In reply to a post by Chrysalis:

ignition snrm does move around without SRA. Bitswapping can turn tones off without SRA.

The modem estimates attainable based on snrm.

So if bitswapping reduces tones available for use then existing tones will reduce snrm to maintain sync.

I didn't say anything about the attainable rate or the SNRM.

On your line, the SNR per subcarrier has noticeably dropped over those months of scrutiny. Yet QLN has remained virtually unchanged (except in frame 5 of your excellent animation - what happened there? The noise floor dropped up to -20dBm. Just as if other subscribers in your cable bundle were disconnected while your line stayed connected.)

Putting those two factors together - the static noise floor and yet falling SNR across the subcarriers - suggests that (downstream) transmit power was cut. Would you have kept a record of transmit power? The xdslcmd command only reports an aggregate - the ACTATP (Actual Aggregate Transmit Power) in each direction, but it should still indicate a cut in power (if there was one).

Max Attainable Rate a.k.a. ATTNDR is calculated in basic form using the following equation:

http://huaweihg612hacking.files.wordpress.com/2013/0...

According to G.993.2 (see page 143 in [1]) to obtain a Bit Error Rate (BER) of 10^-7 - the accepted minimum error rate- the SNRGAP in that equation is set to 9.75dB.

Since the Target SNRM on your line has dropped from 6dB to around 3dB, the basic ATTNDR equation illustrates how and why, with a lowered TARSNRM, the bit-loading still edged higher.

With max data rates of 80/20 in the channel profile, this has resulted in a significant number of surplus bits per frame.

That's my reading any way!

cheers, a

[1] http://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-R...

Edited by deleted (Thu 30-May-13 19:38:48)

yeah could be that also.

In reply to a post by asbokid:

On your line, the SNR per subcarrier has noticeably dropped over those months of scrutiny. Yet QLN has remained virtually unchanged

For SNR, are you talking about over the entire period, or just within the single 6 month period without a resync?

In either case, I'm not really seeing the same thing that you are - except for the first 2 or 3 frames, which do seem a little higher. Perhaps because these were on a 40/10 package (before 80/20 became available, but after the 17a bandplan was turned on)? Maybe that's a little unfair.

I've done another animation of just 4 frames: the very first SNR frame of an 80/20 sync, then the start and end frames of that 6 month single-sync period, then the first sync after that 6 month period. It really doesn't seem that much different to me. At least not compared to the attainable values of 84, 81, 72, 83.

4-Frame SNR animation

(except in frame 5 of your excellent animation - what happened there? The noise floor dropped up to -20dBm. Just as if other subscribers in your cable bundle were disconnected while your line stayed connected.)

Certainly a drop to -140dBm is an indication that there is *no* noise whatsoever.

Actually, I don't know what happened there. I just looked back, and the modem must have resync'ed a good 2 months before I captured that set of stats - and I'm pretty sure we were on holiday on the date that resync would have happened.

The QLN figures are only gathered at the time of sync, aren't they? Perhaps the cabinet was reset in some way, and every line (including mine) needed to resync. Note that there isn't a change in tones 0-350, suggesting that the ADSL modems (exchange-fed) are still out there.

Putting those two factors together - the static noise floor and yet falling SNR across the subcarriers - suggests that (downstream) transmit power was cut. Would you have kept a record of transmit power? The xdslcmd command only reports an aggregate - the ACTATP (Actual Aggregate Transmit Power) in each direction, but it should still indicate a cut in power (if there was one).

I do indeed have the power values recorded in the Plink files, but they don't vary much... although I have seen the aggregate value vary by as much as 0.2dB between consecutive xdslcmd attempts (one --show, and one --stats).

These figures are from the --pbParams output:

Date      TX Power (dBm)
Date      Aggr     D1     D2     D3
02/2012   13.0     9.7    7.7    6.8
03/2012   13.0     9.6    7.7    6.8
06/2012   12.9     9.7    7.8    7.0
07/2012   13.0     9.7    7.7    7.0
09/2012   13.0     9.4    7.9    7.0
11/2012   13.1     9.7    7.8    6.9
12/2012   13.1     9.7    7.8    6.9
01/2013   13.1     9.7    7.8    6.9
05/2013   13.1     9.7    7.8    6.9

Every indication of the aggregate power was either 12.9, 13.0 or 13.1dBm.
The 3 power values for D1, D2 and D3 haven't changed since November.

Max Attainable Rate a.k.a. ATTNDR is calculated in basic form using the following equation:

http://huaweihg612hacking.files.wordpress.com/2013/0...

According to G.993.2 (see page 143 in [1]) to obtain a Bit Error Rate (BER) of 10^-7 - the accepted minimum error rate- the SNRGAP in that equation is set to 9.75dB.

Since the Target SNRM on your line has dropped from 6dB to around 3dB, the basic ATTNDR equation illustrates how and why, with a lowered TARSNRM, the bit-loading still edged higher.

With max data rates of 80/20 in the channel profile, this has resulted in a significant number of surplus bits per frame.

That's my reading any way!

This bit I'll have to look into tomorrow or Saturday, but note that it is my actual SNRM that dropped from 7dB down to 3dB - my target SNRM is still 6dB, I think. Doesn't the target stay at 6dB on Openreach cabinets?

Ta!

4-frame animation link doesn't work for me.

Hmm. Worked here agains.

How about this one?

In reply to a post by WWWombat:

I'm not really seeing the same thing that you are - except for the first 2 or 3 frames, which do seem a little higher. Perhaps because these were on a 40/10 package (before 80/20 became available, but after the 17a bandplan was turned on)? Maybe that's a little unfair.

Aha! So we should only look at graphs from roughly the beginning of the year?

I've done another animation of just 4 frames: the very first SNR frame of an 80/20 sync, then the start and end frames of that 6 month single-sync period, then the first sync after that 6 month period. It really doesn't seem that much different to me. At least not compared to the attainable values of 84, 81, 72, 83.

4-Frame SNR animation

You're right, the SNR drop couldn't account for such a variation in ATTNDR.

I just looked back, and the modem must have resync'ed a good 2 months before I captured that set of stats - and I'm pretty sure we were on holiday on the date that resync would have happened.. The QLN figures are only gathered at the time of sync, aren't they?

Yup, during initialisation.

Perhaps the cabinet was reset in some way, and every line (including mine) needed to resync. Note that there isn't a change in tones 0-350, suggesting that the ADSL modems (exchange-fed) are still out there.

Sure. Your line must have been first to sync! Or at least it beat the disturbers to sync.

I do indeed have the power values recorded in the Plink files, but they don't vary much... although I have seen the aggregate value vary by as much as 0.2dB between consecutive xdslcmd attempts (one --show, and one --stats)...

Every indication of the aggregate power was either 12.9, 13.0 or 13.1dBm.
The 3 power values for D1, D2 and D3 haven't changed since November.

Well tx power is not the answer then!

This bit I'll have to look into tomorrow or Saturday, but note that it is my actual SNRM that dropped from 7dB down to 3dB - my target SNRM is still 6dB, I think. Doesn't the target stay at 6dB on Openreach cabinets?

Yeah, the line-profile sets a target SNRM of 6dB for all lines. You're right viz actual SNRM vs target SNRM. I was trying to shoehorn your SNR figures into the ATTNDR equation. Although the outcome to subcarrier bit-loading and ultimately ATTNDR is the same, whether the margin is forced or has found its own level.

cheers, a

Edited by deleted (Fri 31-May-13 20:10:05)