Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: jaydub on July 17, 2018, 12:20:23 AM
-
There is a lot of posts on here that state that crosstalk is one of the major reasons for degradation of people's max attainable / actual sync rate.
I've not really seen any mention of what impact it has on other line characteristics.
As crosstalk has to impact the noise floor, presumably it would impact both the SNRM and QLN graphs on DSLStats.
I have certainly seen some degradation of max attainable on my line, since I first got VDSL back in October 2016, when my may attainable downstream was ~110Mbps and this has now dropped to ~86Mbps and I can see a significant reduction in SNRM from 12.9dB down to 7.4 dB now.
I don't have any QLN graphs that predate the demise of MDWS, so can't do an equivalent comparison. If i did have them back to 2016, would I also expect to see a significant change in the noise floor on the QLN graphs?
-
If you calculated the average (arithmetic mean) over the entire usable bandwidth, I would expect a change to be noticeable as more cross-talk degrades the circuit's performance.
I have a small utility, anf, that performs the calculation. If you have access to a C-compiler, then you would be welcome to have a copy of the source code . . .
-
Thanks B*cat for the offer, but I don't have a C-compiler and it was more of a general question anyway.
I was just trying to establish whether the Connection speed, SNRM and QLN graphs were the ones that would be most obviously impacted by crosstalk and, in the case of QLN, how obvious the impact would be.
I just wish I had a QLN graph from when I had a d/s attainable of 105Mbps to compare against.
-
I have a small utility, anf, that performs the calculation. If you have access to a C-compiler, then you would be welcome to have a copy of the source code . . .
Does anf work on linux? Could you send me?
-
Does anf work on linux? Could you send me?
Yes and yes. (You have a PM.)
-
Thank you very much. I don't understand results.
DEBUG 1: data_file is /mnt/dane/Statystyki ADSL/qlndiagmodewithoutmask3.txt
DEBUG 2: feof(dfp) returns false
DEBUG 3: Tone number QLN
DEBUG 4: sigma is -60890.5000, total number of sub-carriers is 512
DEBUG 5: sigma / subcarrier_count is -118.9268
-118 dBm/Hz per sub-carrier
-
I undersant. It merely calculates average noise. I've thought it says if noise is crosstalk or rather RFI.
-
Don't use the -d (the debug) flag.
The output is telling you that for your circuit, the average noise-floor is -118 dBm/Hz per sub-carrier. Should it change from that value, a more positive result (say -110 dBm/Hz per sub-carrier) indicates that the circuit has become noisier, a more negative result (say -120 dBm/Hz per sub-carrier) indicates the the circuit has become quieter. It is not a precise measurement, just a useful guide.
At the beginning of this month, the average noise-floor was determined to be -129 dBm/Hz per sub-carrier. A moment ago it was determined to be -128 dBm/Hz per sub-carrier. I.e. a little bit noisier. I am not concerned by that change, it is quite normal.
As a point of reference, the noise floor for a basic telephony circuit (i.e. that over which we have an xDSL service operating) is -140 dBm/Hz.
-
I wouldn't mind having a squint at this anf utility, too, pls :)
-
I wouldn't mind having a squint at this anf utility, too, pls :)
You have a PM. ;)
-
It's likely a silly question :blush: but is there any way of telling from where the crosstalk is coming?
I won't ask for the gadget in view of my rapidly decreasing and aging brain cells. :D
-
. . . is there any way of telling from where the crosstalk is coming?
Not really. At the most basic level we can say it is originating from all other xDSL service carrying pairs within the same cable as the pair being examined.
So an experiment to identify from whence the cross-talk is originating would be to turn off the xDSL service being carried on every other pair within the same cable and quantify how the circuit of interest now operates. That would be the base-line. Then turn on each of the other circuits in turn and again quantify how the circuit of interest is operating. The difference between the current metric and the base-line is of interest. (The smaller would be the better and zero would be ideal.)
-
Gulp!