CAT5e Cable signal integrity for DSL signal

[victor]

I want to run cat5 cable in my house to replace the current telephone wire. I hope to increase the SNR for my ADSL connection.

So I was thinking in using one pair, and I read somewhere that I should use the brown or orange pair, but I was thinking that leaving the other pairs floating isn't it bad? can I just use the 4 pairs in parallel? Should I just loop every pair back? ground then?

What do you guys think? It is my desperate plan hopeless?

[alm]

Telephone over Cat 5 should be fine. I wouldn't put multiple pairs in parallel, that would screw up the impedance. Resistive loss is not your problem for those few meters of cable inside your house. If you want to do anything with the free pairs, then I would ground them, although I think telephone people will usually leave them floating.

If it's just a few meters, then I don't expect better wiring to make much of a difference. Is the current wiring the twisted telephone wiring that telcos use, or is it the cheap flat cables with parallel wires (the wiring you see here)? The latter is quite terrible for more than a couple of meters, but I don't think replacing a few meters of twisted telepohen cable by Cat 5 is going to make any difference unless the old cable is damaged.

The same argument that applies to expensive audiophile power cables also applies here: a few meters of better cable on your side is generally not going to do a whole lot compared to the kilometers of cable on the telco side. Unless the current wiring is of really poor quality.

[Rufus]

I want to run cat5 cable in my house to replace the current telephone wire. I hope to increase the SNR for my ADSL connection.

Probably a waste of time.

Disconnect all house wiring at the point the line enters your house, connect the modem there and check the modem status. The difference (if there is any) is the most you will get by improving house wiring.

[Stonent]

I want to run cat5 cable in my house to replace the current telephone wire. I hope to increase the SNR for my ADSL connection.

Probably a waste of time.

Disconnect all house wiring at the point the line enters your house, connect the modem there and check the modem status. The difference (if there is any) is the most you will get by improving house wiring.

Yes, if your telephone company has not put a lock on the junction box, this would be your best bet.

I found this picture which illustrates it perfectly, using the DSL would be similar. Sometimes the TNI box is locked externally, sometimes it isn't. If it's not locked outside, they usually have a customer side and phone company side, the phone company side is frequently locked so you can't unscrew the terminals to the incoming line.  Either way, you may have a test jack in the box. Unscrew the terminals for the outgoing line (to the house interior) and try the test jack.

[edavid]

I want to run cat5 cable in my house to replace the current telephone wire. I hope to increase the SNR for my ADSL connection.

1. Definitely try connecting your modem at the demarc first, as per the other suggestions

2. This will only help if your existing wiring is not twisted pair

3. CAT3 is more than good enough, and easier to run

4. If you are currently using microfilters, you will get much more improvement from switching to a splitter at the demarc

[victor]

The wiring is pretty new, but it is not twisted pair, and it goes from the street wiring to the modem outlet, we don't use any phone in the house, other outlets are disconnected from the line, so I don't have any filter/spliter.

The wire is like that, it is much thicker than the cat5, like power cords.

[edavid]

Wow!  I haven't seen that used for inside wiring before.   By all means, try CAT3 or CAT5.

[free_electron]

CAT5 works perfectly fine. Any twisted pair will do.

Even if you don't use a phone : install the line filter at point of entrance. it keeps unwanted spurious out.

if you got bad S/N : check the splices. those stupid crimp connectors sometimes give bad contact with loss of lower band channels.
especially if the wrong punch-down tool has been used or the splice contacts have been used too many times...

[edavid]

Even if you don't use a phone : install the line filter at point of entrance. it keeps unwanted spurious out.

I disagree, you should not be using a filter between the line and the DSL modem.

[Psi]

My parents have 4x  50 pair phone cables running throughout the house connecting every room to a central office.
This cable is pre CAT standards so it probably isn't even CAT1 rated. Dad installed it 40 years back from left over cabling from his job at the airport.

There's a PABX with around 10 phones throughout the house and these all run down multiple pairs along with 1000mbit gige networking to every room and a VDSL phone lines which is routed through a few of the cables to the office.

The gig ethernet doesn't run as fast as it should, around 20MB/s is the most you get, but other than that there's no noticeable problems sharing all those signals on a really old 50 pair cable.

Even if you don't use a phone : install the line filter at point of entrance. it keeps unwanted spurious out.

I disagree, you should not be using a filter between the line and the DSL modem.

He means install the splitter/filter at the demarcation point and run a new cable from there for the modem.
That way it isolates the phone wiring as soon as possible.

[edavid]

The gig ethernet doesn't run as fast as it should, around 20MB/s is the most you get, but other than that there's no noticeable problems sharing all those signals on a really old 50 pair cable.

I don't understand how that's possible.  1000BT doesn't have a fallback mechanism.  Either it works at full speed, or it doesn't.

Even if you don't use a phone : install the line filter at point of entrance. it keeps unwanted spurious out.

I disagree, you should not be using a filter between the line and the DSL modem.

He means install the splitter/filter at the demarcation point and run a new cable from there for the modem.
That way it isolates the phone wiring as soon as possible.

The OP doesn't have any phones, so he should just disconnect the inside wiring.  There's no need for a splitter.

[Psi]

The gig ethernet doesn't run as fast as it should, around 20MB/s is the most you get, but other than that there's no noticeable problems sharing all those signals on a really old 50 pair cable.

I don't understand how that's possible.  1000BT doesn't have a fallback mechanism.  Either it works at full speed, or it doesn't.

i dunno either, but i remember getting a reliably 50MB/s between PCs using a patch cable but only ever 20MB/s through the house RJ45 wall sockets which use that 50 pair cable.
Maybe it was intermittent/cycling packets loss due to crosstalk from signals on other pairs.

[G7PSK]

All the splitter/filters that I have seen are straight through for the ADSL and filter the phone or rather block the ADSL signal from the phone itself. Whenever I have to ring BT about the broadband and due to my location that is rather often, the first thing they ask is how many phone you have is there a micro filter and have you unplugged all the phones, if there are no phones you do not need a micro filter, at one time I had a separate phone line for the broadband as the burglar alarm on the other line prevented the ADSL from working, now that the contract on the alarm system has expired it has been changed and do not need 2 lines but noe need the filters on the phone's.

[victor]

I don't understand how that's possible.  1000BT doesn't have a fallback mechanism.  Either it

TCP check byte integrity using bit checksum, of a packet is corrupted it is re transmitted, so in theory every corrupted byte get at least twice the time. So real speed is less than the bus rate

[madires]

I don't understand how that's possible.  1000BT doesn't have a fallback mechanism.  Either it

TCP check byte integrity using bit checksum, of a packet is corrupted it is re transmitted, so in theory every corrupted byte get at least twice the time. So real speed is less than the bus rate

A standard telecom 50 pair cable could be considered to match Cat3. That's sufficient for 10BaseT. 100BaseT might work for short distances (but out of specs). That's why there are tons of frame errors when trying to run 1000BaseT. And those frame errors decrease the bandwidth.

[dfmischler]

There can't be very many frame errors or the throughput would be a lot worse (the minimum TCP retransmission timer is 1 second, and backs off based on round trip time and repeated retransmissions).

[madires]

Stacked protocols make debugging harder because one could mitigate the problems of the other or worsen it :-) You can create a similar situation by setting one side to 100BasseT full-duplex and the other to 100BaseT half-duplex.

[peter.mitchell]

I think changing to cat5 won't make much difference, the big difference will be re-terminating corroded/oxidized terminations.

[SeanB]

New cable will help if the old one is so old that it predates you or your father. At work the DSL phone cable is as follows. 297m from the central office in the street is a 200 pair paper insulated copper cable with a lead sheath to a DP frame in a building near me. Then it is 35m of 50 pair paper cable to a street manhole, from there 56m of new 50 pair cable ( replacing a 20 pair paper cable that got so bad even a 50 baud telex would not work) to the building DP panel. From there it is 5m of 20 pair phone indoor cable to the next DP, which then has the 30m of CAT5 cable that leads to the server. Connections are done with IDC connections at all joints, those in the street have been made with Scotchloks, as the original 40 year old 3M punch block is very nasty, punch one down and 2 pop off elsewhere in the strip, so the pairs were cut out and scotchloked together by the TELCO guys a long time ago out of frustration. Inside it is a smaller punch panel, either Krone or Quante ( where the tooling to punch down is ultra expensive, so I made my own one out of a small flat screwdriver specially ground to fit the slots to punch down on the 10 strips there) and is under 20 years old. Now reiable noise free phones, reliable DSL and the frame relay works well and reliably. There have been issues with the old paper cable with use of 0+16 line splitters, as these supply 200V to power the electronics at the far end. Thankfully they are all gone now with an exchange upgrade, as they would happily burn the cables if moisture got in there.

[grumpydoc]

I want to run cat5 cable in my house to replace the current telephone wire. I hope to increase the SNR for my ADSL connection.

So I was thinking in using one pair, and I read somewhere that I should use the brown or orange pair, but I was thinking that leaving the other pairs floating isn't it bad? can I just use the 4 pairs in parallel? Should I just loop every pair back? ground then?

What do you guys think? It is my desperate plan hopeless?

You don't say where in the world you are based which might help with possible teleco quirks.

I wouldn't underestimate the effect that even a few metres of untwisted cable can have - especially if there's a lot of electrical noise in the vicinity. 1.5m of untwisted cable lost me about a megabit and a half when I had ADSL (have VDSL now) - quite a big deal when the fastest the line would go was 3Mbs^-1

Otherwise I agree with everyone who has said you probably won't see much improvement swapping decent quality twisted pair telephone wiring for cat5e.

However I don't think signal to noise ratios are your problem - at least not when that snapshot of your line stats was taken and perhaps not in the way you think.

One piece of information which would be useful is how long your "local loop" is - i.e the length of the line from the exchange to your premises.

For a line attenuation of 48dB downstream I'd expect 3.5 to 4.5km (roughly 10-12dB attenuation per km). If a lot shorter than this you have a bad line. If longer a very good one.

Then looking at the SNRM 13.8dB is actually quite high - don't forget this is not the signal to noise ratio but the SNR margin - i.e how much worse the SNR can get and the modem still be able to maintain the current sync speed.

When DSL modems negotiate the sync speed they don't negotiate at the fastest speed possible but slightly lower - this allows a margin for the line noise to get a little worse without needing to re-sync. Most lines vary in their noise levels throughout the day so if this were not done you'd get huge numbers of line re-syncs.

The default margin is 6dB for most telecoms companies on a new line - the system then gradually "learns" where it needs to be set for reliable operation.

Finally note the "max attainable" speed is currently over 8Mbs^-1 despite a sync of 4Mbs^-1

So it looks as though either you are on a fixed 4Mbs^-1  service for some reason or you line has a lot of noise variation through the day. The default SNRM starts at 6dB and is adjusted in 3dB steps in the range 3-15dB, thus a current SNRM of 13.8dB might suggest a 12 or 15dB target for your line which isn't great.

Those who have said you don't need a filter if you don't have any phones connected are correct. The filters connect the DSL modem directly to the line and then provide a low pass connection for the phone - their main function is really to keep DSL signals away from the phone. However if the modem is plugged in close to where the line enters the building and the extension wiring is still present it might be worth putting a filter in to isolate the extension wiring. Or just disconnect it.

You need to do a couple of things. First check you're not on a fixed speed service. Then repost stats immediately following a re-sync (this means you can tell what your default SNRM target is).

 Then I'd get a copy of DMT tool and look at the stats over a few complete 24 hour periods. If the SNRM is falling much lower during some point in the day that might give you a clue. If it's overnight then MW radio pickup is a common culprit and if your local loop has long overhead runs there might not be much you can do about it (isolating even unused extension wiring stops it acting as an antenna which is why it can be helpful).

If the SNRM is worse during the day there might be interference from electrical equipment in use at that time - perhaps in your house, perhaps somewhere close to the line running back to the exchange.

You also need to look at what sort of speeds you get when you use the "test" socket if you have one. For UK BT connections that's the one inside the master socket - using that will isolate all the extension wiring. I'd be slightly careful about interpreting the result given the stats that you posted but if you get a big jump up which is then maintained over at least 24 hours it might well be the extension wiring that is the cause of your low sync speed.

[victor]

I want to run cat5 cable in my house to replace the current telephone wire. I hope to increase the SNR for my ADSL connection.

So I was thinking in using one pair, and I read somewhere that I should use the brown or orange pair, but I was thinking that leaving the other pairs floating isn't it bad? can I just use the 4 pairs in parallel? Should I just loop every pair back? ground then?

What do you guys think? It is my desperate plan hopeless?

You don't say where in the world you are based which might help with possible teleco quirks.

I wouldn't underestimate the effect that even a few metres of untwisted cable can have - especially if there's a lot of electrical noise in the vicinity. 1.5m of untwisted cable lost me about a megabit and a half when I had ADSL (have VDSL now) - quite a big deal when the fastest the line would go was 3Mbs^-1

Otherwise I agree with everyone who has said you probably won't see much improvement swapping decent quality twisted pair telephone wiring for cat5e.

However I don't think signal to noise ratios are your problem - at least not when that snapshot of your line stats was taken and perhaps not in the way you think.

One piece of information which would be useful is how long your "local loop" is - i.e the length of the line from the exchange to your premises.

For a line attenuation of 48dB downstream I'd expect 3.5 to 4.5km (roughly 10-12dB attenuation per km). If a lot shorter than this you have a bad line. If longer a very good one.

Then looking at the SNRM 13.8dB is actually quite high - don't forget this is not the signal to noise ratio but the SNR margin - i.e how much worse the SNR can get and the modem still be able to maintain the current sync speed.

When DSL modems negotiate the sync speed they don't negotiate at the fastest speed possible but slightly lower - this allows a margin for the line noise to get a little worse without needing to re-sync. Most lines vary in their noise levels throughout the day so if this were not done you'd get huge numbers of line re-syncs.

The default margin is 6dB for most telecoms companies on a new line - the system then gradually "learns" where it needs to be set for reliable operation.

Finally note the "max attainable" speed is currently over 8Mbs^-1 despite a sync of 4Mbs^-1

So it looks as though either you are on a fixed 4Mbs^-1  service for some reason or you line has a lot of noise variation through the day. The default SNRM starts at 6dB and is adjusted in 3dB steps in the range 3-15dB, thus a current SNRM of 13.8dB might suggest a 12 or 15dB target for your line which isn't great.

Those who have said you don't need a filter if you don't have any phones connected are correct. The filters connect the DSL modem directly to the line and then provide a low pass connection for the phone - their main function is really to keep DSL signals away from the phone. However if the modem is plugged in close to where the line enters the building and the extension wiring is still present it might be worth putting a filter in to isolate the extension wiring. Or just disconnect it.

You need to do a couple of things. First check you're not on a fixed speed service. Then repost stats immediately following a re-sync (this means you can tell what your default SNRM target is).

 Then I'd get a copy of DMT tool and look at the stats over a few complete 24 hour periods. If the SNRM is falling much lower during some point in the day that might give you a clue. If it's overnight then MW radio pickup is a common culprit and if your local loop has long overhead runs there might not be much you can do about it (isolating even unused extension wiring stops it acting as an antenna which is why it can be helpful).

If the SNRM is worse during the day there might be interference from electrical equipment in use at that time - perhaps in your house, perhaps somewhere close to the line running back to the exchange.

You also need to look at what sort of speeds you get when you use the "test" socket if you have one. For UK BT connections that's the one inside the master socket - using that will isolate all the extension wiring. I'd be slightly careful about interpreting the result given the stats that you posted but if you get a big jump up which is then maintained over at least 24 hours it might well be the extension wiring that is the cause of your low sync speed.

This have a really amount of good information.
Yeah, my contract is for 5mbit they told me is the  greatest speed that I could get in my area.
I'm checking my line status regularly, as you said  the SNR margin varies quite a bit during the day now it is around 17db. The attenuation is always 48db

[grumpydoc]

Yeah, my contract is for 5mbit they told me is the  greatest speed that I could get in my area.
I'm checking my line status regularly, as you said  the SNR margin varies quite a bit during the day now it is around 17db. The attenuation is always 48db

OK,  the question you really need to be asking is "why is my SNRM so high?"

I wouldn't expect the reported line attenuation to change (although sometimes you do get small variations - perhaps a couple of dB either way).

You do need to know whether you are on a fixed speed or rate adaptive connection. Your ISP should be able to tell you but it might be just as fast to force a re-sync when the SNRM is high. At 17dB I would really expect the line to sync faster then 4Mbs^-1 - if it goes even a little bit higher you have to be on a rate adaptive service.

If it sticks at 4192kbs^-1  it's probably a 4 megabit fixed speed. 5 megabits would have been a reasonable estimate for your line, but is perhaps a bit unusual for a fixed speed service (and you're not getting that fast anyway). Obviously whatever you do in that case it isn't going to get any faster. If you can get the ISP to switch you to a rate adaptive service and you might see a better speed (I say might because without knowing how the line noise varies it's not possible to say).

If you are on a rate adaptive connection you really do need to find out why the SNRM is so high - I can only repeat what I said earlier, especially getting at least a complete 24 hours stats with DMT tool (which draws pretty graphs for you to post).

It would still be useful to know your local loop length, and the line stats after a re-sync (especially if the SNRM was as high as 17dB just before).

[Rufus]

OK,  the question you really need to be asking is "why is my SNRM so high?"

The two stats the OP has posted show 4192/573 down/up speed with 4db difference in downstream margin. He is almost certainly on a rate limited service so any cable improvement is just going to up the SNRM not speed.

The max rates in the stats indicate he could probably get 7/8 Mbit down if the service allowed it.

[SeanB]

I have 15dB of margin. and am rate limited to 1Mb/s. Could easily do 6-8Mb, but will wait for the price to drop from Insane to just painful. Paying $20 per month for 1Mb is still the best ( cheapest as it is the base option) and no cellular provider gives a data rate lower than ADSL currently, and no uncapped offerings at all, but if it comes to pass........

Friend just got his bill, $800 overage on a normal $100 bill and that is from all out of rate data, just 100M past the soft cap and no SMS or anything about this either. Do not use the phone for a month and they call, but give them money no way do they tell you. Heard of a guy with a R750 000 bill for a month from data charges while roaming.

[jamesb]

Looks like the attainable rates are fairly good - where exactly are you hoping to improve your speed?
If you are hoping to get anything faster than your current connection rate, or your "maximum / attainable" rate, then you need to contact your Telco to see if that is even possible (based on distance to SLAM & the technology type).