Kitz Forum

Broadband Related => ADSL Issues => Topic started by: Weaver on December 06, 2020, 07:23:00 AM

Title: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 06, 2020, 07:23:00 AM
Last weekend a really clued-up engineer came to visit me regarding the aforementioned long-standing sequence of faults (https://forum.kitz.co.uk/index.php/topic,24614.0.html) on lines 2 and 4.
See: https://forum.kitz.co.uk/index.php/topic,24614.msg426258.html#msg426258 (https://forum.kitz.co.uk/index.php/topic,24614.msg426258.html#msg426258).

Just before he left, he fitted a MkIV SSFP instead of the plain ‘telephony’ front that I had before. He also plugged in a nasty DSL lead into my modem because the SSFP has the usual RJ45 socket in it for DSL and in the existing set-up I had a BT-plug-to-RJ11 custom cable plugged into my modem, and this would not fit into the SSFP. I had no need for an SSFP and he must have been confused.

Anyway, he swapped out my cable and the plain plain ‘telephony’ front that I had before. A week went by and I asked janet to unplug the MkIV SSFP and put things back the way before, because I didn’t need an SSFP, didn’t like the poor quality cable and it was also too long. The moment the swap-over had been completed - shock my downstream speed dropped by 250k which was very suspicious. Closer investigation revealed- horror! The hollow SNR-vs-tones fault had returned !

So it seems we have a partial answer: either the blank BT-socket telephony front plate has been burned by the lightning strike of last February, or else my BT-plug-to-RJ11 cable has.

So I need to replace both, and likewise on the other lines too.

Looking at the top of the line 4 NTE, it reads ‘Master Socket 5C’; I didn’t realise I had an NTE5C on that line. The engineer did not swap out the NTE5 while he was here last week; it was done some while ago and I have forgotten. I know because the NTE5 right now post visit has Janet’s label on it saying ‘line 4’ on the top surface, that is on the back part not the SSFP or other removable front part, so that proves it’s the old NTE5 still present.

When AA staff did their customary initial check procedure, they got me to swap modems over to check for faulty modems, which I did, and thus ruled out a bad modem. This did not however rule out the faceplate front or modem-to-faceplate cable, but what can possibly go wrong with that, so who needs to be concerned? AA was also being considerate because I cannot swap kit over in these sort of tests on my own without assistance from my wife, so AA’s kind understanding plus trust in my being clueful (sometimes) means that AA staff do not pester me with checks every time, but rather just trust me. Normally users have NTE-front-lower-halves that are removed in order to get access to the test socket and these are temporarily removed in order to rule out duff filters and also disconnect all extension wiring. So anyway, who would swap out a plain straight-through front-plate as in my case?

BT engineers always remove the front lower half and go straight into the test socket so that explains why they cannot see the problems.

I really need to have all the NTE5s replaced if they have not already been checked ?



Any thoughts?

I can’t see without a magnifying lens and a bright light, but I’m wondering if some darkening is visible on the connectors.



In the following image below, line 4 is on the left, line 2 on the right:

(https://i.postimg.cc/1znZ3XXG/29-FF28-D0-4-E97-4-A2-C-97-FF-0880-EBFF937-D.jpg)

Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 06, 2020, 08:05:25 AM
On an unrelated note, downstream speeds have iirc never been higher in all recorded history with 10.8 Mbps downstream and upstream not tooo bad either with a reasonable 1.30 Mbps, both from speedtest2.aa.net.uk. The https://testmy.net/upload (https://testmy.net/upload) speedtester reports 1.394 Mbps upstream which is good but not outstanding - I should be getting 1.55 Mbps.

Sync rates:
  #1: down 3122 kbps, up 528 kbps
  #2: down 3066 kbps, up 576 kbps
  #3: down 3096 kbps, up 396 kbps
  #4: down 3184 kbps, up 566 kbps

Egress speeds down, as set by A&A:
  #1 3.11 Mbps down,
  #2 3.05 Mbps down,
  #3 3.08 Mbps down
  #4 3.17 Mbps down
Total combined downstream IP rate: 12.041 Mbps; ‘efficiency’ downstream =  89.69% (but see warning below)

Egress speeds upstream, as set by my Firebrick currently:
  #1: 443632 bps
  #2: 483962 bps
  #3: 332724 bps
  #4: 478920 bps
Total combined upstream IP rate: 1.739238 Mbps; I make that an efficiency upstream of 80.15% (but see warning below) adjusted for TCP gives ~ 92.1% down and ~ 82.4% up but those adjustments are only valid if there are no TCP timestamps in use and IPv4 is in use, not IPv6, and also the speed testers have not already made these corrections already. (I know the testmy.net upstream test has not applied upstream rate correction to allow for IP and TCP headers because it quotes a figure for bytes transferred per second and test data length.)

What do you think about those efficiency figures: 89.69% down, 80.15% up ? (those are not comparing like with like, that is IP and TCP headers need to be subtracted to get a TCP-payload (presumed) from the speed testers to compare with an IP PDU rate, unless the speed testers have already done this. I can subtract either 2.667% or 3.467% assuming IPv4 with TCP without or with TCP timestamps.

Allowing for TCP and IPv4 headers with no timestamps I get 92.1 % downstream and 82.35% upstream. If IPv6 +TCP without TCP timestamps is assumed then add about 1.6% to the above numbers; and add even more still if timestamps are being used.



Looking back at the image above (https://i.postimg.cc/1znZ3XXG/29-FF28-D0-4-E97-4-A2-C-97-FF-0880-EBFF937-D.jpg), I realise that it is a pain that the original installing engineer placed the NTE5 for line 2 right next to that for line 4 because now it is a pain unclipping the modern screwless front plates as you can’t get to one of the sides to depress the [?] thingy ??? to release the modern front plate.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: tubaman on December 06, 2020, 09:05:56 AM
Have you tried removing the MK4 SSFP and plugging your own modem lead into the test socket. That should narrow down if the socket or your lead is the culprit.
Also, it is of course possible that the SSFP is doing something as it has a common mode rejection filter in there.
 :)
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 06, 2020, 09:35:29 AM
No, the SSFP isn’t doing anything as when I swapped the SSFP out the badness of the SNR-vs-tones curve promptly appeared, so the non-SSFP condition is broken and the MkIV filter isn’t doing anything ‘good’. Unless I’m missing something.

I think I have the correct fitting cable in my box of tricks, which my wife has for some strange reason named ‘Bert’s Barrow’, and will do as you suggest.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: burakkucat on December 06, 2020, 04:18:26 PM
Interesting developments.  :)

Not something that would be obvious, unless one was physically present and able to inspect at first-hand (or first-paw).

As for what has happened, I haven't a clue.  ???
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 06, 2020, 04:38:56 PM
The moment I went back to my existing ‘front plate’ and cable, the modest characteristic hollow appeared immediately, and the new MkIV SSFP and cheap cable the shape of the graph is perfect. Highest synch rate ever seen downstream, as best I can remember anyway.

Do you think I’m on the right track ?

By the way, I have not seen the engineer’s notes - I’ll ask A&A for them.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: burakkucat on December 06, 2020, 04:49:23 PM
Do you think I’m on the right track ?

An experiment has been performed where two variables were changed at the same time. As one scientist to another, I ask . . . (well, you decide what I ask!  ;)  ) 

For your use case, you have no need for anything telephony related . . . and that include the series connected capacitive/resistive shunt across the pair (which is designed to provide a "finger-print" to the exchange equipment. (Most likely a System X remote concentrator.))

If I was in an identical situation, with four data-only lines, I would be very tempted to remove all of the NTE5s and substitute an appropriate box, connection, with Krone strips. The four incoming lines would be terminated on a Krone strip and then four local cables, with plugs 6P2C (pins 3 & 4), would go directly to the modems.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 06, 2020, 04:56:24 PM
Indeed, I have changed two variables. But I have to recruit my beloved who has been out feeding donkeys before the light fades and she will be able to find me a suitable cable. If all else fails then I can swap parts with line 2 which also has a BT-to-RJ11 custom cable on it, by that photo that Janet took.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 07, 2020, 02:56:58 PM
And guess what, line 2 has now gone bad, again, in spectacular fashion. I initially put this down to the cable or front plate evil syndrome seen on line 4. However, as luck would have it, I mentioned this to AA, and they came back telling me they had seen a battery contact fault on that line. So an engineer is being dispatched.

I have ordered a good supply of custom BT431-to-RJ11 custom cables and some NTE5C mk2 units too just to get the ‘telephony’ lower half fronts. I am trying to get ultra short 0.5m and 1m long cables made out of CW1308 or failing that 4mm BT Cat5.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: burakkucat on December 07, 2020, 11:27:59 PM
I am trying to get ultra short 0.5m and 1m long cables made out of CW1308 or failing that 4mm BT Cat5.

Your patch cables (BS431A ("BT") plug to 6P2C plug) should be constructed with twisted-pair, stranded-core, wires and not solid-core wires, that CW1308 possesses.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 08, 2020, 12:49:02 AM
Ah, I got this wrong, again,  :-[  I remember now your advice on the disadvantages of solid core. I should have asked you before ordering them. At least they are only £4 and hand-made to be a custom spec. The alternative was a BT cat 5 cable so I am told. I ordered from the same custom maker who supplied me last time.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: burakkucat on December 08, 2020, 01:06:15 AM
As long as they are installed and then left alone, not flexed nor repeatedly moved, then you probably will not experience any problems. 
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 08, 2020, 01:52:14 AM
Can you think of a part number, or whatever the correct term is, for the correct stranded type?

Once plugged in, the ultra-short cables to the modems are never molested, because the modems are never moved, so there is more of a chance we will get away with it. My stupid non-existent memory.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: burakkucat on December 08, 2020, 06:40:11 PM
Can you think of a part number, or whatever the correct term is, for the correct stranded type?

Nothing comes to mind.  :-[

The pair of wires within the cable should be twisted and the wire gauge should be appropriate. Consider a Cat5e patch cable. Cut off the two plugs. Slit the entire length of the sheath. Lift out four twisted pairs of wires. They would be appropriate for the task.

Quote
Once plugged in, the ultra-short cables to the modems are never molested, because the modems are never moved, so there is more of a chance we will get away with it. My stupid non-existent memory.

Hmm . . . I've just had a thought. Your detection device indicates an imminent atmospheric electrical discharge. Mrs Weaver rushes into the office to disconnect the modems.  :-\
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 08, 2020, 11:33:22 PM
No that’s true, I forgot about that.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: neil on December 09, 2020, 08:49:24 PM
Interesting developments.  :)

(or first-paw).


🐈
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: siofjofj on December 10, 2020, 08:47:57 AM
Can you think of a part number, or whatever the correct term is, for the correct stranded type?
I use  these (https://www.ebay.co.uk/itm/0-5m-ADSL-2-High-Speed-Broadband-Modem-Cable-RJ11-to-RJ11-50cm-White/264359271605) cables which appear to be twisted pair (certainly not flat anyway) and are more flexible than the solid core CW1308 so I assume are stranded. The RJ11 plugs have strain relief boots fitted too which is nice. You could perhaps use one with an adapter such as  this (https://www.ebay.co.uk/itm/RJ11-to-BT-Plug-Adaptor-Connect-ADSL-DSL-Cable-to-BT-Telephone-Phone-Socket/351044909620).
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 10, 2020, 01:06:32 PM
Those kenable ones you quoted look very like the Tandy/ADSL Nation ones that I have used for a long time. I prefer RJ45-RJ11 custom ones though as the RJ45s are a very slightly better fit.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: burakkucat on December 10, 2020, 10:19:32 PM
I thought you were looking for leads with BT431A and 6P2C (or 6P4C) plugs. Something like these (https://cpc.farnell.com/pro-signal/31034br/bt-plug-to-rj11-1m-black/dp/TE05823)?
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on December 11, 2020, 01:23:07 AM
Perfect, thanks. And nice and short.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on January 02, 2021, 06:25:31 PM
Have swapped out all the modem-to-wall socket cables the ‘telephony’ front plates on lines 2, 3, 4 for new ones. Line 1 has an NTE5a and an Andrews and Arnold straight-through front faceplate. Disappointment: there was a ‘hollow’ failure on 2020-12-25 and later on I fixed it, by some miraculous intervention, when I swapped modems’ network connections, the problem went away. So the burned out faceplate or cable theory seems discredited as a only new kit was in use.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: j0hn on January 02, 2021, 11:04:14 PM
Do you have a full compliment of 4 x A&A data only rj45 faceplates?
I'm sure I still have 1 lying around that I don't use any more.

I couldn't find any online last time I looked.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on January 03, 2021, 02:38:36 AM
I should have three somewhere - do you need one ?
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: j0hn on January 04, 2021, 11:29:58 AM
I should have three somewhere - do you need one ?

The opposite.
I was going to offer you mine.

I can't see me using xdsl again unless we needed to move home soon.
Title: Re: Odd-shaped SNR-vs-tones graph - towards an answer!
Post by: Weaver on January 04, 2021, 06:41:00 PM
Yes please J0hn.

Have pmed you.