Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[burakkucat]

<Cough!> (Another fur-ball.) "30 mbit" (30 millibit)? The pedantic b*cat suggests 30 Mbps . . .

Chucks mega bits of cold fish at the kitteh.  Does that help clear the furball? 

The grumpy old kitteh pukes . . .    . . . "I don't like fish."
However a quaff of beer will be beneficial.

[burakkucat]

I really can't be sure of what I'm seeing in the QLN graph. 

If l*lion would please post the raw data (from Telnet Data ---> QLN), I'll take a closer look and then re-plot the graph . . .

[kitz]

Chucks mega bits of cold fish at the kitteh.  Does that help clear the furball? 

The grumpy old kitteh pukes . . .    . . . "I don't like fish."
However a quaff of beer will be beneficial. 

Yeah but I was attempting to be funny and so couldnt say chuck mega bits of any type of liquid.   
I shall pour some cream shall I?

[Bald_Eagle1]

You are connected to a Huawei DSLAM, as can be seen from these Discovery Phase band plans:-

Discovery Phase (Initial) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3971)


That data also confirms that the modem's firmware hasn't been updated yet.

The tones will change slightly when/if it does get updated to these:-

Discovery Phase (Initial) Band Plan
US: (7,32) (871,1205) (1972,2782)
DS: (33,859) (1216,1961) (2793,3970)

The firmware updates does seem to help with longer/noisier connections, providing slightly improved speed and/or connection stability.


Attenuation is too high to make any use of the highest frequency tones though.

DS SNRM suggests that a little more speed should be available, but Interleaving depth is quite high at 1273, suggesting a slightly error prone connection.

Impulse Noise Protection (INP) is quite high at 8.00 - it usually starts at zero & then 3.00 when Interleaving is applied at the lower levels.

delay is also high at 16.00. That usually starts at 8.00 when Interleaving is applied.


It does appear that DS sync speed has been capped at 30Mbps:-

Downstream rate = 29996 Kbps

This MIGHT be a temporary measure, following frequent resyncs or external engineering works.


The best thing to do would be leave it connected for a few days, ideally monitoring for changes in the stats, either via DSLStats for 'live' viewing or HG612 Modem Stats for longer term monitoring.

Your speeds COULD be restricted by the effects of crosstalk.

How far are you from the cabinet?


I'm between 1000 & 1100m from my cabinet & you can see my stats in the attachments.

[loonylion]

QLN is attached

Distance I'm not sure, I believe around 650M assuming it follows the road. I should be on the HW firmware listed in the sticky on this board, it says V100R001C01B028SP10 for version.

error counts:

CRC errors   3931    0
FEC errors   37    44
HEC errors   987    0

PPP uptime 8 hours 18 minutes

[burakkucat]

I've downloaded the QLN data but it will now have to wait until later, as I am on my way out for the day . . . 

[loonylion]

To update, SNR margin isnt really changing (might be going up/down by 0.1 periodically), speed is rock solid at 30Mbps (happy with my units this time, B*cat? ) but error counts reported by DSLStats look rather nasty.

      Per second   Per minute   Per hour     Per day

CRC   Up   0      0.10      6.02      144   
   Down   0.01      0.58      34.7      833   

FEC   Up   0      0      0      0   
   Down   10.3      619      37170   892079   

HEC    Up   0      0      0      0   
   Down   0.06      3.63      218      5226   

ES   Up   0      0.10      5.74      138   
   Down   0      0.01      0.45      10.9   


figures are uptime averages.

[burakkucat]

(happy with my units this time, B*cat? )

Purrfect.

I've now looked through the raw QLN data and re-plotted the graph, attached below. After a spot of fiddle-faddling, I haven't been able to deduce all that much . . . other than that you are affected by Radio 4 at 198 kHz, Radio 5 at 693 kHz and Radio Wales at 882 kHz.

Considering the transmitter sites map, below, are you located somewhere equidistant to the Radio 4 & 5 broadcast origin at Droitwich and one of the Radio Wales broadcasting sites? Perhaps somewhere on the arc north of Droitwich and just east of the Welsh border?

[loonylion]

I'm between Stroud and Cirencester, and on a hilltop, so you might be on to something, B*cat. If that is the problem (or a contributor), what, if anything, can I do about it? My internal wiring is cat-5 ftp (although the shield is floating at both ends) and I'm using an ADSLNation faceplate.

[burakkucat]

Ah, so I was looking at the wrong segment of the arc! 

To be honest, there is probably not a lot that can be done to minimise that RF ingress. Due to your elevated location and the spatial arrangement in relationship to those broadcast transmitting sites you will always be prone to such interference. Obviously a well balanced pair will reject any common-mode interference. So really the best you can do is to site the modem (or modem/router) right by the NTE5/A and convert that low power (wanted) differential-mode xDSL signal into an Ethernet signal "as soon as possible".

One thing that you could try is to fit a BT80-RF3 into the circuit, immediately before the NTE5/A. Another experiment would be to ground the shielding of the FTP CAT5e cable.

[loonylion]

Will definitely look into earthing the internal wiring then. modem is right by the master socket and connected to it with a cat-5 RJ-11 cable (that was really hard to find ) I saw mention of earthing the unused pair in the wire from the pole somewhere on this forum, would that likely help at all? (considering that wire is probably where the interference gets in; pole to cabinet is underground) Will also look at acquiring an RF3.

[burakkucat]

Will definitely look into earthing the internal wiring then.

Don't do anything permanent and irreversible. You need to perform the experiment, then decide. 

I saw mention of earthing the unused pair in the wire from the pole somewhere on this forum, would that likely help at all?

Like you, I have read les-70's report of his experiment. However I have no personal experience of the technique. 

Will also look at acquiring an RF3.

And, perhaps, a SSFP Mk 2 as well? 

[loonylion]

Will definitely look into earthing the internal wiring then.

Don't do anything permanent and irreversible. You need to perform the experiment, then decide. 

Of course, ATM I just need to figure out where the closest usable earth point is.

Will also look at acquiring an RF3.

And, perhaps, a SSFP Mk 2 as well? 

Would that be better than the ADSLNation plate I currently have? openreach fitted their own faceplate (not sure if it was a mk1 or 2) after breaking the NT-5 trying to get the ADSLNation plate off by brute force (stripped screw on one side) but after the engineer had gone dad put the ADSLNation plate back on because he paid for it

EDIT:

new graph, this looks bad.



so is the drop from 40ish Mbit/s in November (when the line was provisioned) to 35ish Mbit/s in December/Jan and then further as shown in my graphs on the first page unrecoverable?

[kitz]

Whilst the FECs look nasty, it's far more important to look at CRCs and err secs.  It's hard to judge just looking at fecs, as it's possible that interleaving & error correction could just be doing its job as it's meant to.

As regards to overall loss if syc, unfortunately it's a fact that just about all if us here on fttc has seen some sort of deterioration that is attributable to x-talk.    BE has already mentioned how it's affected his line, and over about 8 months I lost near 20Mbps from my headline rate.  Crosstalk seems to have impacted some lines far more than anticipated, and there's many who can't wait for BT to introduce vectoring.   

I've no doubt that you will have lost some speed as your DSLAM starts to fill, but there does look like something else is perhaps going on with your line.  Bcat is far better than me when it comes to identifying the effects of radio interference, so I will leave you in his capable paws and his opinions in respect of that particular area.

[loonylion]

Thanks Kitz.

CRCs/min and error seconds arent zero but they arent high either, there were 4 downstream error seconds some hours ago but otherwise zero, and upstream CRCs are frequent (every 5-20 minutes) and come in ones and twos. Theres a peak of 6 downstream errors (CRC) about an hour ago, but otherwise zero.