Options for Domestic Wired Networking / Broadband

[prpr]

The noise margin sets the fallback mode for the ADSL protocol: the higher the noise margin, the cleaner the signal and the greater the bandwidth the ADSL can make use of.

No, a higher noise margin gives you a lower sync speed and therefore bit-rate, but you didn't show us that graph. Something else must be going on.

So far as I know, my bit of the line card is my bit of the line card and not shared with anybody - anyone know different?

You are correct.

Presuming that's the case, I don't see how the dropouts on the local line can be blamed on other people's traffic.

They can't. The prior post was hogwash.

 

[MikeSh]

The prior post was hogwash.

Can a question be hogwash?
One for assume vs presume perhaps.

 

[Black Hole]

No, a higher noise margin gives you a lower sync speed and therefore bit-rate

You've got that the wrong way around. It's a higher margin (better signal to noise ratio), not a higher noise floor.

 

[prpr]

You've got that the wrong way around. It's a higher margin (better signal to noise ratio), not a higher noise floor.

Believe what you will, but I'm right and you're wrong. Go and do some research elsewhere.

 

[Ezra Pound]

. . . and in the blue corner at 240 pounds we have . . .

 

[Black Hole]

Believe what you will, but I'm right and you're wrong. Go and do some research elsewhere.

OK:

Noise margin (also known as Signal-to-noise ratio margin, SNR) — is used to measure line quality and defines a minimum limit at which the signal level is above the noise level. The limit value of noise resistance for data transmission is 6 dB, a value below which the ADSL connection is not guaranteed at all. If the noise resistance is lower than 6 dB, the communication may be interrupted frequently.
If the noise resistance is higher than 10 dB, the line has good parameters for data transmission. The higher the value, the better the line quality. The 'Noise margin' value should be 6 dB and higher. It can reach values up to 30 dB if a short wire is used.
The field 'Noise margin' displays two values - the first number for the direction to the subscriber (downstream), and the second number for the direction from the subscriber (upstream).

'Noise margin' parameter value [dB]:

• up to 6 dB is a bad line, there are synchronization problems;
• 7 dB to 10 dB failures are possible;
• 11 dB to 20 dB is a good line, no problems with synchronization;
• 21 dB to 28 dB is a very good line;
• 29 dB and above is a perfect line.

Where I think we might be talking at cross purposes is illustrated by this remark in the same article:

if the line length is fixed, the noise margin decreases with increasing speed, and vice versa

The above is a statement that the noise margin will decrease with increasing speed (signal bandwidth), not a statement of the effect of a decreasing noise margin on available bandwidth (and therefore speed). I stand by what I said: the lower the noise margin, the lower the bandwidth you have available and therefore the lower the bit rate.

This becomes complicated by the way the ADSL connection reacts. If it detects a very high noise margin, it notches up the data rate which then brings the noise margin down. My line is adapting to my line conditions, and the noise margin plot has to be tempered by the current data rate, but my theoretical data rate is not changing over the duration of those plots so they are comparable.

However, the plots were intended to record my black-outs, not how good or poor my connection is at any particular time.

Today's chart (post 156) shows an unusually good receive noise margin (for me), which is why the speed tests show better than my typical 3Mbps (or less).

That's where I got it wrong: the "unusually good receive noise margin" might have been due to my data rate being notched down (it wasn't), and the speed test result might have been due to my data rate being notched up (it wasn't), so that leaves contention!

 

[everthewatcher]

This becomes complicated by the way the ADSL connection reacts. If it detects a very high noise margin, it notches up the data rate which then brings the noise margin down.

Same with VDSL - the ends of the link negotiate a line speed based on the noise margin. The data rate is increased until the noise margin drops to a certain value. Suppliers either use a 6dB noise margin downstream and/or limit the speed. BT use a 3db noise margin which lets them offer basic download rates of >50Mbps.

Currently I'm with Plusnet (using a BT Smart Hub) and they limit the speeds to maxima of 40/10, both of which I'm getting. Currently the noise margins are 7.1 dB down / 7.2 dB up.

 

[Ezra Pound]

TKO to BH?

 

[prpr]

TKO to BH?

No, quite the opposite, because he STILL hasn't posted his sync. speed vs. SNR margin plot.
Here's mine from an interference 'event' a couple of weeks ago (I usually run my line on a 3dB margin to get the extra speed, but I'm stuck elsewhere and don't want to fiddle remotely in case of total loss).
I really don't know why I pander to the patheticness of all this ignorance, but just to shut you both up:

 

[prpr]

And for those will still disbelieve, go and read (and understand) pages like this:
https://www.increasebroadbandspeed.co.uk/SNR-tweak

 

[Black Hole]

Okay okay, I have to move to the PC that's currently doing the logging so I can post snapshots from it (note the record is not continuous).

Here's an example of a jump in S/N ratio with no change in the negotiated line rate (this snapshot is representative but not unique):



I had to search hard to find an example capturing a change in the negotiated line rate (other than a complete drop-out):



I don't really know what you are getting so aerated about. Yes, originally I was a bit naive about all this, but I have fallen in and I think we're both talking about the same thing but from a different perspective.

https://www.increasebroadbandspeed.co.uk/SNR-tweak

What does that have to do with it? You and your fancy modem can set the minimum S/N ratio you are prepared to accept, thus allowing a higher line rate to be negotiated. So far as I know, I can't (update: no, I can't find any corresponding setting in my router control panel).

I agree that:

• Increasing the line rate decreases the S/N ratio (in the absence of any other factors);
• If the S/N ratio increases because of an improved environment, the system might renegotiate an increased line rate thus causing the measured S/N ratio to go down as a consequence.

I do not agree that:

• Somehow I do not understand this;
• That a decreased S/N ratio is the cause of an increased line rate, which is how your initial comments read to me.

But I reiterate: what mostly annoys me (having accepted 3Mbps is about the best I am likely to do at this line length from the exchange) are the drop-outs!

 

[Trev]

I think what BH is saying, and it seems pretty logical to me (and taken to the extreme for simplicity) is that if the S/N ratio were to be reduced to zero dB, the signal (if any) will be totally lost in the noise therefore no transfer will take pace. To put it in simple speak. "Download 'speed' will be zero."
Please don't tell me that my basic understanding of SNR is wrong after believing it for getting on for 60 years with a background in radar when receiver SNR is pretty much everything.

 

[Ezra Pound]

I think I can see why BH and prpr were seeing the numbers differently, it is in two term SNR and SNR Margin, they are obviously not the same, I think everyone would agree that higher SNR figures are good, e.g. a SNR of 6dB is better than a SNR of 3dB, however a SNR Margin of 3dB is better than a SNR Margin of 6dB

 

[Black Hole]

however a SNR Margin of 3dB is better than a SNR Margin of 6dB

I don't understand what a Signal-to-Noise Ratio Margin is.

Now, I admit to not having a background in analogue, RF, etc, so this is more hunch than certainty, but my understanding is that signal-to-noise ratio is received signal power divided by received noise power (converted to dB, ie logarithmic). And noise margin is the total received power (in dB) minus the noise power (in dB)... in other words very much the same thing expressed slightly differently (because subtracting logarithms is equivalent to division).

The noise plots in my posts quote "noise margin" in dB, which I take to mean the overall signal (in dB) minus the noise level (in dB), and which I also take to be equivalent to SNR. Is that wrong?

 

[Ezra Pound]

I wouldn't say that this is the only way to define SND Margin, but in this case the broadband supplies are using it to mean a safety buffer, so if you reduce the safety buffer you can get higher connection speeds


From the link in #170

 

[Black Hole]

however a SNR Margin of 3dB is better than a SNR Margin of 6dB

So why is 3dB better than 6dB? It seems to me all you are saying is that by defining your minimum margin as 3dB you are permitting a higher data rate, potentially with more error correction required. In other words 3dB is better than 6dB because with 6dB there is a bit of room left for more data. If that is the correct interpretation, we're all on the same page from that point of view.

But again, that's a target, not an actual measured value, and there is nothing I can do (short of replacing my router) to alter my target noise margin. My plots are the actual noise margin, not the target noise margin. I can only surmise what my target noise margin is from the observation the measured noise margin hovers around 12dB.

 

[Ezra Pound]

So why is 3dB better than 6dB?

because 6240 is a bigger number than 5568

It seems to me all you are saying is that by defining your minimum margin as 3dB you are permitting a higher data rate

Yes

potentially with more error correction required.

Yes, which is why you would only reduce the SNR Margin if you can do it without extra error correction

In other words 3dB is better than 6dB because with 6dB there is a bit of room left for more data.

No, because with 3dB there is a bit of room left for more data as long as no extra error correction is required

But again, that's a target, not an actual measured value, and there is nothing I can do (short of replacing my router) to alter my target noise margin. My plots are the actual noise margin, not the target noise margin. I can only surmise what my target noise margin is from the observation the measured noise margin hovers around 12dB.

Your argument in post #164 was saying that someone had got it the wrong way around, my posts were a way of explaining a possible reason, nothing more

 

[Trev]

My head hurts.

 

[Black Hole]

So does mine

 

[Black Hole]

I've re-read that link, and the quoted table shows the connection speed results from an experiment when setting different target margins. Let's not call them SNR margins, because that really doesn't make any sense - it's either a target signal to noise ratio or a target noise margin.

The experiment adjusts a Billion router. Setting a margin of 3dB gets a faster line rate than a margin of 6dB. All is as I expect it to be.

The article also says that if you set the margin too low, the line may not be stable and suffer drop-outs...