Print

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

[sheddyian]

For info/discussion :

A while ago, I replaced the section of phone line that carries the ADSL signal from "normal" phone cable to CAT5.  I had an old CAT5 patch cable that had solid cores, not stranded as modern ones tend to be, so I cut the plugs off that and punched it into the IDC

This gave me a very slight improvement - my line attenuation, previously 11.4 up dropped consistently to 11.3 UP, though the DOWN stayed the same at 22.

More recently, I was wondering about screening out interference.  I've no specific problems with interference, I was just curious if I could achieve any further speed gains on the line by making it "quieter".

I don't have any screened CAT5 or CAT6, and as the CAT5 was now tacked to the wall, I didn't want to rip it all out again to try something else.

I then thought about the spare pairs in the CAT5 cable - after all, I'm only using 1 pair.  The others were sitting there doing nothing.  Would they make a useful "screen"? So firstly, I commoned all the unused pairs together at both ends.

At the socket end, I then connected these commoned unused pairs to a single wire, which I fed out into the garden immediately outside, and attached to a screwdriver that I shoved into the ground as a makeshift earth.

It's been like that for some time, and I can't say I could detect any measurable difference.  Anyway, today, an electrician friend gave me a proper earth rod, about a metre long, so I thought I'd replace the screwdriver "earth" with something real, and see if it made any difference.  The answer : probably not!

Here are the results anyway, for what it's worth


                          Synch    Attenuation    S/N Margin
                         UP   DOWN   UP   DOWN     UP  DOWN
Screwdriver earth       1020 20933  11.3   22      9.3    6
No earth                1020 20709  11.3   22      9.7    6.1
Proper Earth            1020 20799  11.3   22      9.3    6
Proper Earth            1020 20917  11.3   22      9.7    6


(The 2nd "Proper Earth" is just the same setup as the first "Proper earth", but following another reboot, when nothing else had been changed)

I think all this really shows is that there is a bit of fluctuation, which could be mistaken for improvements/impairments following fiddling

Ian

[burakkucat]

I can't say that any of the results are significant. However I like your experimental style. 

I think Walter may be interested to see your null results because, quite some time ago, he was pondering if earthing the steel reinforcing catenary wires in a standard Openreach drop-cable would provide some degree of screening. (My feeling is no.)

[kitz]

Interesting post - thanks  

The results would seem to indicate (the 2nd proper earth) that the fluctuations seen are just 'normal variance', but hey some of us like to play and experiment just to see if we can get that little bit more and sharing experiences that may be helpful to others.

I'll add another 'experiment' result...  I have a CAT5 extension of appx 20m.  Whats possibly unusual though is that the other pairs have been stripped out.

(as can be seen in this pic)


I cant take credit for this because it wasnt my hand that got sore after pulling out the the additional pairs.. and TBH it was done more to make the cable a bit more flexible to go around quite a few 90 degree bends...  but the result was exactly the same before and after stripping out the wires.
In fact the whole run of CAT5 doesnt make any different at all. Plugging the router at the master socket and/or at the end of the CAT5 extension run gives the same readings.

[burakkucat]

Plugging the router at the master socket and/or at the end of the CAT5 extension run gives the same readings.

Which is good, exactly as one would wish! 

Edited to add: I've just noticed something "odd" in your picture. The two cores have been fed through the post which is designed to take a cable tie, to fasten the cable into place. 

[kitz]

>> Which is good, exactly as one would wish!

Yes sorry, should have added a happy smiley face at the end of my sentence

>> The two cores have been fed through the post which is designed to take a cable tie, to fasten the cable into place.

lol it works..  the 2 cores threaded through the post nicely since I didnt have a plastic tie to hand 

[roseway]

I've got my adsl router connected to a socket on the end of about 10 metres of Cat5e connected to the A/B terminals of a filtered faceplate. Previously the router was connected directly to the faceplate. Just the same as above, the length of Cat5e made no discernible difference.

[sheddyian]

I can't say that any of the results are significant. However I like your experimental style. 

Thanks!

I think Walter may be interested to see your null results because, quite some time ago, he was pondering if earthing the steel reinforcing catenary wires in a standard Openreach drop-cable would provide some degree of screening. (My feeling is no.)

I can see the logic in that, a similar thing to what I was trying to achieve with my spare-pairs earthing scheme.  Maybe I saw no difference because I'm not suffering from interference.  Maybe it just didn't work!  I'm watching the errors to see if they are in any way affected by my improvised screen & earth.

The results would seem to indicate (the 2nd proper earth) that the fluctuations seen are just 'normal variance', but hey some of us like to play and experiment just to see if we can get that little bit more and sharing experiences that may be helpful to others.

Early on with my tinkering, I was getting excited/frustrated when I'd adjust or change something, reboot and find I'd gained an extra 100k speed, only for it to disappear again or get worse later on at the next reboot!  Later realised it was, as you say, "normal variance" 

I've played around with ferrites, with different RJ11 modem leads (I did get a small improvement there that was repeatable!) different filters etc.  So, yes, I like to play and experiment to see if I can get that little bit more

Ian

[sheddyian]

I mentioned above that I'd tried ferrites and different RJ11 cables, might as well give a bit of detail about that in case anyone is interested 

I took apart a dead PC PSU, cut the ferrites out (sort of polo mint shaped metal-like things with copper wire wound round), cut the copper wire off, and tried looping various ADSL related cables through them to see if any numbers changed.  Short answer : no.  Longer answer : I'd read that it could make things worse, the theory being that it ends up filtering out frequencies you want.  I saw neither improvement nor impairment with ferrites over the incoming phone line, on the RJ11 lead, even on the RJ45 patch cable (idly wondered about interference!)

What I possibly saw was a slight improvement in line speed with additional ferrites on the ADSL modem power supply lead.  I noticed that the little switch mode PSU emitted a lot of MW radio noise for a short distance, so considered that extra ferrites might improve things.  It may have done, but I'm uncertain.

What I do intend to try is running the ADSL modem off a battery, then there'll be no mains-bourne or PSU noise getting so near.  I need to get the right sized power connector, then I thought I'd connect it to a 12volt  portable drill battery and see if there's any benefit.

Of course, if there is significant benefit, I've got to work out how I'd make that setup usably permanent!

RJ11 leads : I noticed the lead supplied with the modem was very thin indeed.  It had 28AWG stamped on it.  I found a 26AWG lead from a dead modem, and tried that.  Both leads are flat 2 core, very basic.  The thinner 28AWG lead is just under a metre long, the 26AWG is a bit longer, but no more than 2 metres.

The tests were done in order shown here, from top to bottom

           
                         S/N Margin (UP)
Original 28AWG lead        8.2
26 AWG lead                9.9
Back to 28AWG              8.1
26AWG                     10.1



I was quite impressed it seemed to make a repeatable difference, although I can't say it did anything for my actual synch speeds.

I later tried a fancy screened RJ11 cable from ADSL nation, and found gave me a lower S/N margin than the plain 26AWG lead, so I''ve stuck with that.

Ian

[burakkucat]

What I do intend to try is running the ADSL modem off a battery, then there'll be no mains-bourne or PSU noise getting so near.  I need to get the right sized power connector, then I thought I'd connect it to a 12volt  portable drill battery and see if there's any benefit.

That would be a good experiment to perform. I have suggested using a lead-acid car battery in the past, as the idea of using a portable power tool's battery had never occurred to me. 

[les-70]

  I tried some similar tests a long while ago.  The tests are much easier to assess if you can log and graph the SNR with DMT or RouterStats.  With those you can add and remove the influence a few times with no resync and be sure to see if there is a real effect.  I gained about 0.2db SNR by earthing other conductors. You may get error rate differences but they are very hard to be sure of as they are so variable any way.  I will try your your power supply suggestion. I currently have a few ferrites on the ADSL2 lead, they make a barely detectable drop in SNR, less than or about 0.1 (would be a lot more with VDSL I guess) but I think that they do reduce the error rate a little.

    Further to this post,  with ferrite on the power supply leads I find no effect on SNR but an indication that the error rate may have halved.  I don't trust a bank holiday Sunday to have normal noise so I will  wait for a proper weekday to confirm that.

[burakkucat]

I don't trust a bank holiday Sunday to have normal noise so I will  wait for a proper weekday to confirm that.

b*cat nods in agreement. 

[les-70]

 Further my last post I have had the ferrite on and off the power supply lead (the 12v one) over the weekend and also this morning (a normal looking weekday).  I think the ferrite on the power supply has given a small reduction in noise.  The number of CRC per 15mins seems to drop from about 7 to 4 most of time ( I run at 3db sync).  However once a while there is burst to about 20-30/15mins. These burst don't seem to have changed with the ferrite, but a longer test period would be needed to be sure.  The average impact on total CRC has been about 10-20% better over successive of 2 hours on and off.   I conclude that it is a modest but good thing and I can't see how it would give any bad effects. I imagine that other people would find varying impacts depending on power supplies characteristics and sources of noise that through the power line.

[sheddyian]

That would be a good experiment to perform. I have suggested using a lead-acid car battery in the past, as the idea of using a portable power tool's battery had never occurred to me. 

This plan is thwarted by the unusual power connector on my DSL-2780.  It looks like a typical DC power connector, but has external diameter of 3.7mm and internal of 1.3mm.  Can't find anywhere to buy just 1 or 2 of such things

(Seem quite hard to find anywhere selling any of them actually, found some surface mount sockets of that size though!)

As the modem is under warranty, I'm not keen on opening it up to solder wires to it, nor to snip the lead on the power supply, experiment with a battery then rejoin it 

Ian

[burakkucat]

Are you absolutely sure of the dimensions? 

Maplin have a range in stock, in particular 1.3mm/3.45mm, 1.7mm/4.0mm, 1.7mm/3.45mm, etc.

The closest I can find to the dimensions you've mentioned is 1.35mm/3.5mm.

[sheddyian]

Are you absolutely sure of the dimensions? 

Maplin have a range in stock, in particular 1.3mm/3.45mm, 1.7mm/4.0mm, 1.7mm/3.45mm, etc.

Well, having just rechecked, I did make a bit of a mistake... but it still seems I have an odd dimension of plug, unless I'm interpreting the readings wrongly.  I don't have a known sized plug to measure and confirm what the tolerances are/exactly which part is being measured, though I have one that seems a "typical" sort of plug

The 2 pics of the black plug are the one in question for my DSL-2780.  The third picture of the cream plug is just an example test, which seems to me to be a very typical 5.5mm plug (old PSU from analogue cordless phone!)

So, I read the dimensions of the black plug to be 3.8mm external, 1.3 or 1.4mm internal.  I think.....   

Ian