Print

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

[drewk]

Is there a way to increase the range of the SNR-per-tone graph? My SNR values are too high to be plotted on the graph (not a bad problem to have on an ADSL line, I guess...)

[Bald_Eagle1]

It may be useful to post the SNR log as a text file from the Telnet data tab, to see what was recorded as the highest value.

[roseway]

I'm afraid that the highest value of 70 dB is currently hard coded because I couldn't see this value ever being exceeded. Just to be sure, BE1's suggestion is good.

[kitz]

wow @ >70.  You really must be next door to the exchange. iirc even I didnt get above 55dB when on adsl2+.

I take it you arent in the UK otherwise PCB would have cut in.  Its also a tad bumpy for some reason, you expect a PSD mask to smooth that over particularly in the early tones to get a nice curve inverted 'U' shape.

[drewk]

I'm afraid that the highest value of 70 dB is currently hard coded because I couldn't see this value ever being exceeded. Just to be sure, BE1's suggestion is good.

Okay, I was just wondering if there was an easy way to do it, but it's not necessary.

Yes, I really am very close to the CO. That's why I was quite miffed in recent weeks trying to get someone from the telco to fix the noise on my line. I suppose they assumed that it could not possibly be a fault on their portion, and the speed was "good enough" so why bother.

However they did send me a new modem (DSL-2750B) as a consolation prize, which I was quite happy to discover could be used with the DSLStats software. That was very helpful in monitoring the line noise and solving the problem, namely a "high open", which I remedied by placing a 7k resistor across the line. The small amount of sealing current through the resistor has eliminated the line noise for the past couple of weeks and the maximum provisioned speed has been restored. Speed and performance have been consistently solid.

There are a few glitches with this router's firmware. It was obviously rushed into production. That may account for the strange SNR numbers at times. Or possibly it is the effect of the 7k resistor and the constant 5 ma of current through the line...

BTW, for anyone having intermittent noise issues which the telco can't fix, you may want to try a resistor across the line if you are comfortable with doing that sort of thing. Start with a 15k and decrease as needed, monitoring the SNR graphs in DSLStats.

Thanks for the great software. It helped a lot.

[drewk]

wow @ >70.  You really must be next door to the exchange. iirc even I didnt get above 55dB when on adsl2+.

I take it you arent in the UK otherwise PCB would have cut in.  Its also a tad bumpy for some reason, you expect a PSD mask to smooth that over particularly in the early tones to get a nice curve inverted 'U' shape.

Yes, this is in the States and only a few blocks from the exchange. The line is provisioned for 1 Mbps but capable of 10 Mbps. I have the option of getting a FiOS line but I'm sticking with copper. 1 Mbps is all I really need for now.

[boost]

BTW, for anyone having intermittent noise issues which the telco can't fix, you may want to try a resistor across the line if you are comfortable with doing that sort of thing. Start with a 15k and decrease as needed, monitoring the SNR graphs in DSLStats.

Thanks for the great software. It helped a lot.

I'd be interested to understand a bit more, if you don't mind humouring a noob? Why does this work? Are the line voltages the same in the US/UK, etc?

Cheers!

[drewk]

I'd be interested to understand a bit more, if you don't mind humouring a noob? Why does this work? Are the line voltages the same in the US/UK, etc?

A noisy condition can be caused when there is oxidation at a splice point or terminal connection. See the attached graph showing the initial very noisy SNR conditions on my line. Passing a current through the line can clear this if there is not too much corrosion. My SNR graph is a smooth line now since having installed the resistor a couple of weeks ago.

You can actually hear this effect if you have phone service on the ADSL line: when you pick up the receiver you may initially hear some crackle, which then goes away. This is because current flows through the line only when you lift the receiver. At other times there is no appreciable current through the line, even with the ADSL modems operating.

Placing a resistor across the line will allow a current to flow at all times, not just while the handset is lifted. This constant current can "seal" a line from noise caused by oxidized terminals or splices, which will improve the SNR figure and, eventually, the sync speeds.

The value of the resistor should be small enough to allow a few ma of current to flow through the line: if the open line voltage is 50v (typical) a 15,000 Ohm resistor will pass about 3 ma of current (I=E/R); smaller resistor values will pass greater current through the line. If the resistor value is too low the phone may fail to ring on incoming calls... I have about 5 ma of current on my line now. I expect that 10 or 20 ma may be the reasonable limit if the condition is really bad. The resistor should be rated 1 or 2 watts for these currents.

I suppose the line could deteriorate further if there is a junction or terminal exposed to the elements somewhere along the way. But at that time I will either decrease the resistor value further (increase the current) or hope that the noise level becomes bad enough so that the telco people will be prompted to fix it. I am interested to see how long the line will hold up. So far it has been rock solid even during recent downpours.

BTW, after getting rid of the noise on my line it took several days for the sync speeds to be re-negotiated back to top speed.

[kitz]

Thank you drewk for the above post and explanation. 

I think a lot of us on here have seen the effects of a HR/High Open fault.  Ive had one on my own line where the upstream SNRm would slowly decline and errors start to rack up.  Then after I lifted the receiver the SNRm would suddenly jump back up to what it should be and all would be good again for a while. In more severe cases, adsl will only work when the phone is off the hook. Im no telco engineer, but I believe its called the wetting current.

[JGO]

Im no telco engineer, but I believe its called the wetting current.

AFAK this was by analogy with "mercury wetted" switches/relays which is how the problem of signal currents without sufficient available power to disrupt an oxide film is solved elsewhere.

I have also heard of the 15k resistor across the line solution, with dire warnings not to push the current much higher than 3 ma as this will upset BT !    If it works it is a clear indication of a fault so need only be a temporary expedient anyway.

[konrado5]

But drewk has downstream SNR margin fluctuations and no upstream fluctuations. It is unusual on HR faults.

Best regards
konrado5

[drewk]

But drewk has downstream SNR margin fluctuations and no upstream fluctuations. It is unusual on HR faults.

Best regards
konrado5

My upstream SNR is not valid. The graphed value is simply that which is obtained during initial sync, so it never varies on the graph. I have to issue the CLI ADSL diagnostic command to get the current  upstream SNR value. I don't think there is any way to have it continuously plotted in DSLStats.

[drewk]

I have also heard of the 15k resistor across the line solution, with dire warnings not to push the current much higher than 3 ma as this will upset BT !    If it works it is a clear indication of a fault so need only be a temporary expedient anyway.

Yes, act accordingly with BT.

Over here I doubt the telco would ever notice an extra 10 or 20 ma of current in the line. But if that does get them to investigate the line, so much the better. I've done my part reporting the problems initially, including a 2-hour phone session with their call center half way around the planet (somewhat absurd considering that I am within walking distance to the telco CO). I won't be doing that again.

[boost]

Thanks, drewk.

When you say across the line, are you bridging both legs? Where, exactly, are you installing the resistor?

I have the same fluctuation and I wonder if this is the final push that's gets me onto fastpath

[konrado5]

drewk: could you attach your Hlog and QLN (essentially HR fault Hlog and QLN)?

Placing a resistor across the line will allow a current to flow at all times, not just while the handset is lifted.

Have you confused with capacitor?