cancel
Showing results for 
Search instead for 
Did you mean: 

New Fibre Install - Great download, Poor upload speeds. - Please advise

WWWombat
Grafter
Posts: 1,412
Thanks: 4
Registered: ‎29-01-2009

Re: New Fibre Install - Great download, Poor upload speeds. - Please advise

Right, back to answer now...
Quote from: Thetwohoughs
Here we go!  Wink
I've also added another bitloading. Are the SNRperband of any use to you? I can upload those if required.

No, SNR per band isn't that much use, unless you see some serious divergence.
As someone else asked, the graphs with the complete set of tones (out to 4000) is best.
Your data from the "pbParams" starts to show a bit of the problem.
Quote
[tt]Yours
  VDSL Band Status   U0 U1   U2   U3 U4 D1   D2   D3
  Line Attenuation(dB): 5.6 27.2 44.4 N/A N/A 13.8 35.1 61.1
Signal Attenuation(dB): 5.6 26.6 42.1 N/A N/A 15.4 34.8 61.3
        SNR Margin(dB): 6.2 6.3 6.3 N/A N/A 5.9 5.9 5.9
         TX Power(dBm): 0.4  -15.6 4.5 N/A N/A 9.8 7.7 6.1
[/tt]

Attenuation in D3 looks low, by about 10dB. And the reason is obvious when you look at your Hlog graph once extended in frequencies. That dip at tone 2900 shouldn't be there, and (I agree with @EJS) shows a "bridge tap".
The dip in the downstream tones explains why your downstream bitloading isn't using the D3 frequencies as expected. As a consequence, the bit-loading algorithm has filled up tones in the D1 and D2 spectrum, so those tones contribute more bits than normal; the standard algorithm tries to use bit-loading in the higher spectrum if at all possible - I think it causes less crosstalk in the lower spectrum, leaving it for longer lines that can only use the lower spectrum.
However, that dip alone doesn't explain everything about the U2 spectrum. There the graph is just showing a few sporadic results!
For comparison, here's an old line of mine (650m of copper). That seems very comparable, except for D3.
Quote
[tt]My Line 1
  VDSL Band Status   U0 U1   U2   U3 D1   D2   D3
  Line Attenuation(dB): 5.4 27.7 40.8   N/A 14.1 34.4 53.1
Signal Attenuation(dB): 11.4 27.1 40.0   N/A 14.1 34.4 53.1
        SNR Margin(dB): 11.2 11.2 11.2   N/A 10.8 11.0 10.8
        TX Power(dBm): -3.7 -16.8   6.3   N/A   8.5   7.9   5.4
[/tt]

I've attached the Hlog graph too - but it dates back so far that it didn't ever display the upstream values.
Still, Hlog shows a smooth line as normal, with values *very* similar to yours. It really highlights the problem dip of yours at tone 2900.
And here's a more recent line; just under 400m and just under 80Mbps:
Quote
[tt]My Line 2
Attainable Net Data Rate:     22994 kbps     79732 kbps
Actual Aggregate Tx Power:       7.1 dBm     13.0 dBm
====================================================================================
  VDSL Band Status   U0 U1   U2   U3 U4 D1   D2   D3
  Line Attenuation(dB): 4.4 22.6 34.0   N/A   N/A 11.7 28.4 43.5
Signal Attenuation(dB): 4.4 21.3 32.6   N/A   N/A 16.4 28.2 43.4
SNR Margin(dB):   7.3   7.3   7.3   N/A   N/A 6.3 6.3 6.3
TX Power(dBm):   -1.6 -20.6 6.4   N/A   N/A 9.6 7.8 7.0
[/tt]

The hlog graph for this line is more recent, so includes the upstream. You can see that it is meant to follow a relatively similar curve.
Quote from: ejs
Perhaps the low upload speed is due to the amount of Upstream Power Back Off (UPBO). That means the transmit power of your upstream has been reduced so that when it arrives at the cabinet, it's the same strength as the upstream from someone with a much longer line. If this wasn't done, then your upstream signals would be much stronger, when received at the cabinet, than those from a longer line, and then your upstream could interfere with the upstream from longer lines.

I think UPBO has something to do with the upstream problems here - but only in the sense that it prevents the modem from adapting to the unusable U2 spectrum.
I think UPBO is acting as it should for a line of this length. This would have reduced transmit power, so have lower bitloading than the ideal (as we see in U1). It is expecting U2 to take up the slack, but it appears that the "bridge tap" issue is making U2 considerably less usable than expected.
Downstream, bitloading transfers to the lower spectrum to make up for D3. Upstream, I think, has UPBO getting in the way.
Quote from: Thetwohoughs
G.INP enabled after a forced resync by the cab 24 hours ago?

The appearance of bearer 1 suggests G.INP has indeed been turned on.
While G.INP retransmission is the primary error correction method applied, they also turn on a small amount of interleaving and FEC. Unlike old-style DLM intervention, this interleaving adds about 0.2ms of delay, rather than 8-16ms. The FEC uses less than 5% of bandwidth, rather than 20%.
Quote from: Thetwohoughs
Massive rain and wind storm just hit.. this has shown up on the stats  Huh Does this indicate a line issue?

The initial increase in SNRM indicates a reduction in noise, which isn't a normal issue from a storm. It could be a reduction in crosstalk because someone else has turned off their modem.
The downward glitch near the end may represent a resync. If so, it could indicate a line issue, but it is hard to tell from that alone. Graphs of the error rate (of all the error varieties) would probably help.
Quote from: ejs
I think that kind of dip on the HLog graph shows a "bridged tap", and the tone at the bottom of the dip can be used to work out the length of the bridged tap, which I think works out as very short, maybe about 5m.

I agree, though I think I make it about 4m. My source on bridge tap lengths comes from JDSU: http://www.viavisolutions.com/sites/default/files/technical-library-files/sctpsbridgedtap_an_tfs_tm_...
Quote from: Thetwohoughs
There is an extension that's connected via the Front Plate on the master socket, this runs through the attic then back down to supply the house phone. I've disconnected it before with no noticeable change to the connection stats.
The Openreach Modem is fed via the Filter on the MK3 Faceplate. It's also worth noting that the Engineer had his tester plugged into the test port of the NTE5A and could only get a max upsync of around 8000 - 9000 with the cab, does this rule out internal wiring?

It very likely does rule out the internal wiring. If you want to be sure, disconnect the faceplate, and sync the modem with it connected to the test socket. If the Hlog graph is the same shape, then the problem is outside.
Was that engineer someone sent to install your FTTC connection? Or to fix it? I'm surprised that his tester wasn't able to see the tap: it ought to be visible in their own Hlog graphs, and appear as a blip in the TDR reflections.
Plusnet Customer
Using FTTC since 2011. Currently on 80/20 Unlimited Fibre Extra.
Thetwohoughs
Grafter
Posts: 55
Registered: ‎05-12-2015

Re: New Fibre Install - Great download, Poor upload speeds. - Please advise

Top man, thanks very much for that detailed explanation, it'll take me a while to fully digest!
I've attached a Hlog showing the router resynced after unplugging the extension and using a filter from the test socket.
I see very little difference, The 'Bridge Tap' is a strange one, as from where I'm sat viewing the phone line back to the pole, there appears to be no foreign wiring, where is this faulty piece of line likely to be?
The drop wire literally goes from the master socket, runs 30cm along the window sill, goes out through the side of the window frame, is clipped to the brickwork for 1-2 metres then goes straight the the pole on the roadside where it joins the main 50 pair feed. I wonder if openreach could connect me up to a new line back to the cab if one's available?
The Openreach engineer who came out to look at the upload issue had no graphs to view, his box just told him upstream and downstream sync values, plus power statistics from the line. He seemed to have very little knowledge of FTTC and left me rather disappointed. The original installation was carried out by Kellys.
Plusnet have offered me a second visit for an Engineer, but I'm leaning towards leaving this until the BT men have been out to replace the tree chaffed cable?
Cheers for all the help. It's greatly appreciated.
WWWombat
Grafter
Posts: 1,412
Thanks: 4
Registered: ‎29-01-2009

Re: New Fibre Install - Great download, Poor upload speeds. - Please advise

Agreed - the Hlog is unchanged, so the problem is on the Openreach side.
Quote from: Thetwohoughs
The 'Bridge Tap' is a strange one, as from where I'm sat viewing the phone line back to the pole, there appears to be no foreign wiring, where is this faulty piece of line likely to be?

In one I diagnosed recently, the fault was in the wiring inside the DP, where the second pair in the drop wire was connected in the circuit with the first pair. The drop was about 60m - which pretty much matched the predicted length from the Hlog.
Unfortunately, the Hlog itself can only tell you the length of the extra cable - the length determines the delay with which reflected signals come back and interfere with the real signal. In that recent case, an engineer worked out *where* the fault was by using a TDR meter - a time-domain reflectometer. Some blips on that graph let the engineer know how far away any significant splits in the cable occur, and it can usually identify different joints.
In your case, 4m could be anywhere.
Quote
Plusnet have offered me a second visit for an Engineer, but I'm leaning towards leaving this until the BT men have been out to replace the tree chaffed cable?

On balance, it is probably worth letting BT fix their cable, and keep Plusnet's option under your sleeve. But you somehow need to keep some pressure on them. Getting Plusnet to identify any job numbers and keep following up.
Plusnet Customer
Using FTTC since 2011. Currently on 80/20 Unlimited Fibre Extra.
w23
Pro
Posts: 6,347
Thanks: 96
Fixes: 4
Registered: ‎08-01-2008

Re: New Fibre Install - Great download, Poor upload speeds. - Please advise

Quote from: WWWombat
In your case, 4m could be anywhere.
Quote
4m is the kind of length you might expect for an internal extension rather than something connected at the cabinet or any junction on the D-side (though those can't be completely ruled out), does the external BT line come straight into the master socket or could there be an old junction box somewhere?
Call me 'w23'
At any given moment in the universe many things happen. Coincidence is a matter of how close these events are in space, time and relationship.
Opinions expressed in forum posts are those of the poster, others may have different views.
Thetwohoughs
Grafter
Posts: 55
Registered: ‎05-12-2015

Re: New Fibre Install - Great download, Poor upload speeds. - Please advise

Quote from: WWWombat
Agreed - the Hlog is unchanged, so the problem is on the Openreach side.

Thanks wombat, I actually found that thread the other day where the orange and white from the dropwire had been accidentally joined to the black and yellow in a joint somewhere, super interesting stuff  Cool
I'll ensure that the Openreach engineer who comes out next brings with him a TDR meter, hopefully this guy will be more clued in than the first was.
I've reviewed the test that Plusnet did from there and, and that shows as 'No Bridge Tap detected' does this mean their test is flawed?
Quote from: w23

4m is the kind of length you might expect for an internal extension

I've had the ONLY extension from the master socket unplugged from the facia plate of the NTE5A, followed by the modem powered off at the mains then plugged into the test socket,after powering the modem back on the blip is still there in the graph. This completely rules out any internal wiring?
The one place that stands out for me is where my drop wire joins the pole on the roadside, there is a length of wire that runs half way down the pole to a sealed pipe shaped black joint, I wonder if the Bridge tap exists somewhere in this section.
There are no other boxes or strange wiring loops elsewhere unfortunately, I wish this had been a simple fix caused by some internal stuff  Undecided
Thetwohoughs
Grafter
Posts: 55
Registered: ‎05-12-2015

Re: New Fibre Install - Great download, Poor upload speeds. - Please advise

I'm still no further ahead with this, Plusnet offered a second engineer, but are concerned that If the first engineer was unable to solve the issue, then the second one will have the same problem. Should a pair test have picked up the line tap shown in the H Log graph?
The tree damaged cable still hasn't been replaced, but my router has been synced with the cabinet for 22 days now, so I can't see the tree damage having affected my Phone/BB service.