Wed 35688 Kbps 6007 Kbps
18/11/11 23:57 Fri 36249 Kbps 5999 Kbps
24/10/11 12:24 Mon 36377 Kbps 8316 Kbps
10/10/11 10:26 Mon 34956 Kbps 8301 Kbps

Speedtest.net continued to give the 8Mbps+ readings. NetWorx realtime upload graph was steady at the reported speed of each.

On the new tbb Flash test the upstream seems correct. Over 14Mbps on a 15.2Mbps sync.

In reply to a post by RobertoS:

On the new tbb Flash test the upstream seems correct.

Unless you're using a Mac, where Adobe Flash for Mac seems to have issues with the tbb speed test. (Current workaround is to use Google's Chrome browser for Mac which includes its own embedded flash, but one day that will get updated).

alot of upload issues on speedtest sites are caused by a too small tcp send window.

on speedtest.net eg. all the uk servers have the problem as well as the BTw tester, but ookla made a BETA .nl server on speedtest.net which has a bigger buffer and as such produces full speeds for uploads on the test.

Just out of curiosity, what size window are they using (to save me getting wireshark fired up...)?