Squeezing a few more megs from a slow FTTC

I want to try and squeeze a bit more speed - especially up - from my FTTC. Currently I'm on a 40/10 deal and the router says:

Attainable: 40058 - 13156 (down - up)
Current is : 38447 - 9997

INP is 40 - G INP is ON & OFF - SNR 8 & 8 - Attenuation 20 & 27 - Length is ~ 550m

If I attach my FrirzBox direct to the Openreach 'test' socket I get a small improvement (that disconnects an unused extension).

The attainable increases to 41828 although the up speed sees a slight drop to 13025.
The actual speeds were 39998 down and 9997 up. So the maximum available by my package.
IMP was 38, SNR 7 & 8, all other figures the same.

A very small download speed increase of 1000 isn't going to be noticed, but a one third increase in upload and I might be willing to pay the extra fiver.

So what more can I do to increase this increase?

Two things to take into account.

According to the OR address checker my neighbours have better predicted speeds than I do. Those are further from the cabinet than I am so that shouldn't be correct. (Yes, I'm 99% sure I know the duct run and direction.)

My prediction shows (for clean line) 60 - 41.6 down and 13.7 - 8.5 up. So my actual speeds are at the low end of down and the high of up!

Next door is closer to the cabinet by around 12 metres worth of cable. The predictions are 69.8 - 49 down and 18 - 11.9 up. Fair enough, they're closer, how about the other next door who are another 10 meters away? The exact same predictions. So why do I - in between - have lower predictions? The next two houses have a lower prediction but its faster than mine at 64.8 - 44.9 and 15.6 - 10.

Normal times I'd bang on their doors and ask their speeds...

The other thing is my internal cabling. I asked the OR installer nicely and he jointed the incoming pair to an Ethernet cable running about 10 or 12 metres to a more suitable location for the Master socket (and my router). So I could shorten my line length and any interference issues by having the Master socket relocated to where the line comes into the house, where it was designed to go. Then I'm thinking of having a modem and connect that back to the router (now in router only mode) at its current location using the Ethernet.

But at the end of the day the signal sill has to travel the same distance to my router. Will I be gaining / losing anything or does the fact that I have the modem as close to the cabinet and Ethernet internally going to help speed me up?

Thanks!

Upgrade to an 80/20 service.

Distance is obviously only part of the story, cable type, joints, etc can all impact speed. It is quite possible that your line has an extra joint or 2 causing it to be slower than your neighbours. It is possible that any number of things could be impacting your line.

Also possible you have some noise in your own house impacting the line - maybe the TV is causing noise or something similar.

Openreach are not going to investigate this as your line is within their expected parameters. If they investigated every line that was slower than a neighbour then that would be all they would have time to do and could result in massive lengths of cabling being replaced.

Also, the attainable is I think normally what your line would get at 0dB SNR - so bearing in mind that no line syncs at that you would never hit the attainable. You may only actually get a small amount of that up attainable available to you - you might even find that the line could rebalance and move slightly more to the downstream impacting the upstream - that could mean you would get a small boost in both that would be negligible.

EDIT : On the question of length to the router. Shortening the line could well improve things - it is the length to the router that counts, what you do with ethernet after the router would not make any impact on your connection (well, using PowerLine might but we aren't talking about that). So, shortening it to where the master socket was originally MAY improve things - it also may make no noticeable difference whatsoever.

Edited by ian72 (Tue 26-Jan-21 13:29:46)

Also, the attainable is I think normally what your line would get at 0dB SNR -

I don’t think that’s correct Ian. As someone who looks at attainable rates pretty often, I‘d think it was an approximation based on a 6db margin.

The OP says they’d like more upload ... the guestimate says 13up, which would be an improvement on the 9999 they have now ...

I am not certain but have seen other posts where people can't reach the "attainable" rate. I wouldn't want to see the OP assuming that was the speed it would go when there seem to have been examples posted where people don't get to the attainable rate.

The OP needs to decide whether an extra £5 per month is worth it for what could be a relatively small increase.

The attainable rate is accurate on lines running fastpath and with G.INP.

It is inaccurate with Interleaving enabled.**

The max attainable uses the current target SNRM set by the DLM.
If the target margin drops the attainable increases, and vice versa.
There's no 0dB involved the the calculation.

**Some newer Lantiq chipsets have an accurate attainable even with Interleaving enabled.

Edited by j0hn83 (Tue 26-Jan-21 14:23:31)

In reply to a post by Zarjaz:

As someone who looks at attainable rates pretty often, I‘d think it was an approximation based on a 6db margin.

So what do you think that is for the high or low prediction? I dropped my SNR from 8 to 7 by using the test socket but only gained 1000 Mbps. But the upload SNR stayed the same. (Someone can explain why that is?)

The OP says they’d like more upload ... the guestimate says 13up, which would be an improvement on the 9999 they have now ...

Yes, and in answer to your other post, yes getting an 80/20 connection will do that. My router says the line is capable of 13 odd up. The biggest question is whether going to the trouble and expense of having the Master socket moved, buying a new modem and recabling will add a further worthwhile speed increase.

When you say you look at "attainable rates" often, do you mean what the router reports rather than the OR address checker suggests?

I see if I do a check using my phone number I get the same figures. The only difference being observed speeds: 32.29 down, 12.85 up but that was on 31.03.2020. I was last tested for bridge tap (U), VRI (N) and NCEFaceplate (N) on 14.01.2021.

Just to add further to my confusion over how OR are calculating predicted speeds, the last house in my row has the fastest predictions of all at 69.8 - 49 down, 18 - 11.9 up. Which makes no sense to me.

If you went for the higher package you should achieve around 50Mb downstream sync with a 3dB SNRM target.

That could be part of the reason why your neighbours have a higher estimates than you.
My estimates climbed above my neighbours when i dropped to a 3dB target.

Then I'm thinking of having a modem and connect that back to the router (now in router only mode) at its current location using the Ethernet.

But at the end of the day the signal sill has to travel the same distance to my router. Will I be gaining / losing anything or does the fact that I have the modem as close to the cabinet and Ethernet internally going to help speed me up?

The signal doesn't travel the same distance to your router.
The xDSL signal ends at the modem. xDSL drops over a long distance.

The link between the modem and router is Ethernet and you'll get a line run indoors before that drops below a 100Mb link speed.

If it's a short distance between this joint and the modem it may make little difference.
The shorter the line carrying the xDSL link the better though.

I'm also not a fan of having 500m+ of CW1308 twisted pair then a random length of cat5 in the home, all part of the xdsl link.
I recommend CW1308 all the way to the modem! Cat5e from the modem to the router.

What are the impacted figures?

From what you have said, there is very little you can achieve inside the property so the reason for slow speeds are likely to be external and could be down to gauge of the incoming wires, large amounts of noise ...

Did the installer use a pair in the Cat5e (Ethernet) cable? Does not matter which, just that it is a pair. And the extra few metres - small when compared to 550, but if the socket was moved and the router still right next to it, you might see a small 200-500k increase, stressing might.

Disconnect the unused extension, and plug into the face plate. the small differences in Upstream between 13025 and 13126 will change regularly and it could be that a noise source changed between checks. With that you might see a sync between 12.5 and 13.2. If you believe the upload increase is worthwhile, go for it, you will see a small increase in downstream too, and if everything stays stable, possibly the downstream SNR might drop to 5, 4, 3 dB with an increase in speed.

In reply to a post by j0hn83:

If you went for the higher package you should achieve around 50Mb downstream sync with a 3dB SNRM target.

I could live with that and worth the trouble. But what's my current SNRM? I know my SNR (8dB) but that's not the same, is it? If my SNRM is above 3dB now, how do I get it down?

The signal doesn't travel the same distance to your router.
The xDSL signal ends at the modem. xDSL drops over a long distance.

The link between the modem and router is Ethernet and you'll get a line run indoors before that drops below a 100Mb link speed.

I think I know this really. They're different signals, the xDSL will degrade but once I've translated it to Ethernet I can move it through the house with less loss.

I'm also not a fan of having 500m+ of CW1308 twisted pair then a random length of cat5 in the home, all part of the xdsl link.
I recommend CW1308 all the way to the modem! Cat5e from the modem to the router.

I'm afraid builders were involved...